RBF_Cover (for eps) - National Water Research Institute

2RBF 

The National Water Research Institute Presents 

The Second International 

Riverbank Filtration Conference 

Riverbank Filtration: 

The Future Is NOW! 

PROGRAM & ABSTRACTS 

Edited by: 

GINA MELIN, National Water Research Institute 

September 16-19, 2003 

Hilton Cincinnati Netherlands Plaza ✦ Cincinnati, Ohio USA

2RBF 

The National Water Research Institute Presents 

The Second International 

Riverbank Filtration Conference 

Riverbank Filtration : 

The Future Is NOW! 

PROGRAM & ABSTRACTS 

Edited by: 

GINA MELIN, National Water Research Institute 

September 16-19, 2003 

Hilton Cincinnati Netherlands Plaza ✦ Cincinnati, Ohio USA

Published by the 

NATIONAL WATER RESEARCH INSTITUTE 

NWRI-2003-10 

10500 Ellis Avenue ✦ P.O. Box 20865 

Fountain Valley, California 92728-0865 

(714) 378-3278 ✦ Fax: (714) 378-3375 

www.NWRI-USA.org

Conference Planning Committee 

Chair: CHITTARANJAN RAY, Ph.D, P.E., University of Hawaii at Mañoa 

Conference Coordinator: GINA MELIN, National Water Research Institute 

EDWARD J. BOUWER, Ph.D., 

The Johns Hopkins University 

PAUL ECKERT, Ph.D., 

Stadtwerke Düsseldorf 

WILLIAM D. GOLLNITZ, 

Greater Cincinnati Water Works 

THOMAS GRISCHEK, Ph.D., 

University of Applied Sciences 

Dresden 

DAVID L. HAAS, P.E., 

Jordan, Jones & Goulding 

THOMAS HEBERER, Ph.D., 

Technical University of Berlin 

STEPHEN HUBBS, P.E., 

Louisville Water Company 

Conference Sponsors 

HENRY C. HUNT, CPG, 

Collector Wells International, Inc. 

RUDOLF IRMSCHER, Ph.D., 

Stadtwerke Düsseldorf AG 

RONALD B. LINSKY, 

National Water Research Institute 

JÜRGEN SCHUBERT, 


RODNEY A. SHEETS, 

United States Geological Survey 

THOMAS SPETH, Ph.D., P.E., 

United States 

Environmental Protection Agency 

✦ Environmental & Water Resources Institute of the American Society of Civil Engineers 

✦ Greater Cincinnati Water Works 

✦ International Association of Waterworks in the Rhine Catchment Area 

✦ Jordan, Jones & Goulding 

✦ Louisville Water Company 

✦ Stadtwerke Düsseldorf AG 

✦ United States Environmental Protection Agency 

✦ United States Geological Survey 

in collaboration with 

✦ International Association of Hydrogeologists Commission 

on Management of Aquifer Recharge 

i

Foreword 

In 1999, the National Water Research Institute (NWRI) invited American and European 

experts to the first International Riverbank Filtration Conference, held in Louisville, 

Kentucky, to discuss and promote riverbank filtration, a low-cost, relatively unknown water 

treatment technology in the United States. The result of this conference was Riverbank 

Filtration: Improving Source-Water Quality, a 365-page book written by over 30 experts and 

jointly published by NWRI and Kluwer Academic Publishers in 2002. 

The editors of Riverbank Filtration — namely, Chittaranjan Ray, Gina Melin, and Ronald 

Linsky — saw a need to expand on the topics raised in the book, specifically because the 

United States Environmental Protection Agency is developing the proposed Long Term 2 

Enhanced Surface Water Treatment Rule, which applies to public water-supply systems that 

use either surface water or groundwater under the direct influence of surface water as their 

raw-water source. As a result, NWRI organized the Second International Riverbank 

Filtration Conference, which featured over 40 presenters from around the world and was held 

in Cincinnati, Ohio, in September 2003. 

This conference came about through the dedication and assistance of numerous individuals 

and organizations worldwide. NWRI gratefully acknowledges the efforts of all those involved 

with planning, organizing, and sponsoring the conference, especially the 15-member 

Conference Planning Committee. NWRI also extends special thanks to the conference 

speakers, moderators, and panelists, whose expertise provided invaluable insight into the 

status and needs of riverbank-filtration technology. Finally, NWRI would specifically like to 

thank the Greater Cincinnati Water Works and Jordan, Jones & Goulding for providing 

countless hours and manpower towards organizing this conference. 

The extended abstracts provided at the conference were the contributions of conference 

presenters. Abstracts were edited only when obvious errors were detected or when printing 

requirements necessitated action. The opinions expressed within the abstracts are those of 

individual authors and do not necessarily reflect those of the sponsors. 

NWRI would like to extend sincere thanks to Gina Melin, Editor and Conference 

Coordinator, and Tim Hogan, Graphics Designer, for their efforts in bringing the abstracts to 

press and ensuring that the quality of each and every abstract reached its fullest potential. 

Ronald B. Linsky 

Executive Director 


iii

Conference Schedule & Abstracts Contents 

Wednesday, September 17, 2003 

All sessions will take place in the Continental Room 

7:00 am Registration Mezzanine Level 

7:00 am Speaker Breakfast Rosewood Room 

8:00 am Welcome Continental Room 

Ronald B. Linsky, National Water Research Institute, California 

8:15 am Keynote Address 

Riverbank Filtration: The American Experience . . . . . . . . . . . . . 1 

Edward J. Bouwer, Ph.D., The Johns Hopkins University, Maryland 

8:45 am Session 1: Costs 

Moderated by William D. Gollnitz, Greater Cincinnati Water Works, Ohio 

The Costs and Benefits of Riverbank-Filtration Systems . . . . . . . . . . . . . . 3 

Stephen A. Hubbs, P.E., Louisville Water Company, Kentucky 

9:15 am Session 2: Operations 

Moderated by Chittaranjan Ray, Ph.D., P.E., 

University of Hawaii at Mañoa, Hawaii 

Bridging Research and Practical Design Applications . . . . . . . . . . . . . . . . . 7 

David L. Haas, P.E., Jordan, Jones & Goulding, Inc., Georgia 

Construction and Maintenance of Wells for Riverbank Filtration . . . . . . . 17 

Henry C. Hunt, CPG, Collector Wells International, Inc., Ohio 

Aquifer Storage and Recharge Pretreatment: 

Synergies of Bank Filtration, Ozonation, and Ultraviolet Disinfection . . . 23 

Robert S. Cushing, Ph.D., Carollo Engineers, Florida 

Evolution from a Conventional Well Field 

to a Riverbank-Filtration System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 

John D. North, Cedar Rapids Water Department, Iowa 

11:15 am Session 3A: Hydraulic Aspects 

Moderated by Rudolf Irmscher, Ph.D., Stadtwerke Düsseldorf, Germany 

Groundwater Flow and Water-Quality – A Flowpath Study 

in the Seminole Well Field, Cedar Rapids, Iowa . . . . . . . . . . . . . . . . . . . . . 35 

Douglas J. Schnoebelen, Ph.D., United States Geological Survey, Iowa 

The Use of Aquifer Testing and Groundwater Modeling 

to Evaluate Changes in Aquifer/River Hydraulics at 

the Louisville Water Company . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 

David C. Schafer, David Schafer & Associates, Minnesota 

12:15 pm Lunch Rosewood Room 

v

vi 

1:30 pm Session 3B: Hydraulic Aspects 

Moderated by Rudolf Irmscher, Ph.D., Stadtwerke Düsseldorf, Germany 

Plugging in Riverbank-Filtration Systems: 

Evaluating Yield-Limiting Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 

Stephen A. Hubbs, P.E., Louisville Water Company, Kentucky 

Application of Different Tracers to Evaluate the Flow Regime 

at Riverbank-Filtration Sites in Berlin, Germany . . . . . . . . . . . . . . . . . . . . 49 

Dr. Gudrun Massman, Free University of Berlin, Germany 

2:30 pm Session 4: Siting 

Moderated by Henry C. Hunt, CPG, 

Collector Wells International, Inc., Ohio 

Siting and Testing Procedures for Riverbank-Filtration Systems . . . . . . . . 57 

Samuel M. Stowe, P.G., CPG, International Water Consultants, Inc., Ohio 

Water Quality Management for Existing Riverbank Filtration Sites 

along the Elbe River in Germany. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 

Prof. Dr.-Ing. Thomas Grischek, University of Applied Sciences Dresden, Germany 

3:30 pm Session 5: Dynamics 

Moderated by Prof. Dr.-Ing. Thomas Grischek, 

University of Applied Sciences Dresden, Germany 

Using Models to Predict Filtrate Quality at Riverbank-Filtration Sites – 

What Is the Adequate Level of Modeling?. . . . . . . . . . . . . . . . . . . . . . . . . . 69 

Chittaranjan Ray, Ph.D., P.E., University of Hawaii at Mañoa, Hawaii 

The 100-Year Flood of the Elbe River in 2002 

and Its Effects on Riverbank-Filtration Sites . . . . . . . . . . . . . . . . . . . . . . . 81 

Dipl.-Ing. Matthias Krueger, Fernwasserversorgung Elbaue-Ostharz GmbH, Germany 

Temporal Changes of Natural Attenuation Processes 

During Bank Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 

Paul Eckert, Ph.D., Stadtwerke Düsseldorf AG, Germany 

An Update of the City of Guelph’s Response to Regulation 459/00: 

Effective Natural In Situ Filtration of Several Groundwater 

Under the Direct Influence of Surface-Water Supplies . . . . . . . . . . . . . . . 91 

Dennis E. Mutti, P.E., Associated Engineering Limited, Canada 

On Bank Filtration and Reactive Transport Modeling . . . . . . . . . . . . . . . . 93 

Dr.-Ing. Ekkehard Holzbecher, Humboldt University, Germany 

6:00 pm Reception Pavilion Foyer 

6:30 pm Dinner Pavilion Ballroom 

Dinner Speaker: 

Hydraulic Sensitivities and Reduction Potential Correlated 

with the Distance Between the Riverbank and Production Well . . . . . . . . 99 

Bernhard Wett, Ph.D., University of Innsbruck, Austria

Thursday, September 18, 2003 

All sessions will take place in the Continental Room 


8:00 am Keynote Address 

Riverbank Filtration: The European Experience . . . . . . . . . . . . . . 

Prof. Dr.-Ing. Martin Jekel, Technical University of Berlin, Germany 

105 

8:30 am Session 6: Microorganisms 

Moderated by Edward J. Bouwer, Ph.D., The Johns Hopkins University, Maryland 

Using Microscopic Particulate Analysis for Riverbank Filtration . . . . . . . 111 

Jennifer L. Clancy, Ph.D., Clancy Environmental Consultants, Inc., Vermont 

Transport and Removal of Cryptosporidium Oocysts 

in Subsurface Porous Media. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 

Menachem Elimelech, Ph.D., Yale University, Connecticut 

Laboratory and Field Strategies for Assessing Pathogen Removal 

by Riverbank Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 

Monica B. Emelko, Ph.D., University of Waterloo, Canada 

Fate of Disinfection Byproduct Precursors and Microorganisms 

During Riverbank Filtration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 

W. Joshua Weiss, The Johns Hopkins University, Maryland 

Assessment of the Microbial Removal Capabilities 

of Riverbank Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 

Vasiliki Partinoudi, University of New Hampshire, New Hampshire 

11:00 am Session 7: Organics Removal 

Moderated by Richard J. Miltner, P.E., 

United States Environmental Protection Agency, Ohio 

Riverbank Filtration: A Very Efficient Treatment Process 

for the Removal of Organic Contaminants?. . . . . . . . . . . . . . . . . . . . . . . . . 137 

Dr.-Ing. Heinz-Jürgen Brauch, DVGW-Technologiezentrum Wasser, Germany 

Organics Removal by Riverbank Filtration 

at the Greater Cincinnati Water Works Site . . . . . . . . . . . . . . . . . . . . . . . . 143 

Jeffrey Vogt, Greater Cincinnati Water Works, Ohio 

12:00 pm Lunch Pavilion Ballroom 

Lunch Speaker: 

Potential Uses of Riverbank Filtration 

for Regulatory Compliance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 

Stig Regli, United States Environmental Protection Agency, Washington, D.C. 

149 

vii

viii 

1:15 pm Session 8: Emerging Contaminants Removal 

Moderated by Monica B. Emelko, Ph.D., University of Waterloo, Canada 

Transport and Attenuation of Pharmaceutical Residues 

During Bank Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 

Andy Mechlinski, Technical University of Berlin, Germany 

Attenuation of Pharmaceuticals During Riverbank Filtration . . . . . . . . . . 155 

Traugott Scheytt, Ph.D., Technical University of Berlin, Germany 

The Fate of Bulk Organics and Emerging Contaminants 

During Soil-Aquifer Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 

Dr. Jörg E. Drewes, Colorado School of Mines, Colorado 

Ethylenediaminetetraacetic Acid Occurrence and Removal 

Through Bank Filtration in the Platte River, Nebraska . . . . . . . . . . . . . . . 163 

Jason R. Vogel, Ph.D., United States Geological Survey, Nebraska 

3:15 pm Session 9: Public Policy and Regulatory 

Moderated by Richard J. Miltner, P.E., 

United States Environmental Protection Agency, Ohio 

Riverbank Filtration as a Regional Supply Option 

for the United States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 

Leo Gentile, P.G., CPG, Jordan, Jones & Goulding, Inc., Georgia 

Application of the Long Term 2 Enhanced Surface Water 

Treatment Rule Microbial Toolbox at Existing Water Plants . . . . . . . . . . . 173 

Richard A. Brown, Environmental Engineering and Technology, Inc., Virginia 

Draft Protocol for the Demonstration of Effective Riverbank Filtration . . . . 175 

William D. Gollnitz, Greater Cincinnati Water Works, Ohio 

Source Water Protection and Riverbank Filtration 

in the Dyje River Basin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 

Prof.-Dr. Petr Hlavinek, Brno University of Technology, Czech Republic 

5:15 pm Session 10: Research Needs Panel Discussion 

Moderated by Ronald B. Linsky, National Water Research Institute, California 

Panelists: 

Kellogg J. Schwab, Ph.D., 

Johns Hopkins Bloomberg School of Public Health, Maryland 

Peter Fox, Ph.D., Arizona State University, Arizona 

Monica B. Emelko, Ph.D., University of Waterloo, Canada 

Prof. Dr.-Ing. Thomas Grischek, 

University of Applied Sciences Dresden, Germany 

Prof. Vladimir Rojanschi, Ecological University Bucharest, Romania 

6:30 pm Reception Rosewood Room

Friday, September 19, 2003 


8:00 am Session 11: Case Studies “Lessons Learned” 

Moderated by Stephen Hubbs, P.E., Louisville Water Company, Kentucky 


Flowpath Study Field Design: Methodology and Evaluation. . . . . . . . . . . . 187 

Bruce Whitteberry, P.G., Greater Cincinnati Water Works, Ohio 

The Hungarian Experience with Riverbank Filtration . . . . . . . . . . . . . . . . 193 

Ferenc Laszlo, Ph.D., 

Institute for Water Pollution Control, Water Resources Research Centre, Hungary 

Nitrate Pollution of a Water Resource – 

15 18 

N and 0 Study of Infiltrated Surface Water . . . . . . . . . . . . . . . . . . . . . 197 

Frantisek Buzek, Ph.D., Czech Geological Survey, Czech Republic 

Microbial Growth in Artificially Recharged Groundwater: 

Experiences from a 4-Year Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203 

Ilkka T. Miettinen, Ph.D., National Public Health Institute, Finland 

Evaluation of the Existing Performance of Infiltration Galleries 

in the Alluvial Deposits of the Parapeti River . . . . . . . . . . . . . . . . . . . . . . 207 

Dip.-Eng. Alvaro Camacho Garnica, 

Bolivian Association of Sanitary Engineers, Bolivia 

Sensitivity and Implication of Microscopic Particulate Analysis – 

A Collector Well Owner’s Perspective. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 

Barry C. Beyeler, City of Boardman, Oregon 

Combined Use of Surface Water and Groundwater 

for Drinking-Water Production in the Barcelona Metropolitan Area. . . . . 217 

Jordi Martín-Alonso, Barcelona’s Water Company, Spain 

11:30 am Conference Wrap-Up 

Edward J. Bouwer, Ph.D., Johns Hopkins University, Maryland 

Martin Jekel, Ph.D., Technical University of Berlin, Germany 

ix

Acronyms 

ADA ß-alaninediacetic acid 

AOC Assimilable organic carbon 

AOX Adsorbable organic halogen 

ASR Aquifer Storage and Recovery 

DBP Disinfection byproduct 

DOC Dissolved organic carbon 

DTPA Diethylenetrinitrilopenataacetic acid 

EDTA Ethylenediaminetetraacetic acid 

GAC Granular activated carbon 

GC Gas chromatography 

GWUDI Groundwater under the direct influence of surface water 

HAA Haloacetic acid 

HPI Hydrophilic carbon 

HPO-A Hydrophobic acids 

LT2ESWTR Long Term 2 Enhanced Surface Water Treatment Rule 

MAP Microbially available phosphorous 

MS Mass spectrometry 

NASRI Natural and Artificial Systems for Recharge and Infiltration 

NTA Nitrilotriacetic acid 

NOM Natural organic matter 

NPEC Nonylphenolpolyethoxycarboxylate 

O&M Operation and maintenance 

PDTA 1.3-propylenedinitrilotetraacetic acid 

PhAC Pharmaceutically active compound 

PHREEQC pH redox equilibrium equation 

RBF Riverbank filtration 

THM Trihalomethane 

TOC Total organic carbon 

USEPA United States Environmental Protection Agency 

USGS United States Geological Survey 

UV Ultraviolet 

xi

xii 

Units of Measure 

cfu Colony-forming unit. 

ft Foot 

m Meter 

m 3 /d Cubic meters per day 

m 3 /s Cubic meters per second 

MGD Million gallons per day 

mg/L Milligrams per liter 

ntu Nephelometric turbidity unit 

pCi/L Picocurie per liter 

pfu Plaque-forming unit. 

µg/L Micrograms per liter

Conference 

Abstracts

Keynote Presentation 

Riverbank Filtration: The American Experience 

Edward J. Bouwer, Ph.D. 


Baltimore, Maryland 

Riverbank filtration (RBF) is a process during which surface water is subjected to subsurface flow 

prior to extraction from vertical or horizontal wells. Most RBF systems are located along 

riverbanks. Alternative systems can involve lakes and infiltration ponds. The objective of this 

keynote address is to provide an overview of the water-quality improvements possible with RBF, 

the motivation for RBF and its promise in the United States, and some of the remaining 

challenges for the reliable implementation of this technology. 

During infiltration and soil passage, surface water is subjected to a combination of physical, 

chemical, and biological processes, such as filtration, dilution, sorption, and biodegradation, that 

can significantly improve raw-water quality. Transport through alluvial aquifers is associated with 

a number of water-quality benefits, including the removal of microbes, pesticides, total organic 

carbon (TOC), dissolved organic carbon (DOC), nitrate, and other contaminants. In comparison 

to most groundwater sources, alluvial aquifers that are hydraulically connected to rivers are 

typically easier to exploit (shallow) and more highly productive for drinking-water supplies. As 

reflected by several recently published reviews (Journal of Hydrology, 2002; Ray et al., 2002a; 

Tufenkji et al., 2002; and Ray et al., 2002b), RBF is receiving increased attention, especially in the 

United States. One motivation for the increased applications of RBF is the need for drinking-water 

utilities to meet increasingly stringent drinking-water regulations, especially with regard to: 

• The provision of multiple barriers for protection against microbial pathogens. 

• Tighter regulations for disinfection byproducts (DBPs), such as trihalomethanes (THMs) 

and haloacetic acids (HAAs). 

A second motivation is the ability of RBF to provide continuous treatment and to buffer against 

accidental spills and terrorist events. 

Since RBF is a natural treatment system and has a long history of use in Europe, it is sometimes 

viewed as a simple process. This simplicity is appealing; however, the geochemical, biological, and 

hydrologic factors that control the removal of dissolved and particulate contaminants during RBF 

are complex. We have learned from experience with groundwater remediation that the effectiveness 

of subsurface processes is inherently site-specific. Many of the papers presented at this conference 

will address the factors influencing water-quality improvements possible with RBF systems. 

One question of interest to users can be stated as, “Is RBF with post-disinfection an acceptable, or 

even preferable, alternative to conventional drinking-water treatment?” In my laboratory, the 

reduction of DBP precursors upon RBF was compared with that obtained using a bench-scale 

Correspondence should be addressed to: 


Professor, Department of Geography and Environmental Engineering 


3400 N. Charles Street • Baltimore, Maryland 21218 USA 

Phone: (410) 516-7437 • Fax: (410) 516-8996 • Email: Bouwer@jhu.edu 

1

2 

conventional treatment train on corresponding river waters. The river waters were subjected to a 

treatment train consisting of coagulation, flocculation, sedimentation, filtration, and ozonation. 

This research showed that RBF performs as well as or better than a bench-scale conventional 

treatment train (based on coagulation chemistry for optimum turbidity removal) with respect to 

the removal of natural organic matter (NOM), particularly precursor material for THM4 and 

HAA6 concentrations. Consequently, a potential major benefit of RBF is as a pretreatment step 

for controlling DBPs. A shift from chlorinated to brominated DBPs occurred during RBF. Since 

brominated DBPs tend to have greater toxicity than chlorinated DBPs, the shift from chlorinated 

to brominated DBP species caused reductions in the calculated risk for bank-filtered waters in this 

study to be lower than corresponding reductions in THM concentrations. Nonetheless, the data 

demonstrate the ability of RBF to reduce theoretical risk due to THMs formed upon chlorination 

in all cases and with substantially better performance than the bench-scale conventional 

treatment train. 

A remaining challenge for the reliable implementation of RBF as a technology is the dynamic 

nature of these complex hydrologic systems. Transient flows in rivers influence flow and removal 

processes during ground passage, which can affect the quality of bank-filtered waters. A more 

detailed understanding of the fundamental processes that govern contaminant transport in RBF 

systems will lead to the reliable design, implementation, and operation of RBF systems. 

REFERENCES 

Journal of Hydrology (2002). Special Issue on Bank Filtration, 266(3-4): 139-284. 

Ray, C., T. Grischek, J. Schubert, J.Z. Wang, and T.F. Speth (2002a). “A perspective of riverbank filtration.” 

Jour. AWWA, 94(4): 149-160. 

Ray, C., G. Melin, and R.B. Linsky, editors (2002b). Riverbank Filtration: Improving Source Water Quality, 

Kluwer Academic Publishers, Dordrecht. 

Tufenkji, N., J.N. Ryan, and M. Elimelech (2002). “The promise of bank filtration.” Environmental Science 

and Technology, 36(21): 423A-428A. 

ED BOUWER has taught environmental engineering courses at The Johns Hopkins 

University since 1985. His research interests include factors that influence the biotransformation 

of organic contaminants, bioremediation for the control of organic contaminants 

at waste sites, biofilm kinetics, the interaction between biotic and abiotic processes, 

groundwater contamination, biological processes design in wastewater, industrial and 

drinking-water treatment, and the transport and fate of microorganisms in porous media. 

At present, he is an editor for or is on the editorial board of several journals, including the 

Journal of Contaminant Hydrology, Biodegradation, and Environmental Engineering Sciences. Bouwer received a 

B.S. in Civil Engineering from Arizona State University, and both an M.S. and Ph.D. in Environmental 

Engineering and Science from Stanford University.

Session 1: Costs 

The Costs and Benefits of Riverbank-Filtration Systems 

Stephen A. Hubbs, P.E. 


Louisville, Kentucky 

Henry C. Hunt, CPG 


Columbus, Ohio 

Jürgen Schubert 

Stadtwerke Düsseldorf 

Düsseldorf, Germany 

The Benefits of Riverbank Filtration 

The history of RBF in modern times is connected to the experience of disease outbreaks in Europe 

in the 1890s, with specific reference to the cholera epidemic in Hamburg, Germany, in 1892. The 

preference for groundwater, RBF, and artificial recharge to surface water stems from this 

experience, noting much more wholesome water from these sources than from surface water. 

A much earlier reference to RBF (albeit far less scientific) is found in the Bible: 

The fish that were in the Nile died, and the Nile became foul, so that the Egyptians 

could not drink water from the Nile … So all the Egyptians dug around the Nile for 

water to drink, for they could not drink of the water of the Nile. (Exodus 7, 21-24) 

Thus, it appears that the benefits of RBF are anything but new! 

The benefits of RBF have been cataloged in recent publications to include: 

• Particle removal. 

• Pathogen removal. 

• Organic and inorganic chemical removal. 

• Peak smoothing in spills. 

• Reduction in DBP formation. 

• Production of a more biologically stable water. 

When faced with the decision to choose an advanced treatment technology for its two treatment 

plants, the Louisville Water Company in Louisville, Kentucky, sought to compare these benefits 

against the benefits of in-plant treatment techniques from both a treatment efficiency perspective 

and cost-based perspective; however, a comprehensive, quantitative measure was difficult to develop. 



Vice President, New Technology 


550 South Third Street • Louisville, Kentucky 40202 USA 

Phone: (502) 569-3675 • Fax: (502) 569-0813 • Email: SHubbs@lwcky.com 

3

4 

The Louisville Water Company also wanted to involve the public in the overall decision process, 

so a simple communication tool was desired that was capable of ranking various treatment 

technologies with regards to the risks they were designed to reduce. Treatment effectiveness was 

simply identified by a series of “+” signs, with one “+” being effective and two “++” being highly 

effective. A negative assignment (“–”) indicated that the treatment process had an overall 

negative impact on the risk of a selected component, and a “0” indicated no impact. 

The risks evaluated in the analysis included: pathogens, such as Giardia and Cryptosporidium; 

DBPs; tastes and odor (2-methylisoborneol and Geosmin); synthetic organic chemicals like 

atrazine and Aloclor; and river-borne spills of industrial chemicals. Operational benefits included: 

reduced regrowth in the distribution system, reduced main breaks from avoidance of extremely 

cold water, avoidance of zebra mussels, and reliability associated with simple operation. These risks 

and benefits were used to compare the RBF process to other treatment technologies, including 

conventional treatment, ozone, ultraviolet light (UV), granular activated carbon (GAC), biological 

active carbon, combinations of all of these, and membranes. 

This matrix was recently modified to include risks associated with the presence of radon and 

arsenic in groundwater. Radon has been detected in RBF water in Louisville at levels near the 

maximum contaminant level of 300 picocuries per liter (pCi/L). The radon level leaving the 

treatment plant is below the detection limit of 50 pCi/L, with the reduction the result of 

out-gassing in open treatment basins and blending with surface water. Arsenic has not been 

detected in RBF water at Louisville. Radon and arsenic risk factors do not exist in Ohio River 

source water. The treatments for radon (aeration) and arsenic (flocculation with ferric) effectively 

reduce these risk factors at an increase in treatment costs. 

A matrix comparing these risks and operational benefits is presented in Table 1. This analysis 

indicates which treatment combinations provide specific water-quality benefits. From this analysis, 

it was obvious that RBF was capable of matching the benefits of in-plant treatment options. 

Table 1. Comparison of Risks and Operational Benefits 

River + Ozone River + GAC River + GAC 

Benefits RBF + UV and UV + UV + Membranes 

Particle Removal + + + + 

Microbial Removal + + + + 

DBP Reduction + + + + 

Taste and Odor + + + + 

Spill Dampening ++ 0 ++ + 

Iron/Manganese – + + 0 

SOC Removal + + + + 

AOC/BDOC Control + 0 + + 

Nitrification Control + + + + 

Temperature ++ 0 0 0 

Operability + 0 0 + 

Residuals 0 0 0 + 

Multiple Barriers ++ ++ ++ ++ 

Reduced Chemicals + 0 0 + 

Radon – 0 0 0 

Arsenic 0 0 0 0 

TOTAL 13 9 12 13 

AOC = Assimilable organic carbon. BDOC = Biodegradable organic carbon. SOC = Synthetic organic chemical.

Cost-Evaluation of Riverbank Filtration 

The capital cost of a RBF system depends on many factors, including aquifer characteristics, type 

of well-screen installation (vertical or horizontal), aesthetic considerations in facility design, and 

distance to the population served. The operational costs vary as a function of water quality and 

required treatment, lift required in pumping, and pump and well-screen maintenance costs. The 

ability of a stream to support a given well-field capacity can be calculated as the yield per unit 

length of riverbank. This capacity is influenced by the composition of the riverbed, riverbed 

scouring characteristics, stream-water quality, and width of the river. 

Large river systems situated in glacial sands and gravels (such as the Ohio, Mississippi, and Rhine 

rivers) can sustain yields up to 8 million gallons per day (MGD) per 1,000 feet (ft) of riverbank. 

Typical installations in these aquifers include 1.5-MGD vertical wells spaced on 200-ft centers, or 

15-MGD horizontal collector wells spaced at 2,000-ft centers. 

The 20-MGD horizontal collector well system constructed in Louisville in 1999 cost $5 million. 

The system included: 

• Seven laterals that are 200 to 240 ft in length. 

• A 21-ft diameter, 100-ft deep caisson. 

• A pump house and controls with one constant speed and one variable speed pump 

(both 10 MGD). 

• Two thousand feet of 42-inch discharge piping to the plant. 

The pump house was designed to include architectural features complimentary to surrounding 

residential neighborhoods. This system can peak at over 20 MGD in warm weather and is operated 

year-round at 17 MGD. 

When this system was being considered, alternative treatment techniques for surface-water treatment 

were also evaluated. Critical treatment functions included efficiency in removing Giardia and 

Cryptosporidium, ability to remove 2-methylisoborneol and Geosmin, and DBP reduction. Suitable 

alternatives were determined from the matrix and evaluated for cost. This process was repeated 

twice, by two separate consultants: once in the initial planning stages of the 20-MGD project 

(1995), and again just before a contract was let to design a 45-MGD expansion of the system 

(2002). 

The surface-water treatment process selected in the final cost analysis as providing comparable 

benefits to the overall benefits of RBF included: conventional treatment, ozone and biological 

treatment in GAC-capped filters, and UV/chlorine/chloramine disinfection. Treatment costs, 

however, were also estimated for separate elements of this treatment scheme (as was the cost of 

membranes) for the 180-MGD treatment plant at Crescent Hill. 

The results of this cost comparison are included in Table 2. The capital costs were amortized over 

20 years in this analysis, and the operation and maintenance (O&M) costs include the treatment 

costs to reduce hardness from 220 to 160 milligrams per liter (mg/L). Based on this analysis, the 

decision was made to proceed with the development of RBF for the entire capacity of the 60-MGD 

B.E. Payne Water Treatment Plant, and to continue considering RBF for the larger 180-MGD 

capacity Crescent Hill Treatment Plant. 

5

6 

Treatment Capital Cost Annual O&M Cost Present Worth Cost 

Alternative ($ million) ($ million) ($ million) 

RBF Conventional 103 1.82 112 

RBF + UV 116 2.34 130 

River + Ozone UV 70 6.36 152 

River + GAC + UV 114 5.46 160 

River + Membranes 

+ GAC + UV 

Table 2. Results of the Cost Comparison 

204 4.96 251 

Capital Costs of Alternative Well-Field Design: Hard-Rock Tunnel Collector 

The difficulty of obtaining riverfront property in developed areas prompted the Louisville Water 

Company to consider alternative designs to the traditional well-field configuration. The design 

selected for constructing the 45-MGD addition to the existing 20-MGD RBF capacity includes 

the construction of a large-diameter (10- to 12-ft) horizontal tunnel in bedrock, below the waterbearing 

sand and gravel aquifer. Vertical wells will be constructed at 200-ft centers, and will 

penetrate the bedrock and discharge by gravity into the tunnel. A single pump station located at 

the treatment plant will extract up to 45 MGD from the tunnel, connecting the 30 vertical wells 

in the system. This construction technique minimizes the impact on landowners, with visible 

construction activities limited to drilling and developing the individual wells. This design also 

allows total flow to be distributed to each well, evenly distributing riverbed plugging stresses across 

6,000 ft of riverbank. 

The cost of this system was compared to the cost of developing a conventional well field with 

submersible pumps and a collecting header. The cost-estimate for the bedrock tunnel system was highly 

impacted by the assigned cost of the hard-rock tunnel. The final design included a cement-lined tunnel 

to protect against intermittent layers of shale encountered in the massive limestone formation. 

The current estimate for this hard-rock tunnel-vertical well extraction system is $33 million for 

45-MGD capacity. The system involves approximately 6,000 ft of riverbank and hard-rock tunnel. 

The comparable conventional vertical well field has been estimated at $23 million. Current plans 

are to design and construct the hard-rock vertical well system, noting advantages in expandability, 

future connection to an additional 180-MGD system for the larger treatment plant, and general 

constructability in the developed riverfront area. 

STEVE HUBBS is a Professional Engineer with 28 years of experience at the Louisville 

Water Company in Louisville, Kentucky. In the early 1980s, he began researching riverbank 

filtration as an alternate source of water for the Louisville Water Company, specifically 

looking at the reduction in disinfection byproduct precursors, river-borne organics, and 

mutagenicity in the riverbank-filtration process. His work continued in the 1990s with a 

focus on pathogen reduction, and his research is now being conducted on the hydraulic 

connection between the riverbed and aquifer, with a focus on riverbed plugging dynamics 

and their influence on sustainable yields from high-capacity riverbank-filtration systems. Hubbs received an 

M.S. in Environmental Engineering from the University of Louisville, and is currently enrolled in the Ph.D. 

program in Civil and Environmental Engineering at the University of Louisville, focusing on the hydraulics 

of riverbank-filtration systems.

Session 2: Operations 

Bridging Research and Practical Design Applications 

David L. Haas, P.E. 

Jordan, Jones & Goulding, Inc. 

Norcross, Georgia 

Michael J. Robison, P.E. 



David R. Wilkes, P.E. 



Background 

The Louisville Water Company operates two water filtration plants: the B.E. Payne Water Treatment 

Plant (Payne Plant), which has a treatment capacity of 60 MGD (227,000 cubic meters per day [m 3 /d]) 

and is located in eastern Jefferson County, Kentucky; and the Crescent Hill Water Treatment 

Plant, which is located closer to downtown Louisville and has a treatment capacity of 180 MGD 

(682,000 m 3 /d). Both are conventional surface-water treatment plants drawing their supply from 

the Ohio River. The Payne Plant has two 60-inch (1.5-meter [m]) raw-water intake pipes located 

on plant property. The intake for the Crescent Hill Water Treatment Plant is located near the 

intersection of Zorn Avenue and River Road, approximately 7 miles (4.4 kilometers) downstream 

from the Payne Plant intake. 

In 1999, the Louisville Water Company started operating a horizontal collector well at its 

Payne Plant. This horizontal collector well consists of a 16-ft (4.9-m) diameter caisson. 

The caisson was constructed down to the top of bedrock and is approximately 100-ft (30.5-m) 

deep. There are seven 12-inch (30.5-centimeter) collector laterals radiating out horizontally from 

the caisson. A pump station was constructed on top of the caisson above the flood plain to pump 

water to the Payne Plant for further treatment. The design capacity of this well is 15 MGD 

(57,000 m 3 /d). 

Jordan, Jones & Goulding, Inc. was retained by the the Louisville Water Company to implement 

RBF for the remainder of the Payne Plant. This second phase of RBF will provide an additional 

45 MGD (171,000 m 3 /d) of capacity. Ultimately, the Louisville Water Company desires to extend 

RBF technology to the Crescent Hill Water Treatment Plant. 

This paper provides a summary of issues that were faced in taking the concept of providing 

additional RBF capacity for the Payne Plant through design. There were three primary areas that 


David L. Haas, P.E. 

Senior Project Manager 


6801 Governors Lake Parkway • Norcross, Georgia 30071 USA 

Phone: (678) 333-0242 • Fax: (678) 333-0828 • Email: DHAAS@JJG.com 

7

8 

design engineers needed to address: 

• Type of RBF collection system. 

• Hydrogeologic yield. 

• Water quality and treatment. 

Type of RBF Collection System 

First, it was necessary to select the most appropriate type of RBF technology to satisfy the technical 

needs of the project economically, as well as other goals such as community acceptance of the 

project. The construction of additional horizontal collector wells, similar to the existing well, was 

considered unacceptable because 15 to 20 of these collector wells would be required to satisfy the 

total treatment capacity for both plants. The construction of this many wells, each with an aboveground 

pump station, would be costly, as well as aesthetically unpleasing. 

Three options were considered that would use tunnels in conjunction with RBF to satisfy the 

needs of the project: 

• Option 1 consisted of driving a tunnel through the alluvial deposits that overlay bedrock 

and installing collector laterals from within the tunnel (soft-ground tunnel option). This 

design concept is depicted in Figure 1. 

• Option 2 included installing a series of horizontal collector wells (Figure 2) that would 

be capped at grade and connected together using a deep tunnel through the bedrock. 

• Option 3 consisted of installing vertical collectors connected together using a tunnel 

through the bedrock (Figure 3). 

With any of these options, the number of above-ground pump stations would be minimized, 

because water from horizontal or vertical collectors would be conveyed through the tunnel to a 

single pump station located on plant property. 

In addition to cost, factors that were considered in selecting the type of RBF collection system 

included construction and maintenance. 

Construction Considerations 

The two main construction issues that were evaluated for Option 1 were the feasibility of 

constructing the tunnel in the alluvial deposits along the riverbank and the feasibility of installing 

laterals safely from within the tunnel. With regard to Options 2 and 3, the main construction issue 

evaluated was rock quality. 

To evaluate these factors, a detailed subsurface investigation was conducted that included the 

following elements: 

• Large-diameter bucket auger borings to bedrock. 

• Grain-size analysis of composite samples obtained from the bucket auger borings. 

• Core borings of bedrock materials. 

• Examination of rock outcrops near the site. 

Large-diameter bucket auger borings were used to collect representative samples of alluvium. 

Because alluvium contains significant amounts of coarse gravel, cobbles, and possibly boulders, the

Ohio River 

Southern Caisson 

(capped at grade) 

19-ft Diameter 

Soft Soft-Ground 

Ground Tunnel unnel 

Tunnel 

Caisson 

Capped 

at Grade 

Ground Surface 

Pump 

Station 

Lagoon 

Existing Collector Well 

Existing 48-Inch 

Raw Water Transmission Line 

New 54-Inch 

Raw Water Transmission Line 

Lagoon 

Lagoon 

Clay 

Lagoon 

Sand and Gravel 

14-ft Diameter Tunnel with 25 Laterals 

Bedrock 

Plan View 1A 

Existing (Two) 

60-Inch Diameter 

Water Intake Pipes 

Northern Caisson 

(Pump Station) 

32-ft Diameter Shaft 

200-ft Laterals 

Existing B.E. Payne 

Water Treatment Plant 

15+0 10+0 5+00 0+00 

Figure 1. Soft ground tunnel concept (Option 1). 

Section View 1B 

large-diameter bucket auger was considered the most appropriate sample collection method. 

Grain-size distribution curves were developed based on sieve analyses of the bucket auger samples. 

The results of the rock-core boring program indicated the presence of limestone and shale. A 

particularly good layer of limestone occurred from 130- to 157-ft (40- to 48-m) below ground 

surface. Limestone beds in this unit tend to be on the order of 2- to 3-ft (0.6- to 0.9-m) thick, 

interbedded with 3- to 4-inch (7- to 10-centimeter) thick layers of shale. All of the limestone in 

this layer has a rock quality designation in the range of 90 to 100 percent, which according to 

Deere and Deere (1989) is classified as “excellent.” 

9

10 

Ohio River 

Pump Station 

Collector Well 


Drop Shaft 

4-ft Diameter 

Based on the results of the subsurface investigation, the construction of either type of tunnel 

system (soft ground or bedrock) was determined to be feasible. 

Maintenance Considerations 

Collector 

Well 

Collector 

Hard Rock 

Hard Roc Rock Tunnel unnel 

Lagoon 

200-ft 

Diameter Laterals 

11 per Collector Well 

7-ft Diameter 

Tunnel 

Existing 


Lagoon 

Existing 48-Inch 

Raw Water 

Transmission Line 

Lagoon 

Lagoon 

Pump Station 

32-ft Diameter Shaft 

Existing (2) 60-Inch Diameter 

Water Intake Pipes 

New 54-Inch Raw Water 

Transmission Line 

200-ft 

Diameter Laterals 

11 per Collector Well Sand and 

Gravel 


Shaft 

in Bedrock 



8+00 4+00 0+00 

Figure 2. Hard-rock tunnel concept using horizontal collector wells (Option 2). 



Ground 

Surface 

Bedrock 

Plan View 2A 

One of the unique challenges of Option 1 would be the long-term maintenance of collector 

laterals. Over time, it is anticipated that the laterals would need to be cleaned to restore their 

yield, as would be the case with any well screen. Several options were considered that would allow 

individual laterals to be taken out of service for cleaning while the rest of the tunnel system 

remained in operation. The selected option consisted of a “wet/dry” tunnel system, with water 

being conveyed through carrier pipes in the “wet” side and O&M being performed from within 

Clay 

Drop Shaft 

4-ft Diameter 

Section View 2B

TBM Exit 

Construction 

Vertical Collectors 

Installedon 200-ft Centers 

RBF Pump Station 

Ohio River 



Capped Well Capped Well 

Clay 

Sand & 

Gravel 

Aquifer 

Bedrock 

Well Casing 

Water Table Level 

Wellscreen 

Tunnel 

Figure 3. Hard-rock tunnel concept using vertical collectors (Option 3). 

TBM Construction 

Existing Collector 

Well/Pump Station 

Approx. 2,000 ft 

Tunnel Extension 

(no wells) 

Existing Low Lift 

Pump Station 

Plan View 3A 

the “dry” side. This construction option would allow laterals to be maintained individually, while 

all other laterals remain in service. Figure 4 illustrates the “wet/dry” tunnel concept. 

Maintenance requirements for Option 2 would involve taking an entire caisson out of service for 

lateral rehabilitation. This concept would reduce the effective capacity of the overall system by 

the capacity of an entire caisson, or approximately one-third of a three-caisson system. 

Option 3 would consist of individual wells that could be individually taken out of service and 

rehabilitated. A packer could be used to isolate the vertical collector from the tunnel. Based on 

maintenance considerations, Option 3 was preferred. 

3B 

To Pump 

Station 

Section View 

11

12 

Figure 4. Wet/dry tunnel concept. 

Hydrogeologic Yield 

Computer modeling was used to predict the safe yield of the aquifer and to optimize the placement 

of collectors for all of the options. For Option 1, collector laterals would be positioned 60 ft 

(18.3 m) on center on the riverside of the tunnel (Schafer, 2000). Using a 1,500-ft (457-m) long 

tunnel with laterals located on 60-ft (18.3-m) centers resulted in an estimated yield ranging from 

44 to 61 MGD (167,000 to 231,000 m 3 /d), depending on water temperature. For Option 2, the 

optimum number of laterals per collector well was determined to be 10 to 11 (Schafer, 2000) and 

the estimated yield ranged from 37 to 51 MGD (140,000 to 193,000 m 3 /d), depending on water 

temperature, for a system consisting of three new collector wells. For Option 3, approximately 

30 vertical collectors would be installed approximately 200-ft (67-m) apart, resulting in an 

estimated yield ranging from 38 to 44 MGD (144,000 to 167,000 m 3 /d) (Shafer, 2002). 

Data from the existing collector well and pumping tests were used as the basis for selecting 

leakance and transmissivity values that were used in calculating the predicted yields. Figure 5 

shows measured leakance data from the existing collector. As shown, leakance declined over the 

initial 2-year period of operation from over 2.0 to 0.148 inverse days due to clogging in the 

riverbed. Within the past year, measured leakance values appear to have stabilized and increased 

slightly. To be conservative in predicting yields for the new RBF addition, the lowest observed 

leakance value was used. 

Water Quality and Treatment 

Tunnel Cross-Section 

Gate Valves with 

Removable Elbow 

Precast 

Concrete 

Tunnel Segments 

12-Inch Lateral 

Concrete Fill 

Two 48-Inch Carrier Pipes 

Water quality obtained from the collector wells in many locations has been documented as having 

many advantages over that obtained directly from surface-water sources (Sontheimer, 1980; 

Hubbs, 1981; Wang et al., 1995; National Water Research Institute, 1999; Kuehn and Mueller, 

2000). In the Louisville Water Company’s case, the existing RBF collector well has shown many

Leakance (inverse days) 

10 

1 

0.1 

10/1/1999 3/31/2000 9/30/2000 3/31/2001 9/30/2001 3/31/2002 9/30/2002 

favorable improvements in water quality compared to the Ohio River source (Wang et al., 2002), 

including: 

• Lower turbidity. 

• Lower TOC. 

• Increased minimum temperature. 

• Reduced microbial contaminants. 

Date 

Figure 5. Measured leakance values corrected to 68-degrees Fahrenheit (October 1999). 

Table 1 summarizes typical water quality from the collector well compared to local groundwater 

and Ohio River sources. 

Table 1. Characteristics of River Water and Groundwater in the Project Area 

Infiltrated Bedrock 

Parameters River Water Groundwater Groundwater 

pH 1 7.7 to 7.9 7.4 to 7.5 7.2 to 7.3 

Total Hardness (mg/L) 1 90 to 205 280 to 290 530 to 582 

Total Alkalinity (mg/L) 1 50 to 110 235 to 250 260 to 280 

TOC (mg/L) 1 2.1 to 4.9 0.3 to 0.6 0.4 to 0.7 

Turbidity (ntu) 1 2 to 1,500 5.0

14 

The turbidity of well water is typically less than 0.08 nephelometric turbidity units (ntu), which 

could allow the Louisville Water Company to bypass the coagulation treatment process and use 

direct filtration for treating water. Reduced turbidity will result in lower coagulant dosages and less 

sludge production. The reduction of organic material in the water will provide lower levels of 

DBPs in the finished water. 

With minimum water temperatures of the Ohio River source near freezing, water main breaks are an 

issue during the winter months. By having a more moderate water temperature and using more RBF 

water, this will ultimately result in fewer main breaks. The Louisville Water Company also noted the 

lack of Geosmin and 2-methylisoborneol in well water during a taste and odor episode that occurred 

in 1999, resulting in the ability to eliminate the need to add powdered activated carbon. 

Well water, however, has other water-quality issues that need to be addressed in the design of the 

system. Data from the Louisville Water Company shown in Table 1 indicates the following 

treatment issues: 

• Increased hardness. 

• Presence of radon. 

• Low dissolved oxygen. 

Additionally, water quality of the aquifer downstream from the Payne Plant has been documented 

as having elevated levels of iron (Schafer, 2000). This will be a consideration as the Louisville 

Water Company proceeds with implementing RBF for the Crescent Hill Water Treatment Plant, 

but is not a factor for the current design project at the Payne Plant. Because the Payne Plant is 

already a softening plant, the increased hardness of RBF water will be treated using existing basins. 

Radon levels of 180 pCi/L have been noted in the existing collector well. While this is below the 

anticipated maximum contaminant level of the future Radon Rule (300 pCi/L), the Louisville 

Water Company desires to set a goal of 0 pCi/L, because the surface-water source currently 

supplying water to its customers does not contain radon. Treatment for radon will involve the 

installation of an in-line aerator. 

Dissolved oxygen from well water is less than 0.1 mg/L. Currently, as water flows through 

the basins of the existing treatment plant, oxygen levels are naturally increased up to about 

5 mg/L. The same in-line aerator used for radon removal can be used to impart higher levels of 

oxygen into the water, if needed. 

Summary 

The results of the subsurface evaluation show that both tunnel options are feasible for RBF. With 

either Option 1 or 3, the collector laterals would be evenly spaced along the entire length of the 

collection system, providing for more even hydraulic head distribution across the riverbank and, 

thus, better overall yield from the installation. With Option 2, the hydraulic head would be 

focused at the collector wells, resulting in less than maximum yield from the aquifer system. 

Maintenance flexibility would be greater for either Option 1 or 3, which would allow each 

lateral/vertical collector to be maintained independently of the operation of the remaining 

laterals/vertical collectors. Cost was a determining factor in selecting Option 3. In optimizing the 

design of the selected RBF system, the design engineers considered both water-quantity and waterquality 

issues. Water-quantity predictions were made using conservative assumptions based on 

data from the existing collector well. Water-quality issues were identified and appropriate 

treatment technologies are being incorporated into the design.

REFERENCES 

Deere, D.U., and D.W. Deere (1989). “Rock Quality Designation (RQD) After Twenty Years.” Contract 

Report No. 6L-89-1, U.S. Army Corps of Engineers. 

Hubbs, S.A. (1981). “Organic reduction — Riverbank Infiltration at Louisville.” Proceedings, American 

Water Works Association Annual Conference, St. Louis, Missouri. 

Kuehn, W., and U. Mueller (2000). “Riverbank Filtration: An Overview.” Journal AWWA, 92(12): 60. 

National Water Research Institute (1999). Abstracts, International Riverbank Filtration Conference, November 

4-5, 1999, Louisville, Kentucky. 

Schafer, D. (2001). Evaluation of Collector Well Production Capacity B.E. Payne Water Treatment Plant, David 

Schafer & Associates, Inc., Project Report. 

Schafer, D. (2000). Hydraulics Analysis of Groundwater Extraction at the B.E. Payne Water Treatment Plant, 

David Schafer & Associates, Inc., Project Report. 

Sontheimer, H. (1980). “Experience with Riverbank Filtration Along the Rhine River.” Journal AWWA, 

72(7): 386. 

Wang, J., J. Smith, and L. Dooley (1995). “Evaluation of Riverbank Infiltration as a Process for Removing 

Particles and DBP Precursors.” Proceedings, American Water Works Association Water Quality Technology 

Conference, New Orleans, Louisiana. 

Wang, J.Z., S.A. Hubbs, and R. Song (2002). Evaluation of Riverbank Filtration as a Drinking Water Treatment 

Process, American Water Works Association Research Foundation Report Number 90922. 

Wang, J.Z. (2003). Personal communication. 

DAVID HAAS is a Senior Project Manager with Jordan, Jones & Goulding, Inc., an 

Atlanta-based consulting firm that offers engineering, management, and planning 

services. Haas has over 18 years of experience in municipal water supply, treatment, and 

distribution system projects. Currently, he is the Project Manager for the Louisville Water 

Company’s 45-million gallons per day riverbank-filtration project at the B.E. Payne Water 

Treatment Plant in Louisville, Kentucky. He is also a contributing author of Riverbank 

Filtration: Improving Source-Water Quality, jointly published by Kluwer Academic 

Publishers and the National Water Research Institute in 2002. Haas received both a B.S. and M.S. in 

Environmental Engineering from the University of Louisville. He is a Professional Engineer in the States of 

Georgia, Kentucky, and Tennessee. 

15


Construction and Maintenance of Wells 

for Riverbank Filtration 




By description, RBF implies that we are designing something that will allow water to be infiltrated 

from a surface-water source through riverbank (and river-bottom) deposits. This allows physical 

(e.g., suspended) particles to be filtered out of source water in an attempt to “pretreat” raw water 

before it reaches a treatment plant or enters a distribution system with some sort of primary 

treatment, such as chlorination. 

While the term “RBF” is a relatively new term in the United States, well and gallery systems have 

been used in international settings to develop water supplies using induced infiltration dating back 

to the 1800s. Many well and infiltration systems have relied on RBF to provide recharge into 

alluvial aquifers to replace groundwater pumped at sites all across the United States, but only 

recently have these sites been coined as “RBF” sites. Water-supply facilities at such sites typically 

include vertical wells, infiltration galleries, and radial collector wells. 

What makes RBF work is that the water level in the aquifer is drawn down by pumping from a 

well or gallery system located adjacent to a surface-water source, and water from the surface-water 

body is then induced to infiltrate into the aquifer as hydraulic gradients are developed from 

pumping. The schematic in Figure 1 shows this general relationship for horizontal collector wells 

and conventional vertical wells. 

Plan 

Elevation 

Screened 

Pipe 

River River 

Horizontal Vertical 

Figure 1. Well systems develop capacity through induced infiltration (Ray et al., 2002a). 

Flow 



Senior Project Manager/Hydrogeologist 


6360 Huntley Road • Columbus, Ohio 43229 USA 

Phone: (614) 888-6263 • Fax: (614) 888-9208 • Email: hchunt@collectorwellsint.com 

Flow 

17

18 

Riverbank Filtration Suitability 

A RBF system is designed to infiltrate water from an adjacent surface-water source, using streambed 

and riverbank deposits to naturally filter out suspended materials from source water. The first (and 

obvious) requirement is that the facilities be placed in close proximity to a source of recharge, such 

as a river. During the feasibility and siting stages of a project, a number of criteria must be 

considered, including: 

• Availability of water from a surface-water source that can recharge the aquifer. 

• An efficient hydraulic interconnection between the river and aquifer. 

• Suitable water quality in the surface-water source. 

• Sustainable flow in the river to match anticipated withdrawal rates. 

A detailed hydrogeologic investigation is required to verify that aquifer conditions are favorable 

for considering a RBF facility to meet project water demands. The investigation must evaluate the 

geology of the aquifer and the interconnection between the river and groundwater in the aquifer 

to determine the feasibility of inducing infiltration from the river, evaluate possible well designs, 

and determine likely well yields. This type of investigation typically includes exploratory test 

drilling, aquifer (pumping) testing, and data analysis to develop the needed information for each 

prospective project site. Based on the results of this investigation, well designs are developed and 

alternatives are compared for feasibility, yield, and cost. These alternative designs are generally 

discussed below. 

Vertical Wells 

Vertical wells represent the most conventional method for developing a groundwater supply in the 

country. These wells consist of a vertical borehole that is usually completed with a screen and riser 

casing to allow water to enter from the formation and be pumped from the borehole via a pumping 

system installed within the riser casing. A diagram of a typical vertical well constructed in an 

unconsolidated (e.g., sand and gravel) aquifer is shown in Figure 2. Vertical wells can be used 

effectively when small to moderate yields are needed, or when a system is growing slowly over a 

longer period of time, such that adding a well to the system every year or so meets growing water 

demands. To meet larger demands, a series of vertical wells must be installed, spreading along 

riverfront areas or grouped into well-field clusters. 

Vertical wells can be constructed using a variety of drilling methods, including mud rotary, reverse 

circulation, cable tool, bucket-auger, dual-rotary, and air rotary. The method selected for each site 

will take into account a combination of well criteria, including well depths, diameters, water-table 

elevations, screen requirements, and potential problems caused by geologic conditions 

encountered. These wells are constructed by drilling a vertical borehole, and then installing the 

desired well riser casing and screen in the borehole. In some cases, an artificial gravel-pack filter 

is also added to help prevent the intrusion of sand and silt from the formation during pumping. 

Collector Wells 

A collector well, also called a horizontal collector or radial well, differs from a vertical well in that 

the well screen is installed horizontally into the aquifer formation from a central reinforced 

concrete caisson that serves as the wet well pumping station. These wells are constructed by 

sinking sections of reinforced concrete (called lifts) into the aquifer adjacent to the river until the 

lower portion of the caisson reaches the design elevation for installing the well screen. Individual

Land 

Surface 

lengths of well screens are then projected out into the aquifer in a variety of patterns near the base 

of the alluvial aquifer or at another pre-selected horizon within the aquifer. Where RBF is desired, 

the pattern of lateral well screens is predominantly beneath the river. This allows the “pumping 

center” for the collector well to be shifted closer to the river, usually in an attempt to increase the 

percentage of river water that is infiltrated into the aquifer. This design capability usually permits 

the highest percentage of river water to be achieved in a riverbank setting. The ability to install 

the well screens horizontally in the aquifer beneath the river permits longer lengths of well screen 

to be installed, per site, typically maximizing the yield possible from each well site. It is common 

for a collector well to produce a yield equivalent to multiple vertical wells from the same well site. 

A general schematic of a typical collector well is shown in Figure 3. 

Well Selection 

Water 

Pump to 


Treatment 

Plant 

Well 

Casing 

Pump House 

Electric Motor 

Approximately 

1-m High 

Rotating 

Shaft 

Cone of Depression 

Pump Bowl 

for a Turbine Pump 

Well Screen 

Gravel Pack 


Figure 2. Typical components of a vertical well (Ray et al., 2002b). 

Original Water Surface 

The hydrogeologic investigation determines the hydraulic characteristics of the aquifer formation 

necessary to determine the potential yield possible from either a collector well or a series of 

vertical wells. This data allows a comparison to be made of well designs to meet project water 

demands, usually considering a single collector well versus a series of vertical wells, with 

connecting pipeline, electrical service, access roads, etc., to produce the equivalent yield (it is 

common for a collector well to produce a yield equal to 5 to 10 vertical wells). As the capital costs 

to install the “complete system” are compared, it is often very competitive to consider a collector 

well to meet moderate to very large capacities, and more competitive to consider a vertical well 

when lower capacities are required (for example, when incremental increases in capacity are 

projected over a number of years). 

River 

19

20 

Laterals 

Alternate Systems 

In addition to vertical wells and radial collector wells, infiltration galleries can be used to develop 

filtered surface-water supplies through RBF. These can include trenching parallel or beneath the 

river and installing screened gallery pipes that can deliver filtered surface water to sumps and wet 

wells. It is also possible to use horizontal directionally drilled wells to induce infiltration from an 

adjacent river. These systems are often installed under low head conditions, such that lower per 

foot yields are obtained, requiring longer gallery lengths to meet higher capacity needs. It is often 

difficult to perform effective well-screen maintenance on these systems. 

Well Maintenance 

Pump Shaft 

Pump 

Pump House 

Central 

Collection 

Caisson 

Well Screens 

(in Laterals) 

Figure 3. Typical components of a radial collector well (Ray et al., 2002b). 

Land Surface 

It is normal for well screens to become plugged with chemical (mineral) precipitates and biological 

growths (e.g., iron bacteria) over time in alluvial aquifers. The rate of plugging can be exacerbated 

if the screen design creates excessive entrance velocities through the slot openings in the well 

screens, so it is important that proper screen design considers the water quality and hydraulic 

characteristics of the aquifer to maintain entrance and approach velocities within acceptable 

ranges to minimize this rate of plugging and extend the intervals between well cleanings. Since 

the lineal footage of well screen in a collector well is longer than for a vertical well, entrance 

velocities are minimized so that the intervals between required maintenance can be extended. 

This typically results in lower O&M costs over the life of the collector well. 

Well maintenance can be accomplished using a variety of methods including mechanical, 

chemical, or a combination of methods. The most effective method to restore well-screen 

openings and well efficiency will vary from well field to well field, and is selected based upon 

details of the well construction (e.g., screen design), groundwater quality, past results, the nature 

of plugging (mineralogic versus biological), and other factors. The use of an ongoing monitoring 

and record-keeping program should allow you to track well performance, identify operating trends, 

and predict optimal times for performing maintenance.

Summary 

Well systems can be constructed in alluvial aquifer systems at many locations to induce infiltration 

from adjacent surface-water sources to provide a pre-filtered raw-water supply. The design of these 

systems should be selected after evaluating site-specific aquifer conditions, water-quality 

objectives, and project water demands to ensure that the most effective system is selected. There 

are many well systems operating in the United States and overseas, demonstrating that RBF can 

be used effectively to meet water-quality and capacity objectives. 

REFERENCES 

Ray, C., T. Grischek, J. Schubert, J. Wang, and T. Speth (2002a). “A perspective of riverbank filtration.” 

Journal AWWA, 94(4): 149-160. 

Ray, C., G. Melin, and R. Linsky, editors (2002b). Riverbank Filtration: Improving Source-Water Quality, 


HENRY HUNT is a Senior Project Manager and Hydrogeologist with Collector Wells 

International, Inc., which specializes in the design, construction, and rehabilitation of 

water-supply systems for public drinking water, industrial process, or power plant cooling 

water. Hunt has over 25 years experience involving riverbank-filtration systems, ranging 

from the inspection and rehabilitation of existing well and infiltration systems to the 

design and construction of new water wells and riverbank-filtration facilities. He has 

special expertise concerning the inspection, evaluation, siting, and testing of radial 

collector wells along many alluvial valleys across the United States and overseas. In addition, he has authored 

a number of papers regarding water-supply development, infiltration galleries, radial collector wells, and 

seawater collector wells, including several chapters in the 2002 publication Riverbank Filtration: Improving 

Source-Water Quality and the newly revised American Water Works Association Manual 21 on Groundwater. 

Hunt received a B.A. in Geology from Lafayette College in Pennsylvania, with a minor in Civil Engineering. 

21


Aquifer Storage and Recovery Pretreatment: 

Synergies of Bank Filtration, Ozonation, 

and Ultraviolet Disinfection 

Robert Stanley Cushing, Ph.D., P.E. 

Carollo Engineers 

Sarasota, Florida 

R. David G. Pyne, P.E. 

ASR Systems 

Gainesville, Florida 

David R. Wilkes, P.E. 

Jordan, Jones & Goulding 


Introduction 

Aquifer Storage and Recovery (ASR) is a technique in which water is stored in subsurface aquifers 

during high-supply, low-demand periods for use during low-supply, high-demand periods. As 

indicated in Figure 1, water is pumped into a confined aquifer, displacing the native water, and 

forms a zone in the aquifer comprised primarily of injected water. Water is subsequently extracted 

from the aquifer when needed to supplement finished water-production capacity. 

Native 

Ground 


Buffer 

Zone 

Stored 


ASR Well 

Confining Layer Confining Layer 

Figure 1. Schematic representation of ASR. 

Stored 


Target Storage Volume 

Confining Layer 


Robert S. Cushing, Ph.D., P.E. 

Partner 

Carollo Engineers, P.C. 

401 North Cattlemen Road • Suite 306 • Sarasota, Florida 34232 USA 

Phone: (941) 371-9832 • Fax: (941) 371-9873 • Email: RCushing@carollo.com 

Buffer 

Zone 

Native 

Ground 


23

24 

Traditionally, ASR has been used to store potable water to meet seasonal limitations in source-water 

supply or to optimize the use of water-treatment capacity. For example, the Peace River Regional Water 

Supply Authority in Florida draws and treats water from the Peace River at a rate limited by minimum 

flows and levels. By storing treated water through ASR during high river flow periods, the Authority 

has been able to more than double the usable yield from this source, while avoiding adverse impacts to 

sensitive downstream ecosystems. Mt. Pleasant Water Works in South Carolina supplements their 

water supply using reverse-osmosis treatment of brackish groundwater. ASR allows the reverse-osmosis 

plant to operate at a relatively constant rate, minimizing the production cost from this facility. 

Emerging ASR applications include storing water types other than finished potable water, including 

partially treated surface water and reclaimed wastewater, as well as different uses for ASR production 

water (such as water-treatment plant source-water, irrigation, and environmental applications). 

Biological and physiochemical processes during ASR storage can remove certain contaminants, 

including THMs and HAAs, leading to ASR applications that provide treatment, as well as storage. 

Water-quality goals for water to be injected into an ASR well vary with application. Generally, the 

least restrictive goal is to produce a water with chemical and particle characteristics that prevent 

excessive loss of permeability in the injection well aquifer. Some applications require injection water 

to meet primary (or public health-related drinking water) standards to protect aquifer quality or to 

meet ASR production requirements. Treatment to secondary or aesthetic water-quality standards may 

also be required. The most restrictive water-quality goals require treatment for trace, unregulated 

contaminants (e.g., low levels of endocrine disrupting compounds). Treatment system complexity, 

O&M requirements, and cost are all strongly related to source-water quality and finished water-quality 

goals, with ASR pretreatment costs ranging well over an order of magnitude, depending on the 

particular application. 

One general treatment approach for meeting ASR pretreatment goals consists of bank filtration 

followed by ozonation and/or UV disinfection (Figure 2). This process combination offers the 

improved reliability of a multiple-barrier approach while maintaining cost and operational 

efficiencies due to the synergistic nature of the individual unit processes. The following discussion 

provides a summary of treatability study results, engineering analysis, and cost-estimates for this 

process configuration. 

Lake Okeechobee 

Figure 2. Integrated bank filtration/ozonation/UV disinfection process train. 

Discussion 

Bank Filtration 

Ozone Contactor/ 

Reaction Basin 

UV Disinfection 

Traditional treatment approaches, such as membrane filtration or conventional treatment, 

produce process residuals that generally have substantial cost and operational implications. In 

addition, these traditional approaches require substantial chemicals for process O&M, an aspect

that is important when considering both reliability and security. A process train comprised of bank 

filtration followed by ozonation and/or UV disinfection produces no residuals and has minimal 

chemical requirements. This process combination offers a number of benefits, including cost, 

flexibility, and multiple barriers to pathogens and other contaminants. 

For this process combination, the barriers to pathogens operate by distinctly different mechanisms, 

enhancing the robustness of the multiple-barrier approach and providing the first example of 

process synergy. Bank filtration provides physical removal and physical/chemical/biological 

inactivation of pathogens, ozonation provides chemical disinfection, and UV provides 

disinfection through a fundamentally different mechanism. 

While bank filtration removes color and taste- and odor-causing compounds, ozonation provides 

a means of further improving these water-quality characteristics. Ozonation also increases 

UV transmittance, decreasing the capital and operational cost of the UV disinfection process. 

Case Study Description and Objectives 

As part of the Comprehensive Everglades Restoration Program, the U.S. Army Corps of Engineers 

— in conjunction with the South Florida Water Management District — have initiated pilot 

studies to evaluate potential treatment options for an ASR project that includes hundreds of ASR 

wells. Treatment systems will be designed to treat surface water prior to injection into groundwater 

storage zones, with ultimate capacity approaching 1.5-billion gallons of water per day. 

The purpose of the pilot-testing project was to demonstrate surface-water treatment technologies 

to meet multiple water-quality goals prior to injection into ASR wells. Pilot-test data and 

engineering analysis were used to develop optimized design criteria, as well as capital and O&M 

costs for full-scale treatment systems. The specific primary objectives of the study were to: 

• Determine raw-water quality during the pilot study. 

• Collect and analyze unit process and treatment train performance data. 

• Establish process feasibility and design criteria for full-scale water treatment systems. 

• Determine preliminary capital and operating cost estimates for each of the unit processes 

considered. 

Methodology 

Two parallel treatment trains were operated during most of the study: 

• Treatment Train 1 — Simulated bank filtration/ozonation. 

• Treatment Train 2 — Simulated bank filtration/UV disinfection. 

During later stages of the study, two additional trains were tested. Simulated bank filtration was 

bypassed and a direct surface-water intake (with wire mesh filtration followed by ozonation and 

UV disinfection) was evaluated. Figure 3 illustrates the treatment trains, with dashed lines labeled 

“Alternative Treatment Train” showing the supplemental testing. 

Water was taken from the St. Lucie Canal (eastern shore of Lake Okeechobee) through an intake 

header that consisted of slotted polyvinyl chloride pipe. The water was pumped from the intake 

pipe through a centrifugal pump and into the simulated bank-filtration system, consisting of a 

wedge-wire filter (Parker TruClean filtration system) followed by a gravity sand filter 

(10-ft × 10-ft area, with 36-inches of 0.45 to 0.55-millimeter diameter sand). Note that original 

25

26 

plans for a pilot-scale bank-filtration intake were changed by the U.S. Army Corps of Engineers 

and South Florida Water Management District. Simulated bank-filtration results were analyzed 

directly and extrapolated to consider the impacts of full-scale bank-filtration intake. 

Results 

Decant 

Water Source Submersible 

Pump 

Figure 3. Pilot process flow schematic. 

Residual 

Wedge Wire 

Filter 

Settling Tank 

Residual 

Sand Filter 

Solids Land 

Applied On-Site 

Alternate Treatment Train 

Holding 

Tank 

Alternate Treatment Train 

Raw-water quality during the study (August 2002 to October 2002) was generally consistent with 

historical trends. TOC levels were high, averaging 27 mg/L, with a maximum concentration of 

49 mg/L. Correspondingly, true color was also elevated, averaging 96 color units, with a maximum 

of 236 color units. Turbidity was variable, ranging from 5 to 57 ntu, with an average of 15 ntu. 

The simulated bank-filtration unit process provided particle removal averaging 91 percent and 

turbidity removal averaging 78 percent. The removal of dissolved constituents (e.g., TOC and 

color) was minimal. As expected, due to the short residence time in the simulated bank-filtration 

system (1 to 3 hours), particle removal through the simulated system was less than expected 

through a full-scale bank-filtration system, and microbial activity and the associated removal of 

organic material was not significant. 

Ozone was capable of removing color to meet and exceed the secondary maximum contaminant 

level of 15 color units and to reduce UV absorbance. Due to the high ozone doses required to 

maintain a residual and rapid decay rates, achieving disinfection credit through ozonation was not 

feasible. In addition, with bromide concentrations approaching 300 micrograms per liter (µg/L), 

achieving significant disinfection and maintaining bromate concentrations below the maximum 

contaminant level of 10 micrograms per milliliter would be infeasible; however, ozone doses used 

to reduce color and improve UV transmittance (from a minimum of 13-percent UV transmittance 

in raw water to greater than 65-percent UV transmittance in ozonated water). 

Due to the synergistic relationship between ozone and UV disinfection, full-scale design criteria 

for the two processes were analyzed concurrently to minimize water production costs. As ozone 

dose goes up, the cost to purchase and operate the ozone system rises; however, due to the increase 

in UV transmittance with higher doses, UV-disinfection system costs decrease. An average ozone 

dose of approximately 6 mg/L and maximum dose of 11 mg/L produced a UV transmittance of 

65 to 72 percent and a minimum production cost (amortized capital, plus O&M). The UV system 

UV 

O 3 

Effluent 

to Canal

was designed to deliver a dose of 140 millijoules per square centimeter to meet all disinfection 

goals for Giardia, Cryptosporidium, and viruses at a design UV transmittance of 60 percent. 

The cost of a 5-MGD bank filtration/ozonation/UV-disinfection treatment facility is approximately 

$4,759,000 for design and construction and $147,000 per year for O&M (or $0.20 per 1,000 gallons 

of water produced). This translates into a normalized capital cost of $0.95 per gallons per day of 

capacity and a production cost of $0.70. 

BOB CUSHING has 14 years of experience in applied environmental science and 

engineering. Throughout his career, he has coupled fundamental concepts with sound 

engineering practices to provide creative, innovative, and enduring solutions to challenges 

faced by water and wastewater utilities. Cushing has been responsible for numerous 

successful treatment facility planning and design projects, as well as studies and programs 

for improving distribution system water quality. Cushing has practiced nationally, 

providing service to a broad cross-section of the industry, from some of the largest and most 

visible utilities (e.g., New York City and Washington, D.C.) to very small applications with important and 

unique issues (e.g., Oray National Fish Hatchery, Utah). He has also been responsible for introducing and 

applying advanced technologies, most notably ultraviolet disinfection for potable water treatment. An 

internationally recognized expert in ultraviolet disinfection, he is responsible for seminal applied research in 

this area, and is project manager and a primary author for the United States Environmental Protection 

Agency’s Ultraviolet Disinfection Guidance Manual. Cushing received a B.S. in Petroleum Engineering and an 

M.S. and Ph.D. in Civil Engineering from the University of Texas at Austin. 

27


Evolution from a Conventional Well Field 

to a Riverbank-Filtration System 

John D. North 

Cedar Rapids Water Department 

Cedar Rapids, Iowa 

Douglas J. Schnoebelen, Ph.D. 


Iowa City, Iowa 

Shelli Grapp 



Roy E. Hesemann 



The City of Cedar Rapids is located in east-central Iowa and has a population of 121,000. The 

Cedar River alluvial aquifer is the sole source of drinking water for the City. Within the last few 

years, the Cedar River — which drains a major agricultural watershed — has experienced extended 

periods of elevated levels of nitrate and herbicides (notably, atrazine and its degradation 

byproducts) that were above the maximum contaminant level for drinking water. Fortunately, 

during these extended periods of compromised water quality in the river, RBF enabled the Cedar 

Rapids Water Department to obtain and treat adequate quantities of water that met all drinkingwater 

standards. The Cedar Rapids Water Department and the United States Geological Survey 

(USGS) have been collaborating in an ongoing study of the river and its hydraulic interconnection 

with the City’s wells. A primary focus of this study is to identify operational and development 

strategies that will optimize RBF and the water-quality benefits it affords. 

The Cedar River originates in southern Minnesota and generally flows in a southeasterly direction 

through east-central Iowa until it discharges to the Iowa River in Louisa County. The Cedar River 

drains a major agricultural watershed, which also includes several urban areas. The entire 

watershed encompasses 7,800 square miles, in which approximately 6,500 square miles are 

upstream from Cedar Rapids’ well fields. Land use in the watershed is predominantly agricultural 

(approximately 90 percent), with major crops being corn and soybeans (Kalkhoff et al., 2000; 

Schnoebelen and Schulmeyer, 1998; Schulmeyer, 1995). Major urban areas in the watershed 

include: 

• Albert Lea and Austin in Minnesota. 

• Mason City, Clear Lake, Charles City, Forest City, Waverly, Cedar Falls/Waterloo, and 

Cedar Rapids/Marion in Iowa. 


John D. North 

Water Utility Director, Cedar Rapids Water Department 

1111 Shaver Road NE 

Cedar Rapids, Iowa 52402 USA 

Phone: (319) 286-5912 • Fax: (319) 286-5911 • Email: J.North@Cedar-Rapids.org 

29

30 

The Cedar River and its water quality have been the subject of extensive studies and monitoring 

by the Cedar Rapids Water Department, USGS, Iowa Geological Survey Bureau, and other local 

agencies. These studies have documented several major water-quality issues that affect the river’s 

suitability for the three designation uses established by the Iowa Department of Natural Resources: 

• Primary contact recreation (e.g., swimming). 

• Wildlife, fish, and aquatic life. 

• Potable water source. 

Water-quality challenges come from both point and non-point sources and include excessive soil 

erosion/sedimentation, elevated levels of nitrate and phosphorous, and fecal coliform bacteria 

counts above the acceptable levels for recreational swimming. The 57-mile segment of the Cedar 

River upstream from the City’s well field to LaPorte City has been placed on the State’s list of 

impaired streams due to elevated levels of nitrate and fecal coliform bacteria. 

The most daunting challenge for the Cedar Rapids Water Department is the increasing trend of 

elevated levels of nitrate, especially in the spring and early summer. During these periods, monthly 

nitrate levels are routinely in the range of 10.0 to 14.7 mg/L as nitrogen (N). The drinking-water 

standard for nitrate is 10.0 mg/L (as N), and a single exceedance constitutes a violation. 

Fortunately, as illustrated in Figure 1, a 2- to 3-mg/L reduction in nitrate levels is generally 

accomplished as the water moves from the river to the wells. This natural reduction, combined 

with the monitoring and management of individual wells, has enabled the Cedar Rapids Water 

Department to avoid any violations of the drinking-water standard for nitrate. 

16.0 

14.0 

12.0 

10.0 

8.0 

6.0 

4.0 

2.0 

0.0 

Jan-02 

Feb-02 

Mar-02 

Apr-02 

May-02 

Jun-02 

Monthly Nitrate Results (mg/L as N) 

Highest Reported Values 

Jul-02 

Aug-02 

Sep-02 

Oct-02 

Nov-02 

Dec-02 

Jan-03 

Feb-03 

Mar-03 

Apr-03 

May-03 

Jun-03 

Jul-03 

Cedar River 

NW Water Plant 

MCL Standard 

Figure 1. Highest recorded nitrate levels: Cedar River as compared to the well water supplies to Cedar 

Rapids’ NW Water Plant (Source: Cedar Rapids Water Department). 

Prior to 1963, the City of Cedar Rapids obtained its water supplies directly from the Cedar River. 

River water underwent treatment processes similar to those used today — lime softening, 

disinfection by chloramination, filtration, and the addition of fluoride and phosphate for corrosion 

control. This provided for safe drinking water, but failed to avoid the intermittent aesthetic 

problems (taste and odor) primarily associated with elevated algae levels in the river. This 

prompted the City to construct and convert its water supplies to a series of shallow wells along the 

Cedar River.

Today, Cedar Rapids obtains all of its water supplies from a series of shallow wells constructed in 

the sand and gravel alluvium along the Cedar River. The City now has four well fields consisting 

of 45 vertical wells and four horizontal collector wells. Total current production capacity of the 

wells is generally assumed to be approximately 65 MGD, but output will vary depending on river 

level and recent operating conditions. Withdrawal rates now average about 36 MGD, with a 

maximum demand of 50 MGD. Approximately 80 percent of the water is used for grain processing 

and other wet industries. 

Vertical wells are drilled to the top of the bedrock with depths ranging from 42 to 72 ft and are 

located about 30 to 900 ft from the river. The two original horizontal collector wells placed in 

service in 1995 consist of a 13-ft diameter center column extending to a depth of approximately 

60 ft (about 2 ft above bedrock) and six lateral screens extending about 200 ft from the central 

column. The two collector wells placed in service this year are of similar construction, except for 

the use of a 16-ft diameter center column. All of the well laterals were constructed so as not to 

extend under the river channel except under high water conditions. 

The USGS and other agencies have conducted extensive studies of the Cedar River and the 

alluvium that serves as the drinking-water supply for Cedar Rapids. These studies have determined 

that the Cedar River is generally a “gaining stream” — that is, water will typically move through 

alluvium to the river; however, pumping of the wells will reverse the normal flow pattern and 

induce the infiltration of water from the Cedar River into the alluvium. The USGS found that 

induced filtration from the Cedar River supplies approximately 74 percent of the recharge to the 

City’s wells. The balance of the recharge is from the underlying aquifer (21 percent) and the 

infiltration of precipitation (5 percent). The USGS also determined that the travel times for the 

movement of water from the Cedar River to the nearest vertical well ranged from 7 to 17 days 

(Schulmeyer, 1995; Schulmeyer and Schnoebelen, 1998). 

The movement of water through sand and gravel formations due to pumping is commonly called 

induced RBF. RBF has been widely used in Europe as a pretreatment process for almost 100 years, 

and there has been recent interest and research concerning its use in North America. The natural 

filtration and biological activity that occur during the movement of water through the alluvium 

affords several water-quality benefits. These include a reduction in turbidity and microbiological 

improvements with the removal/reduction of viruses, bacteria, and protozoan organisms 

(e.g., Giardia and Cryptosporidium), as well as a reduction in the levels of nitrates, herbicides, and 

other potential contaminants. USGS studies of Cedar Rapids’ wells have confirmed that, “The 

filtering efficiency of the aquifer is equivalent to a 3-log reduction rate or 99.99-percent reduction 

in particulates” (Schulmeyer, 1995). 

When the Cedar Rapids Water Department and USGS initiated their cooperative study in 1992, it 

was restricted to a 231-square mile area along the Cedar River that encompasses the City’s four well 

fields. Initially, the primary focus was to develop an understanding of the hydraulic interconnection 

of the river and wells and the factors that affect the quantity and quality of the water drawn from the 

wells. For example, studies regarding new well development would simply attempt to identify sites 

that would yield significant water with relatively low levels of iron and manganese. There was very 

little consideration given to “RBF” and enhancing the water-quality benefits it affords. 

Beginning about 1995, the scope of the study was expanded to include additional physical and 

chemical parameters, as well as the addition of biological monitoring (e.g., Microscopic Particulate 

Analysis). This was prompted by two primary considerations. The ongoing cooperative study 

confirmed that well water was consistently of higher quality than the river; however, the study also 

confirmed that the movement of recharge water through alluvium could also readily transport 

31

32 

contaminants from the river to the wells. Additionally, Iowa regulatory officials issued a preliminary 

determination that the newly constructed collector wells were groundwater under the direct 

influence of surface water (GWUDI) sources. 

To date, major study activities and work products of the USGS/Cedar Rapids Water Department 

cooperative study include: 

• Identification and mapping of contaminant sources in the immediate vicinity of the 

City’s wells. 

• Development of a regional groundwater model covering 231 square miles, which 

simulates groundwater flow under steady-state conditions. 

• Construction of a detailed groundwater flow model, which simulates groundwater flow 

and well recharge under transient conditions (a computer model is now complete and 

being tested). 

• Completion of extensive water-quality monitoring (physical and chemical parameters) of 

the river and wells. 

This research, along with the Cedar Rapids Water Department’s Source Water Assessment study, 

has documented the importance of the river and the potential contaminant threats it poses to our 

water supplies. They have also confirmed the protection and other benefits afforded by RBF. 

Consequently, the cooperative study was recently expanded to include the entire Cedar River 

Watershed. The Cedar Rapids Water Department, USGS, and Iowa Geological Survey Bureau 

have partnered to conduct three separate comprehensive synoptic studies of water quality of 

samples at 64 locations throughout the watershed. A dye tracer/time-of-travel study has been 

completed on the lower main stem of the Cedar River, with plans for more dye tracing on other 

sections at low flow conditions, as well as possible Lagrangian sampling. In Lagrangian sampling, 

the same mass of water is sampled as it moves downstream. The objective of this study will be to 

validate time-of-travel models and to determine the fate of nitrogen compounds as they are 

transported down the river (D.J. Schnoebelen, 2002). 

In summary, the Cedar Rapids Water Department has attempted to develop a better understanding 

of the river and its wells, which would allow the implementation of management and 

protection programs for its drinking-water supplies. Although we did not fully understand its role 

until just recently, RBF is a valuable pretreatment process and is an integral component of our 

multi-barrier strategy for protecting water quality. It has enabled the Cedar Rapids Water 

Department to avoid any violations of the drinking-water standard for nitrate that would 

necessitate the construction and operation of costly nitrate removal facilities. 

Without RBF, the Cedar Rapids Water Department would not have been able to supply quality 

water that meets all drinking-water standards while maintaining the competitive rates required by 

its major customers, the local industries. Although very beneficial, it must be acknowledged that 

RBF does not completely remove contaminants, as evidenced by the presence of nitrates and 

herbicides in Cedar Rapids’ well supplies. RBF, along with watershed protection programs, must 

be a key element of Cedar Rapids’ multi-barrier approach to the protection of its water supplies 

and public health.

REFERENCES AND ACKNOWLEDGEMENTS 

The following references and publications were used in the research and preparation of this report 

(though not all are cited in the text): 


• Ray, C., G. Melin, and R.B. Linsky, editors (2002). Riverbank Filtration: Improving Source-Water 

Quality, Kluwer Academic Publishers, Dordrecht. 


• Becher, K.D., S.J. Kalkhoff, D.J. Schnoebelen, K.K. Barnes, and V.E. Miller (2001). Water-Quality 

Assessment of the Eastern Iowa Basins – Nitrogen, Phosphorous, Suspended Sediment and Organic 

Carbon in Surface Water, 1996-98, U.S. Geological Survey Water-Resources Investigations Report 

01-4175. 

• Kalkhoff, S.J., K.K. Barnes, K.D. Becher, M.E. Savoca, D.J. Schnoebelen, E.M. Sadorf, S.D. Porter, 

and D.J. Sullivan (2000). Water Quality in the Eastern Iowa Basins, Iowa and Minnesota, 1996-98, 

U.S. Geological Survey Circular 1210. 

• Schnoebelen, D.J. (2002). Written correspondence. 

• Schnoebelen, D.J., D.E. Christiansen, and K.D. Becher (2002). “City of Cedar Rapids: Source 

Water Assessment of Public Drinking-Water Supplied from Ground-Water.” 

• Schulmeyer, P.M. (1995). Effect of the Cedar River on the Quality of the Ground-Water Supply for City 

Rapids, Iowa, U.S. Geological Survey Water-Resources Investigations Report 94-4211. 

• Schulmeyer, P.M., and D.J. Schnoebelen (1998). Hydrogeology and Water Quality in the Cedar Rapids 

Area, Iowa, 1992-1996, U.S. Geological Survey Water-Resources Investigations Report 97-4261. 

Iowa Department of Natural Resources 

• “Progress Report One — Cedar River Assessment Survey,” June 2001. 

• “Source Water Assessment and Protection Program and Implementation Strategy for the State of 

Iowa,” March 2000. 

Iowa Department of Natural Resources – Geological Survey Bureau 

• Seigely, L.S., and M.P. Skopec. Written reports of three comprehensive synoptic monitoring programs 

conducted at 62 sites on the Cedar River on August 26, 2000; June 2, 2001; and June 7, 2003. 


• Grapp, S.J., J.D. North, and R.E. Hesemann (2002). “City of Cedar Rapids Source Water Assessment.” 

JOHN NORTH has 30 years of experience in water-quality monitoring and in the 

operation and maintenance of water systems. For the past 11 years, he has worked for the 

City of Cedar Rapids in Iowa, which draws its water supplies from a series of shallow 

alluvial wells located along the Cedar River and drains a major agricultural watershed. At 

present, he is the Water Utility Director for the City of Cedar Rapids Water Department. 

Under his direction, the Cedar Rapids Water Department has conducted several 

comprehensive water-quality research projects and other studies designed to ensure the 

continued availability of an ample supply of safe drinking water for Cedar Rapids. Since 1993, the utility has 

partnered with the United States Geological Survey in a comprehensive ongoing study of the City’s well 

fields and their hydraulic interconnection with the Cedar River. A major focus of the study has been to 

optimize the benefits of RBF. North received a B.S. in Chemistry from Loras College. He has dualcertification 

as a Grade IV water and wastewater plant operator. 

33

Session 3: Hydraulic Aspects 

Groundwater Flow and Water Quality – A Flowpath 

Study in the Seminole Well Field, Cedar Rapids, Iowa 



Iowa City, Iowa 

Michael J. Turco 


Lincoln, Nebraska 

John D. North 



In Iowa, alluvial aquifers near major rivers are a source of water for many communities. The City 

of Cedar Rapids withdraws water from wells completed in the Cedar River alluvium, a shallow 

alluvial aquifer adjacent to the Cedar River. The City of Cedar Rapids is located within Linn County 

in east-central Iowa, and water for the City is supplied by four well fields (East, Northwest, 

Seminole, and West well fields) along the Cedar River. The City has a population of about 121,000, 

and several large industries are major water users. Currently, per capita water usage in the City is 

nearly three times the national average. The City is committed to providing both a high quality 

and quantity of water to its customers. The USGS and Cedar Rapids Water Department have 

been working together in an ongoing research program to better understand water quality and flow 

in the Cedar River and alluvial well fields. Work has been done on both a basin and well-field 

approach and has involved dye tracing/time-of-travel studies on the Cedar River, water-quality 

sampling, geochemical modeling, and groundwater-flow modeling. 

The effect of land use in the Cedar River Basin on both surface-water and groundwater quality is 

an important issue. The Cedar River Basin upstream from Cedar Rapids is approximately 

6,500 square miles. Upstream land use in the Cedar River Basin is over 90-percent agriculture. 

Corn and soybeans are the major crops. Livestock raised in the area include beef and dairy cattle, 

as well as hogs. Runoff from agriculture is of concern, particularly during the spring and early 

summer when many chemicals are applied to cropland. Triazine and acetanilide herbicides are 

commonly applied in the Cedar River Basin, and these herbicides are water soluble and can be 

transported to streams and infiltrate to groundwater. In addition, several studies in eastern Iowa 

have identified nutrients as a major contaminant that has impaired water quality (Goolsby and 

Battaglin, 1993; Hallberg et al., 1996; Schnoebelen et al., 1999; Kalkhoff et al., 2000). In general, 

the majority of nitrogen inputs in the Cedar River Basin are from chemical fertilizers and animal 

manure (Becher et al., 2000). High nitrate levels (greater than 10.0 mg/L) in the Cedar River are 

of particular concern to municipal water operators. The lower Cedar River is listed on the Iowa 

total maximum daily load list for nitrate upstream of Cedar Rapids, Iowa. The Cedar River is the 



Research Hydrologist/Water-Quality Specialist 


Federal Building, Room 269 • 400 South Clinton • Iowa City, Iowa 52240 USA 

Phone: (319) 358-3617 • Fax: (319) 358-3606 • Email: djschnoe@usgs.gov 

35

36 

source of most nitrate detected in the Cedar River alluvial aquifer because of induced infiltration 

from the river due to pumping (Schulmeyer and Schnoebelen, 1998; Boyd, 2000). 

An unconsolidated surficial layer of glacial till, loess, and Cedar River alluvium (alluvial aquifer) 

overlies carbonate bedrock of Devonian and Silurian age (bedrock aquifer) in the study area. The 

alluvial aquifer typically consists of a sequence of coarse sand and gravel at the base, grading 

upwards to finer sand, silt, and clay near the surface. The sand and gravel contain carbonate, shale, 

and ferro-magnesium rich rock fragments. The thickness of the alluvial aquifer ranges from about 

2 to 30 m. The alluvial aquifer is recharged by the infiltration of water from the Cedar River, 

precipitation, and seepage from the underlying bedrock and adjacent hydrogeologic units. In areas 

under the influence of municipal pumping, groundwater flow is from the Cedar River toward the 

well fields; in areas outside the influence of municipal pumping, groundwater flow is toward the 

Cedar River. Results from a regional groundwater flow model indicated that approximately 

74 percent of the water pumped from the alluvial well fields is recharged from the Cedar River, 

approximately 21 percent is recharged from adjacent underlying hydrogeologic units, and approximately 

5 percent of the water is from infiltrating precipitation (Schulmeyer and Schnoebelen, 

1998). Currently, a more detailed groundwater model in the study area indicates that, in some 

places, up to 90 percent of the water pumped from the alluvial well fields is recharged from the 

Cedar River. 

The water quality in the alluvial aquifer within the well field has been characterized with samples 

collected from both monitoring and municipal wells at various times since 1992 (Boyd, 2000; 

Schulmeyer and Schnoebelen, 1998; Schnoebelen and Schulmeyer, 1996). Calcium, magnesium, 

and bicarbonate are the dominant ions. In addition, nitrate, sulfate, silica, iron, and manganese are 

present in significant concentrations in certain wells or at certain times of the year. Previous work in 

the Seminole well field indicated some detections of herbicides and their degradates (breakdown 

products) in shallow monitoring wells (3.8- to 6-m deep) completed in the alluvium as water moved 

from the river into the alluvial aquifer (Boyd, 2000). Atrazine was the most commonly detected 

herbicide in this study. Acetochlor, cyanazine, and metolachlor were also detected, but at smaller 

concentrations than atrazine. Acetanilide degradates were detected at greater frequencies and at 

greater concentrations than their corresponding parent compounds. Fewer numbers of detections of 

herbicide compounds were found in wells completed deeper in the alluvium. 

The infiltration of water with large nitrate concentrations into the alluvial aquifer from the Cedar 

River affects groundwater quality. Recent research was conducted along a flowpath to study RBF 

through a natural wetland area. Groundwater modeling helped locate the flowpath study. The 

study examined the role of a natural wetland in reducing nitrate concentrations as water moves 

from the Cedar River. A real challenge for the Cedar Rapids Water Department is the increasing 

trend of nitrate concentrations in the Cedar River. Nitrate concentrations in the Cedar River 

during the spring are often more than 10 mg/L and can reach 20 mg/L. A 2- to 3-mg/L reduction 

in nitrate often occurs as water moves from the river to the well, but in some wells, this may not 

reduce nitrate concentrations below the 10.0-mg/L maximum contaminant level. Sampling in 

wells along a flowpath occurred quarterly over a period of about 4 years. A comparison of water 

chemistry was made from water analyses from: 

• The river. 

• A monitoring well upgradient of the wetland area and river. 

• Wells in the wetland area. 

• Wells between the wetland area and river.

In addition, a comparison of water-chemistry data from a municipal well located near the wetland 

area and one located nearest the river were compared in terms of water chemistry from previous 

sampling work (Schulmeyer and Schnoebelen, 1998). Results show that nitrate concentrations 

were 4 to 6 times lower in samples from monitoring wells completed in the wetland area than in 

the Cedar River or groundwater in the upland area; however, iron and manganese concentrations 

in samples from the monitoring wells in the wetland areas were an order of magnitude higher 

when compared to the river or upland well. Water samples from the wells and the Cedar River 

generally displayed similar trends (high in the spring and low in the fall), while iron and manganese 

concentrations were more variable. 

As water moves from the river towards the monitoring wells, microorganisms obtain energy for 

metabolic processes by catalyzing the oxidation of organic matter with a progressive series of 

reducing reactions (Stumm and Morgan, 1981). Nitrate can be reduced to elemental nitrogen (N 2) 

by denitrification (Equation 1) or to ammonium (NH 4+) by reduction (Equation 2). Since 

ammonium was only detected in small quantities (less than 0.80 mg/L), denitrification most likely 

is the predominant process. 

4NO 3 – + 5CH2O + 4H + = 2N 2(g) + 5CO 2 + 7H 2O (Equation 1) 

NO 3 – + 2CH2O + 2H + = NH 4 + + H2O + 2CO 2 

(Equation 2) 

Reduction then proceeds from nitrate (NO 3 – ) to Mn +4 , Fe +3 , SO4 –2 , CO2, and N 2. The reduced 

forms of iron (Fe II) and manganese (Mn II) are more soluble in water and are more mobile than 

oxidized forms (Hem, 1985) and, under anoxic conditions, are stable. As nitrate in groundwater 

is depleted, iron and manganese reduction begins. The reduction of Fe +3 to Fe +2 and Mn +4 to 

Mn +2 from aquifer grain coatings can cause large concentrations of these ions in groundwater. 

Ferrihydrite and manganite (MnOOH) occurring as oxyhydroxide coatings on clay and silt 

particles are the most likely oxidized forms of iron (Fe +3 ) and manganese (Mn +3 and Mn +4 ) in the 

alluvial aquifer. Oxidized forms of iron and manganese might occur in the aquifer as crystalline 

minerals, such as hematite (Fe 2O 3) and hausmannite (Mn 3O 4). Iron and manganese may 

co-precipitate with carbonate minerals to cause well fouling. 

Research in the Seminole well field indicates that the location of a well in or near natural wetland 

areas may benefit from the natural reduction of nitrate concentrations, with the disadvantage of 

increased iron and manganese concentrations. Future expansions of the well fields may take 

advantage of natural wetland areas to help reduce nitrate concentrations. In Iowa, most wetlands 

have been drained, but alluvial wetlands associated with bottomland forested and oxbow lake 

areas may persist as they are subject to periodic flooding and are often not suitable for sustained 

agriculture. 

REFERENCES 

Becher, K.D., D.J. Schnoebelen, and K.K.B. Akers (2000). “Nutrients discharged to the Mississippi River 

from Eastern Iowa Watersheds, 1996-97.” Journal AWWA, 36(1): 161-173. 

Boyd, R.A. (2000). “Herbicides and herbicide degradates in shallow groundwater and the Cedar River near 

a municipal well field, Cedar Rapids, Iowa.” The Science of the Total Environment, 241-253. 

Goolsby, D.A., and W.A. Battaglin (1993). “Occurrence, distribution, and transport of agricultural chemicals 

in surface water of the Midwestern United States.” Selected papers on agricultural chemicals in water resources of 

the midecontinental United States, D.A. Goolsby, L.L. Boyer, and G.E. Mallard, compilers, U.S. Geological 

Survey Open-File Report 93-418, p. 1-25. 

37

38 

Hallberg, G.R., D.G. Riley, J.R. Kantamneni, P.J. Weyer, and R.D. Kelley (1996). Assessment of Iowa safe 

drinking water act monitoring data, 1988-1995, Iowa City, University of Iowa Hygienic Laboratory Research 

Report 97-1, 132 p. 

Hem, J.D. (1985). Study and interpretation of the chemical characteristics of natural water, third edition, U.S. 

Geological Survey Water-Supply Paper 2254, p. 264. 

Kalkhoff, S.J., K.K. Barnes, K.D. Becher, M.E. Savoca, D.J. Schnoebelen, E.M. Sadorf, S.D. Porter, and D.J. 

Sullivan (2000). Water Quality in the Eastern Iowa Basins, Iowa and Minnesota, 1996-98, U.S. Geological 

Survey Circular 1210, 37 p. 

Schnoebelen, D.J., K.D. Becher, M.W. Bobier, and T. Wilton (1999). Selected nutrients and pesticides in streams 

of the Eastern Iowa Basins, 1970-95, U.S. Geological Survey Water-Resources Investigations Report 99-4028, 

105 p. 

Schnoebelen, D.J., and P.M. Schulmeyer (1996). Selected hydrogeologic data in the Cedar Rapids area, Benton 

and Linn Counties, Iowa, October 1992 through March 1996, U.S. Geological Survey Open-File Report 

96-471, 172 p. 

Schulmeyer, P.M., and D.J. Schnoebelen (1998). Hydrogeology and water-quality in the Cedar Rapids Area, 

Iowa, 1992-96, U.S. Geological Survey Water Resources Investigations Report 97-4261, 77 p. 

Stumm W., and J.J. Morgan (1981). Aquatic Chemistry — An introduction emphasizing chemical equilbria in 

natural waters, second edition, Wiley-Interscience Publisers, New York, 780 p. 

DOUG SCHNOEBELEN is a Research Hydrologist with the United States Geological 

Survey and has worked on a variety of groundwater and surface-water projects over the last 

13 years. He has served as the groundwater specialist on the White River National Water 

Quality Assessment project in Indiana and as the surface-water specialist for the Eastern 

Iowa Basins National Water Quality Assessment project in Iowa. In addition, he has been 

the Iowa District Water-Quality Specialist since 1994, and is an Adjunct Professor in the 

Geoscience Department at the University of Iowa. His research has focused on the use of 

isotopes and bore hole geophysics in groundwater and, more recently, on the fate and transport of agricultural 

chemicals in surface water and groundwater across eastern Iowa. In particular, he has focused on the fate and 

transport of several pesticide degradate compounds. He has been involved with ongoing research in riverbank 

filtration and geochemical modeling in the Cedar Rapids well field since 1999. Schnoebelen received a B.S. 

in Geology from the University of Iowa, an M.S. in Geology from the University of Tennessee, and a Ph.D. 

in Geology, with a minor in Environmental Science, from Indiana University.


The Use of Aquifer Testing and Groundwater Modeling 

to Evaluate Changes in Aquifer/River Hydraulics at the 


David C. Schafer 

David Schafer & Associates 

Stillwater, Minnesota 

In 1999, the Louisville Water Company in Louisville, Kentucky, constructed a 20-MGD radial 

collector well as part of a pilot study to evaluate using RBF to reduce treatment costs and to ensure 

high-quality water production. The water company currently withdraws up to a total of 240 MGD 

of surface water from the Ohio River at the Zorn Pumping Station and B.E. Payne Water Treatment 

Plant (both in Louisville, Kentucky), and is exploring ways to further improve water-supply 

quality. 

A primary objective of the pilot project was to identify processes that would meet or exceed 

expected water-quality regulations. Considerations included: 

• Removing synthetic organics. 

• Reducing turbidity. 

• Eliminating taste and odor during summer months. 

• Attenuating water temperature extremes in the distribution system. 

Although several options were evaluated (such as GAC and membrane filtration), RBF was the 

only one that addressed all of the Louisville Water Company’s concerns. 

In addition to evaluating water quality, a secondary objective was to assess aquifer/river hydraulics 

and monitor the hydraulic performance over time to detect whether or not changes in capacity 

occurred. Although it was not expected that the well structure would lose hydraulic efficiency, 

there was some concern that riverbed materials adjacent to the well could clog or compact in 

response to groundwater withdrawal. 

The collector well was constructed about 120 ft from the shore of the Ohio River at the B.E. Payne 

Water Treatment Plant site. It incorporated a 16-ft inside diameter concrete caisson extending to 

a depth of about 105-ft below land surface. The caisson penetrated the glacial sand and gravel 

aquifer at the site, bottoming out on the underlying shale and limestone bedrock. Seven 12-inch 

diameter stainless steel horizontal well-screen laterals were installed near the base of the sand and 

gravel aquifer — four laterals 240 ft in length and three 200 ft in length. The four longer laterals 

were oriented toward the Ohio River, while the other three shorter laterals extended 1) upriver, 

2) downriver, and 3) away from the river. 


David C. Schafer 

President 

David Schafer & Associates 

9955 North 101st Street • Stillwater, Minnesota 55082 USA 

Phone: (651) 762-8281 • Fax: (651) 762-8335 • Email: dschafer@prodigy.net 

39

40 

Two 10-MGD capacity pumps were installed to produce water from the collector well. A variablefrequency 

driver was used to power one of the two pump motors to allow the pumping plant to 

readily adapt to changing flow and head requirements. 

In addition to the collector well, about a dozen on-land piezometers were installed around the 

collector well at distances up to 1,200-ft away. Also, a nested piezometer was installed about 

80-ft offshore, adjacent to the well site. The piezometers were used to monitor aquifer water levels 

during the testing and operation of the collector well to provide the data needed to quantify 

aquifer characteristics. 

Following well construction, a 70-day constant-rate pumping test was conducted at a pumping rate 

of 19.4 MGD. Water levels were monitored in both the collector well and observation wells. The 

maximum drawdown observed in the collector well during the test was about 27 ft. The pumping 

test data were used to quantify hydraulic properties — transmissivity and river leakance — and to 

identify baseline performance characteristics. 

The data were analyzed by constructing a groundwater flow model of the site and adjusting model 

input parameters until the model faithfully replicated water levels observed during the pumping 

test. The USGS groundwater flow code, MODFLOW, was selected for the modeling study. The 

aquifer and Ohio River were represented in a model grid covering an area 4,000 ft × 14,000 ft. The 

model domain was subdivided into 119 rows, 53 columns, and 8 model layers. The use of a large 

number of model layers permitted the realistic simulation of vertical resistance to groundwater 

flow — both water exiting the river and entering collector well laterals. Results of the pumping 

test and modeling effort revealed an aquifer transmissivity of 204,000 gallons per day per foot and 

a river leakance of 2.35 inverse days at the prevailing groundwater and surface-water temperatures. 

The collector well was used as the pumping well in five subsequent pumping tests conducted over 

a 3-year period from 1999 to 2002. Data from each pumping test were analyzed using groundwater 

modeling to update the computed values of transmissivity and leakance. The prevailing groundwater 

and surface-water temperatures were different for each test that was conducted. Because water 

temperature affects flow characteristics, it was necessary to correct the results of each test for 

temperature effects so that performance could be compared from one test to another. 

In the model calculations, a simplifying assumption was made that leakance was constant 

everywhere for any given model run. This simplification allowed a relative comparison of one test 

with another. It is known that leakance declines in a non-uniform manner, with greater reduction 

near the pumped well and less reduction away from the well; however, the exact nature of 

leakance distribution is not known, and trying to estimate it with the available data would have 

been speculative. Therefore, the assumption of a single leakance value was made. The resulting 

leakance value can be thought of as an “effective leakance” for comparison purposes. 

Throughout the period of testing, transmissivity remained constant while effective leakance 

declined steadily. Over time, the rate of leakance decline diminished and finally stabilized based on 

the last pumping test conducted in July 2002, after nearly 3 years of well operation. Measured 

leakance values, corrected for water temperature, are listed in Table 1. 

River scour associated with a bank full-flood event in Spring 2002 may have contributed to the 

apparent stabilization (actually, a slight increase in leakance). The average measured leakance in 

July 2002, corrected to the same water temperature as the original pumping test, was about 

0.18 inverse days, more than an order of magnitude lower than the original leakance value.

Table 1. Measured Leakance Values 

Date of Test Effective Leakance 

Data Acquisition (inverse days) 

October 1999 2.35 

March 2000 0.72 

October 2000 0.25 

April 2001 0.20 

September 2001 0.15 

July 2002 0.18 

The corresponding well capacity had declined by about a third in response to reduced river 

leakance. 

The reduction in river leakance is being evaluated on an ongoing basis, but is believed to be 

caused by clogging/compaction of riverbed sediments in response to collector well operation. 

Regular pumping tests and leakance evaluation will be continued in the future to further monitor 

the effects of riverbed clogging and periodic river scour associated with flood events on the Ohio 

River. The Louisville Water Company continues to do research in the area of riverbed clogging 

and leakance reduction. 

Hydrologist DAVID SCHAFER has over 30 years of experience in the groundwater industry. 

He has designed hundreds of high-capacity water supply wells, has analyzed hundreds of 

pumping tests, and is consulted regularly on well development and rehabilitation procedures. 

In addition, he has done extensive groundwater modeling using numerical models, analytic 

element models, and proprietary analytical models that he has developed. At present, he is 

President of his own company, David Schafer & Associates, which he founded in 1999. 

Schafer has published numerous articles on dewatering, well hydraulics, well development, 

and well rehabilitation, and contributed a portion of the well hydraulics section of Groundwater and Wells, 

Second Edition. Schafer received both a B.S. in Mathematics and an M.S. in Computer Science from the 

University of Minnesota. 

41


Plugging in Riverbank-Filtration Systems: 

Evaluating Yield-Limiting Factors 




The RBF process involves the passage of river water containing dissolved and suspended solids 

through the sands and gravels of a connected aquifer. In this process, the removal of suspended solids 

involves dynamics that are similar to slow sand filtration, with both biological filtration and physical 

filtration acting to remove particles. If the entrained particles are not removed, the hydraulic 

conductivity of the riverbed will decrease due to particles plugging the pores of the aquifer. 

Work done at the Louisville Water Company in Louisville, Kentucky, and elsewhere have indicated 

that the bulk of particle removal takes place in the first few feet of the aquifer, which is consistent 

with behavior in a slow sand filter. The dynamics of how a riverbed restores its filtration capacity 

are not fully understood. Anecdotal information from those operating large-capacity well fields 

indicates that these well fields require 3 to 5 years to “settle in” to their sustainable yield, which 

is usually within 50 to 75 percent of the initial capacity of the well field. Recent research at the 

Louisville Water Company has focused on developing a better understanding of the dynamics of 

riverbed plugging and predicting sustainable capacity from high-capacity RBF systems. 

Predicting Sustainable Yield 

The 20-MGD horizontal collector well at Louisville was constructed in 1999, and data on various 

parameters have been collected since that time. These data have been analyzed by traditional 

modeling techniques (Schafer, 2003) and, more recently, by regression techniques, with the 

specific interest of gaining more information about the process of riverbed plugging. 

Data indicate a cyclic pattern in specific yield, assumed to be a function of water viscosity as 

impacted by water temperature. Data also indicate a decreasing trend with time for specific yield. 

The possible causes of this decrease include: 

• Plugging of the riverbed-aquifer interface. 

• Decreased conductance of the bulk of the aquifer. 

• Decreased conductance in the area of the well screen. 

Various parameters from the entire 4-year database at the Louisville Water Company were 

subjected to regression analysis, with specific yield as the dependent variable. Variables included: 

• River stage. 

• Water elevation in the collector caisson. 



Vice President, New Technology 


550 South Third Street • Louisville, Kentucky 40202 USA 

Phone: (502) 569-3675 • Fax: (502) 569-0813 • Email: SHubbs@lwcky.com 

43

44 

• River and caisson water temperature. 

• Time (linear), the square root of time, and natural log of time. 

The influence of the driving head from the river was factored into the specific yield value by 

subtracting river stage from water level in the caisson. This eliminated the need to carry a term 

for river level in the analysis. Thus, the term for specific yield was defined from the data as: 

(Well output in MGD)/(River stage – Caisson water level). 

The square root of time was selected as a parameter based on the hypothesis that: 

• Plugging of the aquifer was the result of the deposition of suspended sediment at the 

aquifer-riverbed interface. 

• The rate of this plugging process would decrease as the recharge area extended farther 

into the river. 

• A balance between plugging and riverbed scouring would eventually be established. 

Subsequent analyses indicated that a better regression fit was realized when the natural log of time 

was used as a parameter, as opposed to linear time or the square root of time. 

When the entire data set was analyzed, it was apparent that typical trends were skewed during 

times of flooding. The database was modified to exclude flooding periods and then re-analyzed. 

Results of this analysis are presented in Figure 1. The regression model provides a good fit with 

actual data, but varies the most at the initiation of pumping and during the period of Weeks 160 

to 180, following a significant flooding event. 

Specific Yield (normalized to mean of 0.491 MGD/ft) 

0.500 

0.400 

0.300 

0.200 

0.100 

0.000 

-0.100 

-0.200 

-0.300 

Ln = Natural log. 

Figure 1. Regression with Natural Log of Time (Cleaned Data) 

SpYield (MGD/ft) = (-0.013) + .0045 (Well Temp. F) + .0018 (River Temp F) - .095 (Ln (Time week)) 

0 20 40 60 80 100 

Period (week) 

120 140 160 180 200 

Actual Data Predicted Data 

R = 0.96 

F = 583 

Specific Yield immediate following 

flood event is greater than predicted 

Period of flooding 

It is suggested that the parameter of natural log of time in this analysis is a good surrogate for the 

measure of plugging in the system. This analysis was extended to include an extrapolation of an 

additional 5 years of data, with the insertion of “jumps” representing the impact of periodic

iverbed scouring. This analysis shows the system stabilizing after 8 years of operation, with a 

specific yield cycling between 0.25 and 0.4 MGD/ft drawdown. This type of extrapolation is highly 

speculative, but fits anecdotal information on sustainable yields from high-capacity RBF systems. 

Riverbed Shear Stress as an Indicator of Riverbed Scouring 

Riverbed scouring plays an essential role in determining the sustainable yield in RBF systems. The 

shear stresses exerted on the riverbed during periods of high flow provide for the transport of 

riverbed materials and the re-suspension and transport of particles trapped during the RBF process. 

The extent of riverbed scouring can be estimated as a function of riverbed shear stress exerted 

during high-flow events. 

Streams typically exert higher riverbed shear stresses near headwaters, with decreasing stresses 

exerted near the terminus of a large water body (such as an ocean). This implies that riverbed 

scouring will decrease near the terminus of a stream, consistent with the formation of river deltas. 

Because of this tendency to deposit fine materials near the mouth of streams, these locations are 

typically not well suited for high-capacity RBF systems. 

Two techniques for estimating riverbed shear stresses were used in this analysis: 

• Measured velocity profiles across the river cross-section. 

• Stream-surface slope measurements. 

The shear stresses exerted on a streambed can also be inferred by looking at the riverbed material in 

transport during high-flow events, as indicated by the particle-size distribution of riverbed sediments. 

A compilation of particle-size data from the Ohio River in the Louisville area is presented in 

Figure 2. Data to the left represent suspended solids, while data to the center and to the right 

indicate bed material. The river at the point where these data were collected is on a bend, with 

Percent Passing 

120 

100 

80 

60 

40 

20 

Figure 2. Particle Size Distribution Analysis – Riverbed and Suspended Solids 

(USGS Data, 1979–1982, Ohio River at Louisville) 

Flow= 239,000 cfs 

TSS=408-466 mg/l 

Suspended Solids Indiana side 

Riverbed 

Flow = 301,000 to 434,000 cfs 

TSS = 482 to 698 mg/l 

Main channel and Kentucky side Riverbed 

0 

0.001 0.01 0.1 1 10 100 

Particle Size 

SS -11/30/1979 

SS -6/10/1981 

SS-1/27/1982 

bed-1375' from KY bank 










SS-6/2/82 mid-stream 

SS-6/2/82 Ind side 

TSS = Total suspended solids. cfs = Cubic feet per second. SS = Suspend solids. KY = Kentucky. Ind = Indiana. 

45

46 

the Kentucky side of the river on the outside of the bend. The prevalence of courser material in 

the riverbed on the outside of the bend reflects higher riverbed shear stress exerted at the outside 

face of the bend. 

These data indicate that the shear stresses exerted against the riverbed are capable of transporting 

riverbed media in the size range of 1 to 10 millimeters. According to available literature, this 

would require a streambed shear stress in the range of 5 Newtons per square meter. 

Stream velocity profiles and stream surface slopes of the Ohio River at high flow conditions 

(383,000 cubic feet per second) were also secured from USGS. Velocity profile data were reported 

down to about 4 ft, the approximate limit of the acoustic doppler current profiler technology used 

for this data collection effort. 

The velocity profiles used in this study for calculating boundary shear stress would be expected to 

reflect a maximum shear stress for any point in the river, as the velocity profile selected was the 

one with the highest velocity in the cross-section. The slope calculation technique, on the other 

hand, would be expected to reflect an average boundary shear stress across both the cross-section 

and length of river selected for calculation. Comparable data sets for these two methods of 

calculation revealed that the velocity profile (using the manipulation yielding the greatest shear 

velocity) provided a lower and less consistent estimate of boundary shear stress than did the 

method based on measured stream slope. These data are summarized in Table 1 from a 2000 USGS 

database: 

Table 1. Comparison of Shear Stress Calculation Methods at Three Locations 

in the Ohio River Near Louisville 

Location Profile-Calculated Shear Stress Slope-Calculated Shear Stress 

(N/m 2 ) (N/m 2 ) 

12 Mile Island 1.63 3.21 

BEPWTP 1.68 5.33 

Zorn Avenue 5.79 5.62 

BEPWTP = B.E. Payne Water Treatment Plant. N/m 2 = Newtons per square meter. 

The inconsistency and lower results from the stream profile calculations may be the results of the 

higher degree of variability obtained from the acoustic doppler measurement technique for 

velocity and, particularly, the inability for this technique to measure velocity at the lower portions 

of the stream. It is also noted that video images of the riverbed indicated surface irregularities (or 

“pocks” on the river bottom). These depressions would be expected to change the velocity profile 

as compared to a theoretical surface where friction is exerted by media roughness only. 

The slope-calculated shear stress values, on the other hand, were consistent and logical for all of 

the measured values. The higher stream flows were associated with higher shear stresses, and shear 

stresses tended to be greater where the river was deeper and narrower. 

The data from the Ohio River were compared to data from the Rhine River. The maximum 

boundary shear stress observed on the Ohio River during the high-flow event was 

7.5 Newtons per square meter. This compares to the average shear stress of the Rhine River of 

10 Newtons per square meter, as reported by Schubert (2002).

REFERENCES 

Schafer, D.C. (2003). “The Use of Aquifer Testing and Groundwater Modeling to Evaluate Changes in 

Aquifer/River Hydraulics at the Louisville Water Company.” Program and Abstracts, Second International 

Riverbank Filtration Conference, G. Melin, ed., National Water Research Institute, Fountain Valley. 

Schubert, J. (2002). “Hydraulic aspects of riverbank filtration: Field studies.” Journal of Hydrology, 266 (3-4): 

154-161. 

STEVE HUBBS is a Professional Engineer with 28 years of experience at the Louisville 

Water Company in Louisville, Kentucky. In the early 1980s, he began researching riverbank 

filtration as an alternate source of water for the Louisville Water Company, specifically 

looking at the reduction in disinfection byproduct precursors, river-borne organics, and 

mutagenicity in the riverbank-filtration process. His work continued in the 1990s with a 

focus on pathogen reduction, and his research is now being conducted on the hydraulic 

connection between the riverbed and the aquifer, with a focus on riverbed plugging 

dynamics and their influence on sustainable yields from high-capacity riverbank-filtration systems. Hubbs 

received an M.S. in Environmental Engineering from the University of Louisville, and is currently enrolled 

in the Ph.D. program in Civil and Environmental Engineering at the University of Louisville, focusing on 

the hydraulics of riverbank-filtration systems. 

47


Application of Different Tracers to Evaluate 

the Flow Regime at Riverbank-Filtration Sites 

in Berlin, Germany 

Dr. Gudrun Massmann 

Free University of Berlin 

Berlin, Germany 

Dipl. Geol. Andrea Knappe 

Alfred-Wegener-Institute Potsdam 

Pottsdam, Germany 

Doreen Richter 



Dr. Jürgen Sültenfuß 

University of Bremen 

Bremen, Germany 

Prof. Asaf Pekdeger 



Introduction 

The inhabitants of metropolitan Berlin (Germany) rely on drinking water derived from groundwater 

within the City’s boundary. Production well galleries are generally located adjacent to the surfacewater 

system and artificial infiltration ponds, and around 70 percent of abstracted groundwater is 

estimated to originate from bank filtration and artificial groundwater recharge (Pekdeger and 

Sommer-von Jarmerstedt, 1998). Berlin’s water production is a semi-closed water cycle with an 

indirect potable reuse system: local wastewater treatment plants discharge treated effluent into the 

surface-water system (Figure 1). Induced by well abstraction, surface water infiltrates into the 

ground and is used for drinking-water production. The raw water requires only minimal treatment 

to ensure good drinking-water quality. After use, water is redirected into wastewater treatment 

plants and released into the surface-water system again after treatment. Hence, the quality of the 

raw water is influenced by a number of factors, including: 

• Temporal and spatial variations of surface-water quality, largely depending on location in 

relation to wastewater treatment plants. 

• Presence, permeability, and thickness of the colmation layer. 

• Lithology, permeability, and geochemistry of the aquifer sediment. 

• Nature of the wells (location, length, and depth of filter screens). 


Dr. Gudrun Massmann 

Scientific Colleague 


Hydrogeology Group • Malteserstr. 74-100 • 12249 Berlin Germany 

Phone: +49-30-83870472 • Fax: +49-30-83870742 • Email: massmann@zedat.fu-berlin.de 

49

50 

TS GWA Tegel 

TS Wannsee 

TS Tegeler See 

WW Spandau 

Spandau WW Jungfernheide 

Charlottenburg 

KW Ruhleben 

Unterschleuse 

WW Kladow 

Havel 

WW Stolpe KW Schönerlinde 

Tegeler See 

Spree 

WW Tiefwerder 

WW und PEA 

Beelitzhof 

Kleinmachnow 

KW Stahnsdorf 

WW Tegel 

Tegeler Flieβ 

PEA-Tegel 

input summer 

KW Ruhleben 

Telfowkanal 

Nordgraben 

Legend 

flow direction 

waterworks 

sewage treatment plant 

surface water treatment plant 

transect / field site 

weir 

lock 

Figure 1. Overview of the surface-water system, flow directions, waterworks, wastewater and surface-water 

treatment plants, and study sites in the western part of Berlin. 

• Hydraulic regime resulting from natural gradients and pumping performances. 

• Water-quality changes that occur during bank filtration. 

A major advantage of the application of bank filtration is the capability of the subsurface to 

remove contaminants during underground passage, either by physical filtration and/or (bio)chemical 

processes, such as adsorption, reduction, or degradation. In addition, the application of bank 

filtration spares natural groundwater resources, which inhibits the rise of deeper, more saline 

groundwater into local freshwater aquifers; however, since a proportion of the surface water in 

Berlin originates from treated effluent released by wastewater treatment plants, it is necessary to 

constantly monitor the fate of potential contaminants during bank filtration. 

Processes accompanying bank filtration and artificial recharge are currently studied in Berlin 

within a multidisciplinary cooperative project at the Berlin Centre of Competence called Natural 

and Artificial Systems for Recharge and Infiltration (NASRI) (KWB, 2002). The project focuses 

on the behavior and removal of, for example, pathogens, microcystins, and organic pollutants, as 

well as pharmaceutically active compounds, during underground passage. To interpret the 

behavior of these non-conservative water constituents, the local hydrogeological and hydraulic 

situation has to be well understood. Three field sites are the focus of current investigations, all in 

the western part of the City (see Figure 1). Two sites are located next to natural lakes (Tegeler 

See/Tegel Lake and Wannsee/Wannsee Lake) and one next to an artificial recharge pond (GWA 

Tegel). Only examples of the Wannsee Lake site will be discussed. 

Several tracers have been successfully applied in Berlin for various purposes. Some of these are 

useful to study water movement and derive mean residence times (e.g., delta deuterium [δD], 

delta oxygen 18 [δ 18 O], temperature [T], chloride [ C1 – ]), while others may be particularly useful 

in identifying the proportion of treated wastewater in surface water (e.g., boron [B], Cl – , δD, δ 18 O) 

or the proportion of either bank filtrate (e.g., ethylenediaminetetraacetic acid [EDTA], galdolinium 

[Gd], strontium [Sr]) or deeper saline groundwater (e.g., B, Cl-, sodium [Na+]) in raw water. 

Panke 

Landwehrkanal 

(KW Marienfelde)

Methodology 

Transects generally reach from the lake or artificial recharge pond to a production well parallel to 

the flow direction. These transects contain a number of observation wells, usually one below the 

lake, some between the lake and production well, and some beyond the well. A brief overview on 

the analysis discussed here is given in Table 1. 

The T-He age dating method uses the ratio of the concentration of radioactive tritium ( 3 H or T) 

derived from atmospheric nuclear bomb testing and its decay product, Helium ( 3 He), in 

groundwater to determine a groundwater age (i.e., the time passed since water had its last contact 

with the atmosphere). 

In case the abstracted water is a mixture of surface water and background groundwater only, the 

percentage of bank filtrate in the well (X) can be calculated as: 

X = [C w–C GW)/(C SW–C GW)]•100 [%] 

With C as the concentration of a suitable tracer in groundwater (C GW), well water (C w), or surface 

water (CSW). 

The simple mixing formula can be used under the premise that the differences between groundwater 

and surface water are large and relatively stable, and well water is a mixture of two water components 

only. 

Results 

Surface-Water System 

Table 1. Overview of Analytical Methods 

Parameter Analytical Method Laboratory Details Provided In 

4 He, 20 Ne, Mass Spectrometry Institute of Environmental Physics, Sültenfuß et al. 

and 22 Ne Pfeiffer QMG 112 University of Bremen (in preparation) 

(Age Dating) 

3 He and 4 He Mass Spectrometry Institute of Environmental Physics, Sültenfuß et al. 

(Age Dating) MAP215-50 University of Bremen (in preparation) 

Cl – Photometry Free University of Berlin 

(Autoanalyser, 

Technicon) 

B ICP-OCE Free University of Berlin 

δ 2 H, δ 18 O Delta S MS Alfred-Wegener-Institute, Meyer et al. (2000) 

Potsdam 

EDTA DIN 38413-PO3 Berlin Water Works 

4 He = Helium 4. 20 Ne = Neon 20. 22 Ne = Neon 22. 3 He = Helium 3. 4 He = Helium 4. ICP-OCE = Ionic Coupled Plasma. 

The strain on surface water becomes larger as it flows through the City and is substituted by 

considerable amounts of treated wastewater. The wastewater influence provokes high 

concentrations of, for example, Cl – , B, anthropogenic Gd, high conductivities and temperatures 

(in winter), and more negative isotope signatures. The highest influence and largest fractions of 

treated wastewater can be seen in the Teltow Canal and Nordgraben (a ditch). The Havel River 

51

52 

contains the lowest concentration of wastewater indicators. For example, Figure 2 shows the 

B concentration and the proportions of treated wastewater in surface water in September 2002. The 

influence of the Nordgraben extends to transect Tegel Lake, while the field site Wannsee Lake is 

under the influence of the Teltow Canal. The percentage of treated wastewater is higher during 

the summer months, when the Ruhleben Treatment Plant discharges into the Teltow Canal. 

Figure 2. (a) Percentage of treated wastewater in surface water (calculated with discharge only; evaporation, 

storage changes, and precipitation neglected). (b) Boron concentration indicating the influence of 

the Teltow Canal in the south and Nordgraben in the northeast for September 2002. 

By combining the results of the discharge measurements with the concentrations of wastewater 

indicators, proportions of treated wastewater can be calculated with a mixing formula for each 

sampling point. In Figure 3, results are shown for Tegel Lake in front of the transect. In October 

2001, a pumping device started operation, which pumps Havel River water into the PEA-Tegel, a 

phophate elimination plant (see Figure 1), when the natural discharge of the Nordgraben is very 

low in summer. This led to a strong decrease of treated wastewater in Lake Tegel near the transect 

(from 33 to 46 percent [Fritz, 2002] to an average of 11.7 +/– 1.9 percent). For surface water near 

transect Wannsee Lake, values of 20.1 percent in the summer (when Ruhleben discharges into the 

Teltow Canal) and 8.5 percent in the winter were calculated. 

Bank-Filtration System (Wannsee Lake) 

At Wannsee Lake, two transects exist, running perpendicular to the shore of Wannsee Lake 

between drinking-water Production Wells 3 (TS Wannsee 2) and 4 (TS Wannsee 1). The wells 

of the local gallery (Beelitzhof) have several filter screens in different porous aquifers separated by 

aquitards. A schematic overview is given in Figure 4. Also shown are the T and 4 He concentrations 

originating from hydrogen bomb testing in the 1960s, as well as uranium and thorium decay 

within the aquifer. In the two deeper aquifers, groundwater is considerably older than 50 years, 

since T is already decayed and the 4 He values are high. In contrast, the shallow wells reflect the 

present atmospheric concentrations of T, while 4 He could not be detected. The resulting age is less 

than the detection limit of 3 months. The production well is a mixture of all aquifers.

Figure 3. Percentage of treated wastewater calculated with B and Cl for surface water in front of transect 

Tegel Lake. 

W E 

3339 

3338 

3337 

BEE201OP 

BEE201UP 

3336 

3334 

3335 

Brunnen 4 

BEE200UP 

BEE200OP 

3332 

m below m above 

ground sea-level 

Figure 4. Tritium and terrigenic 4 He in different observation wells and Production Well 4 of TS Wannsee 1. 

Mean residence times are estimated with the help of (for example) δD and δ 18 O breakthrough 

curves. The surface water shows a clear seasonal signal, with more negative signatures in winter 

(Figures 5a and 6). The exemplary breakthrough curves of Observation Wells 3,337 and 3,335 

(see Figure 4) reflect the surface-water signal. Here, groundwater is made of bank filtrate only. The 

curves illustrate that residence times are rather low at this particular site. Travel times are around 

1 month on the distance from the lake to Observation Well 3,337, which is approximately 

two-thirds of the way to the production well. Well 4 does not show a seasonal signal and has a more 

negative signature. 

53

54 

δ 18 O [ 0⁄00 versus Standard Mean Ocean Water] 

EDTA [µg/L] 

–6 

–7 

–8 

–9 

7 

6 

5 

4 

3 

similar in the lake and Wells 3 and 5. Mixing calculations assuming a residence time of 1 to 2 months 

result in an average percentage of bank filtrate of 61 to 97 percent in Well 3, 10 to 47 percent in 

Well 4, and 88 to 96 percent in Well 5. The large variations are mainly due to the detection limit 

of 2-µg/L EDTA, which results in uncertainties in the actual concentrations of Well 4 and 

background groundwater. An evaluation of other tracers suggest that the proportion of bank 

filtrate in Well 4 is more likely to be in the order of magnitude of 10 percent. 

The differences between the water bodies become clear in the plot of δD versus δ 18 O (Figure 6), 

where Well 4 plots next to the deeper aquifer samples. In contrast, Well 3 and Well 5 plot within 

the surface-water samples (just like the shallow observation wells), but show less seasonal variation. 

δD [ 0⁄00 versus Standard Mean Ocean Water] 

–46 

–48 

–50 

–52 

–54 

–56 

–58 

–60 

–62 

–64 

–66 

–68 

–70 

Figure 6. δD versus δ 18 O at the Wannsee Lake field sites. Analysis of deeper aquifer samples was done from 

September 2000 to November 2001, with remaining data from May 2002 to March 2003. 

Conclusions 

Mar 03 

Jan 03 

Feb 03 

May 02 

Dec 02 

–9 –8 –7 –6 –5 

δD [ 0⁄00 versus Standard Mean Ocean Water] 

Jun 02 

Nov 02 

Sep 02 

Aug 02 

Jul 02 

The combination of different tracers enables the interpretation of the flow regime at sites where 

bank-filtration processes are currently studied in Berlin. With the help of T/He analysis, the ages 

of different water bodies can be estimated. The analysis of tracers showing distinct seasonal 

variations is used to estimate travel times, while water constituents — which are either mainly 

present in bank filtrate or background water — are used for mixing calculations. The results of the 

tracer studies can be used to interpret the fate and behavior of potential contaminants (e.g., drug 

residues, organic pollutants, etc.). 

55

56 

Acknowledgements 

We would like to thank the German Research Foundation, Veolia Water, and the Berlin Water 

Company for financing the NASRI Project. 

REFERENCES 

Fritz, B. (2002). Untersuchungen zur Uferfiltration unter verschiedenen wasserwirtschaftlichen, hydrogeologischen 

und hydraulischen Bedingungen, Ph.D. Thesis, University of Berlin, Berlin, 203 pp. 

KWB (2002). NASRI Natural and Artificial Systems for Recharge and Infiltration, First Progress Report, reporting 

period May 2002 – December 2002. 

Pekdeger, A., and C. Sommer-von Jarmerstedt (1998). Einfluß der Oberflächenwassergüte auf die Trinkwasserversorgung 

Berlins, Forschungspolitische Dialoge in Berlin, Geowissenschaft und Geotechnik, Berlin, pp. 33-41. 

Meyer, T., L. Schönicke, U. Wand, H.W. Hubberten, and H. Friedrichsen (2000). Isotope studies of hydrogen 

and oxygen in ground ice – Experiences with the equilibration technique, Isotopes Environ. Health Stud., p 133-149. 

Sültenfuß, J., G. Massmann, and A. Pekdeger (in preparation). Datierung mit der He3-T-Methode am Beispiel 

der Uferinfiltration im Oderbruch. 

GUDRUN MASSMANN studied Geology at both the Universities of Bremen and 

Edinburgh, specializing in Hydrogeology. After receiving her Diploma, she worked as an 

occupational trainee at the Center for Groundwater Studies at the Commonwealth 

Scientific & Industrial Research Organization (CSIRO) Land and Water in Adelaide, 

Australia, gaining experience on Artificial Storage and Recovery. For the past 4 years, she 

has worked as a scientific colleague for the Hydrogeology Group of the Free University of 

Berlin. Her research interests include hydraulic modeling, hydrochemistry, tracer analysis, 

and isotopes. Massmann finished her Ph.D. in summer 2002 at the Free University of Berlin on a project 

dealing with evaluating and modeling hydraulic and hydrochemical processes accompanying bank filtration 

in a polder region in Germany. She then continued to word on bank filtration, but changed the field site to 

Berlin, where bank filtration plays an important role in drinking-water production.

Session 4: Siting 

Siting and Testing Procedures 

for Riverbank-Filtration Systems 

Samuel M. Stowe, P.G., CPG 

International Water Consultants, Inc. 


Phased, multi-tasked investigations can be structured to collect pertinent data to evaluate yield, 

quality, and (ultimately) the design of RBF systems. The initial phase should: 

• Define project objectives (yield, quality, water use). 

• Identify project concerns (treatment and regulatory issues). 

• Collect available data for preliminary hydrogeological screening/feasibility. 

If the initial phase indicates a reasonable probability that potential favorable conditions exist 

(given project objectives), then subsequent phases can be designed to collect site-specific data for 

comprehensive analysis. Subsequent phases should include: 

• Test drilling. 

• Hydraulic interval testing. 

• Water-quality screening. 

• Streambed characterization. 

• Water-level monitoring. 

• Detailed aquifer and water-quality testing. 

The key parameters to any RBF evaluation are aquifer transmissivity and streambed permeability. 

A good understanding of these parameters, along with the hydrogeological setting, will result in 

the proper evaluation of the expected yield and quality of a RBF system and will allow for the 

thorough evaluation of potential means of development. Understanding the ability of the aquifer 

to provide sufficient RBF to recharge water pumped from a RBF system is key to ensuring that 

long-term capacities can be sustained and that target water quality can be maintained through a 

balance of infiltrated surface water and groundwater. 

Two case studies are presented to provide examples of recent RBF investigations. One is for a 

low-yield system (1.0 MGD), with very difficult access conditions in New Mexico, where rafts, 

helicopters, and portable drilling equipment were used. The other is a high-yield system 

(50 MGD) along a large river in the Great Plains, where more standard access and drilling methods 

(rotary, rotosonic) were used. 


Samuel M. Stowe, P.G., CPG 

President 

International Water Consultants, Inc. 

6360 Huntley Road • Columbus, Ohio 43229 USA 

Phone: (614) 888-6263 • Fax: (614) 888-9208 • Email: smstowe@collectorwellsint.com 

57

58 

Phase 1 Investigations 

Once the project objectives have been adequately defined, existing data that pertain to the 

hydrogeology and hydrology of the region of interest should be collected, assimilated, and assessed. 

This research should focus on the feasibility of developing the required capacity from 

unconsolidated deposits. Information collected for evaluation would include: 

• Well logs. 

• Bridge boring logs. 

• Steam flow and quality data. 

• Records of production wells for other water utilities in the region. 

• Reports on file with state or local agencies. 

• Other information that pertains to the hydrogeological setting of the area. 

Additionally, sites that are identified as potentially favorable should be visited and inspected. 

Phase 1 findings should include the following: 

• Listings of the available data reviewed and other resources used. 

• Discussions of the data and results of any preliminary analyses that were possible, with the 

confidence level of data obtained. 

• Identification of areas that appear to have a favorable potential for testing. 

• Preliminary rankings of prospective sites, with a discussion of relative advantages and 

disadvantages for siting. 

• Evaluations regarding water quality that may be expected from a RBF system located in 

this area, considering depth, proximity to the river, other regional water-quality problems 

experienced, and local features that may impact quality, etc. 

• Assessment of permits that may be required for installation and withdrawal using a 

RBF system. 

• Descriptions and budgets for work procedures recommended to be completed in 

subsequent phases. 


Phase 2 tasks would include drilling, sampling, and testing exploratory borings/observation wells 

at sites identified in Phase 1. Given logistics and site layouts, surface geophysical surveys (seismic, 

electrical resistivity, electromagnetic) could precede actual drilling. Geophysical surveys can be 

helpful in preliminary screening activities. These sites would be selected based upon access (land 

ownership, permission, etc.) and the potential for favorable saturated aquifer thicknesses near the 

river. The primary objective of this Phase is to locate the most favorable site(s) for detailed aquifer 

testing (Phase 3) and the collection of data for hydraulic and quality evaluations for site ranking. 

The selection of appropriate drilling and sampling methods are imperative during this phase. 

Drilling methods should consider: 

• Representative soil samples. 

• Expected drilling conditions (materials, depths). 

• Rate of advancement (timely drilling). 

• Cost-effectiveness.

• Potential for hydraulic testing and water-quality sampling. 

• Site access conditions. 

• Monitoring well installation. 

Potential drilling/sampling methods include: 

• Hollow stem auger with split-spoon sampling. 

• Cable tool with bailed samples. 

• Direct rotary with wash and split-spoon sampling. 

• Rotosonic with dual-tube continuous sampling. 

• Reverse rotary with wash samples. 

Generally, it has been found that rotosonic drilling provides the best overall results for representative 

soil sampling, speed, water sampling, and hydraulic testing. Drawbacks have been costs and 

access. In the rotosonic drilling method, a drill casing is advanced into the ground using 

rotary/vibrasonic techniques. This method does not require the use of drilling mud, so there is no 

mud to dispose of, and ground-surface disturbance is minimal. The rotosonic drilling method 

produces nearly continuous samples of the materials penetrated by the sample tube, and the 

method produces representative samples from unconsolidated, granular deposits. 

Sieve analyses should be performed on selected lithologic samples collected from test borings to 

determine optimum well-screen design and to help evaluate hydraulic conductivity. Hydraulic 

interval testing can also be conducted on test borings to determine the hydraulic conductivity of 

selected intervals and to evaluate groundwater quality. 

Following the collection of all field data, the data are compiled and analyzed to determine the 

most favorable locations for detailed aquifer testing (Phase 3). This determination is based upon 

saturated thicknesses, hydraulic conductivities, and logistics. From Phase 2 results, it can be 

determined with a reasonable degree of certainty that one or more of the sites tested is capable of 

yielding the desired quantity of water. 


At the completion of Phase 2, if results are favorable, a site (or sites) for detailed aquifer testing is 

selected. Phase 3 studies generally include the installation of additional observation wells for 

sampling subsurface materials, monitoring water levels, and conducting a detailed aquifer test, 

along with a test pumping well. Additionally, the evaluation of streambed conditions (width, 

depth, permeability) is generally conducted during this phase. 

Observation wells are positioned in a pattern around the test pumping well at appropriate locations 

and distances selected to facilitate data analysis, with emphasis on RBF. Following the installation 

of observation wells, a temporary test pumping well is installed. It is important that the test 

pumping well be constructed to produce enough water to adequately stress the aquifer. Well points 

can be installed in the river (conditions permitting) to further evaluate streambed permeability. 

The monitoring of water levels in the wells and river should be conducted prior to any pumping 

to evaluate antecedent trends and the correlation of stream level and groundwater levels, with 

levels converted to elevation datum for analysis. Following the collection of background water 

levels, a long-term (2- to 10-days) constant rate test is conducted. A general guideline is to run 

the test 24 hours after steady-state conditions are reached. At the conclusion of the test, the 

59

60 

pumping is discontinued and the recovery of water levels is monitored until at least 90-percent 

recovery is obtained in the observation wells. During the test, samples of the water pumped (and 

river) can be obtained for laboratory analyses. Additionally, the pump discharge should be 

monitored every 6 to 12 hours and river quality every 24 hours for field screening of temperature, 

pH, iron, hardness, turbidity, and specific conductance. 

All data from Phase 3 are analyzed to determine aquifer properties, streambed permeability, aquifer 

character, yield, anticipated quality, and conceptual RBF system design. Yield estimates should 

consider the impacts of river stage/flow and water temperature variations. Based upon the results, 

the preferred location of a RBF system can be recommended based upon an evaluation of benefits 

and drawbacks, including refinements in design. Phase 3 testing may also identify additional 

testing that may be necessary to finalize design. 

Case Studies 

Two case studies are briefly discussed to illustrate the phased approach to siting RBF systems and 

critical evaluation items. Case Study Number 1 was completed in north-central New Mexico along 

the Rio Grande (Figure 1), where difficult access conditions presented challenges for completion 

using conventional approaches. Project objectives were to locate a site that could yield at least 

1.0 MGD of infiltrated surface water from the Rio Grande for potable use (surface-water rights to 

1,200 acre-feet per year). Phase 1 studies (literature review, site reconnaissance) identified up to 

six potentially favorable sites within the canyon. Given difficult site access, the initial Phase 2 study 

involved surface geophysics (electrical resistivity) at the six locations, along with a control site. 

Access to the sites was obtained using an all terrain vehicle and rafts, with base camps set up for 

overnight stays. Based upon the survey, two sites were selected for test drilling. 

Figure 1. Rio Grande Valley, New Mexico. 

Drilling was conducted using a portable modular core rig with wash samples and 2-inch 

outside-diameter split-spoon samples. At Site A, nine borings were drilled, six 1.5-inch diameter 

observation wells installed, and hydraulic testing (high-efficiency suction pumps) completed. As 

results were favorable, Phase 3 aquifer testing was completed with additional observation wells and 

a 66-hour constant rate test (0.144 MGD). 

Following the completion of testing at Site A, the equipment and base camp were mobilized to 

Site B using rafts and a helicopter. Phase 2 testing included the drilling of four borings, interval

testing, and a short constant rate test. As results here were not as favorable as at Site A, detailed 

aquifer testing was not completed. 

Case Study Number 2 was completed along the Missouri River in North Dakota (Figure 2), where 

more conventional investigative methods could be used. Project objectives were to locate a site 

that could yield up to 50 MGD of potable water from a RBF system. The study was being 

completed to evaluate the possibility of replacing an existing direct surface-water intake. Phase 1 

investigations (desktop study, site visit) determined that several sites existed within the area, 

which were promising. Phase 2 investigations were targeted at two sites, given the primary 

logistical concern of distance from existing treatment. Phase 2 included the drilling of three 

borings and conducting one hydraulic test using rotosonic methods. 

Figure 2. Missouri River Valley, North Dakota. 

Given favorable Phase 2 results, Phase 3 investigations were completed at one site with the 

installation of four additional observation wells and a test pumping well. Because of budget 

concerns, Phase 3 drilling was completed using mud rotary methods and a local drilling contractor. 

Testing was completed with the running of a 72-hour constant rate test (1.9 MGD). The analysis 

of data indicated a transmissive aquifer (40,000 square feet per day) in good communication with 

the river. Preliminary results indicate that the site will yield 30 MGD from a single horizontal 

collector well. 

SAM STOWE is President of International Water Consultants, Inc., a subsidiary of 

Collector Wells International, Inc. A hydrogeologist, Stowe has nearly 30 years of diverse 

experience in the groundwater industry. He has been in charge of projects involving 

aquifer-test analyses, riverbank filtration and recharge evaluation, groundwater quality, 

well design, groundwater management, numerical modeling, contamination investigation, 

and remedial action. He has also been involved in groundwater-supply projects for yields 

of nearly 100 million gallons per day and in contamination evaluations ranging from 

industrial organic pollution from landfills to the environmental effects of strip and deep-coal mining. In 

addition, Stowe is highly experienced in evaluations of induced infiltration potential from streams and 

oceans. He has completed hundreds of hydrogeological evaluations for horizontal collector wells in regard to 

yield, quality, and design, having been responsible for siting and designing many horizontal collector wells 

with individual yields ranging from 2 to 40 million gallons per day and is a recognized expert in their 

application. Stowe received a B.A. in Geology from Miami University and an M.S. in Geology from Ohio 

State University. 

61

Session 4: Siting 

Water-Quality Management for Existing Riverbank- 

Filtration Sites along the Elbe River in Germany 

Prof. Dr.-Ing. Thomas Grischek 

University of Applied Sciences Dresden 

Dresden, Germany 

Introduction 

In former times, the management of RBF sites in Germany focused on water quantity — how much 

water could be abstracted and what groundwater levels would result. Control measures were based 

on measurements of river and groundwater levels and pumping rates and on black-box models to 

estimate the portion of riverbank filtrate abstracted (e.g., Luckner and Nestler, 1982). The 

development of computer programs for groundwater flow and transport simulations resulted in 

increasing management applications for RBF sites (e.g., Koster et al., 1994; Heinzmann, 1998; 

Eckert et al., 2000). 

The number of published papers dealing with the complex management of water quality at RBF 

sites is very low. Besides work done by Sontheimer (1991) and Schubert (1999) at sites along the 

Rhine River, a control and management concept for the Hengsen site on the Ruhr River 

(Schöttler and Sommer, 1992) must be highlighted. At the end of the 1980s, water-quality 

management also became a subject for RBF sites on the Elbe River due to problems with river and 

raw-water quality. Initial work done by Müller, Schwan, and others between 1985 and 1989 was 

continued by Nestler et al. (1998). 

Water-quality management must be based on detailed knowledge of groundwater flow conditions 

and monitoring measures to obtain sufficient data on water quality; however, this is not the case 

for every site. Of course, for small waterworks, such investigations and investments may be higher 

than the cost savings resulting from water-quality management. But, the optimization of water 

abstraction can have long-term effects on raw-water quality and treatment cost savings. Waterquality 

management systems are well established at sites where problems with contamination have 

occurred (e.g., high concentrations of nitrate, organic halogens). In such cases, experiences may 

not be published to give the impression of a “no problem waterworks” or because a clear 

description is not easy due to the large amount of data needed to understand the complex system 

and site-specific boundary conditions. 

General Water-Quality Management Measures 

The first step in water-quality management is to clarify which advantage of RBF is the most 

important and to identify the main aims or problems. At one site, the nitrate concentration in the 

raw water should be decreased; at another site, the concentration of DOC, dissolved iron, or an 

organic contaminant should decrease. 



Professor 

Institute for Geotechnics & Water Sciences 

University of Applied Sciences Dresden • Friedrich-List-Platz 1 • 01069 Dresden, Germany 

Phone: +49 351 4623350 • Fax: +49 351 4623567 • Email: grischek@htw-dresden.de 

63

64 

Water-quality management could include: 

• Optimization of the operation of existing abstraction wells. 

• Building additional abstraction wells. 

• Technical measures in the riverbed. 

• (Political) activities to improve river-water quality. 

• Contracts with farmers to change land use in a catchment zone. 

This paper focuses on the operation of existing wells affected by mixing processes in an aquifer, 

flow times and flow path lengths, and the catchment area for landside groundwater. Special 

measures to face problems resulting from contaminant shock loads in a river, as well as droughts 

or floods, are not covered because these are the subjects of other contributions in this volume. 

Strategies based on the operational control of existing abstraction wells could include: 

• Well operation to achieve a maximum volume of utilized aquifer (long flow paths and 

retention times). 

• Preferential operation of selected wells with best raw-water quality. 

• Operation of a limited number of wells with high abstraction rates to increase the 

proportion of bank filtrate in raw water. 

• Continuous operation of selected wells to meet the mean water demand and additional 

operation of other wells in peak periods. 

• Periodic operation of wells to increase the effects of dispersion and mixing in the aquifer. 

An additional technical measure could be a change of the well filter length and depth in an 

aquifer. 

Riverbank Filtration along the Elbe River 

At present, there are eight water facilities along the Elbe River that use RBF to supply more than 

1.5-million people with drinking water and industry with process water. Three of these sites have 

been chosen to carry out further investigations on water-flow and water-quality changes during 

RBF, especially due to the changing water quality of the Elbe River in the 1990s. Between 1992 

and 2002, different research programs focused mainly on the behavior of DOC, EDTA, sulfur 

organic compounds, and carbonic acids. Besides this specialized research, daily problems among 

water companies were investigated. 

All RBF sites along the Elbe River are operated under anoxic conditions. Reducing conditions 

result in the dissolution of iron and manganese along the flowpath between the river and 

abstraction wells. At some sites, denitrification occurring during RBF was found to compensate for 

high nitrate concentrations in groundwater from other sources. 

Due to the closure of large industries and a new water pricing system after the reunification of 

Germany, water demand decreased significantly between 1989 and 1998. Lower abstraction rates 

for bank filtrate and an improvement in river-water quality allow for efficient groundwater 

management and the optimization of pumping schemes to obtain good raw-water quality.

Torgau Case Study 

The highly productive RBF scheme near Torgau in Saxony, eastern Germany, is situated within the 

Elbe River Basin. The basin is filled with Pleistocene deposits to a depth of 10 to 55 m that comprise 

interfingered glaciofluviatile sediments ranging from fine sand and silt to medium sand and gravel 

(Grischek et al., 1998). The deposits are overlain by Holocene river gravels (5- to 8-m thick) and 

meadow loam (2-m thick). The hydraulic conductivity values range from 0.6-2 × 10 –3 m per second. 

The Elbe River is in direct hydraulic contact with the aquifer. Riverbed clogging is low. The RBF 

system consists of 42 vertical wells arranged in nine well groups (Figure 1) and has a capacity of 

150,000 m 3 /d. The distance between abstraction wells and the riverbank is about 300 m. The flow 

time of bank filtrate is between 80 and 300 days. The wells have a 20-m long filter placed in the 

lower layer of the aquifer at 30- to 50-m below ground surface. The site is well equipped with two 

sampling profiles along flowpaths between the river and abstraction wells, with monitoring wells 

in the landside catchment zone. 

5712 

5710 

5708 

5706 

2.3* 

2.0 

3.2 

2.6* 

3.2 

4.4 

4.1* 

1.7 

3.1 

I 

1.0 

1.0 

3.1 

II 

1.2 

1.9 

2.6 

2.7 

III 

1.3 

1.1 

3.0 1.0 

1.4 4.1 

0.8 

3.3 1.0 

0.9 

IV 

1.7 

3.3 

2.9 

2.8 

Mixed water 

DOC < 2.3 mg/L 

1.1 

0.9 

2.8 

0.8 

Lakes 

0.8 

1.5 

1.8 

3.3 

2.7 

4.6* 

2.8 

3.0 

5.1 

Mixed water 4.2 

DOC 2.3...3.0 mg/L 

Elbe River 

V VI 

VII 

1.6 

2.0 

4.3 

2.6 

5.2 5.0 

Mixed water 4.8 

DOC > 3 mg/L 

Legend 

Abstraction well 

3.7 

3.2 

3.0 3.3 

1.9 

1.7 

4571 4573 4575 4577 

Figure 1. Mean DOC concentration in milligrams per liter in groundwater in the catchment zone of the 

Torgau Waterworks from 1995 to 1997. 

The main aims of water-quality management include the maximum attenuation of organic 

compounds during aquifer passage and low concentrations of DOC, dissolved iron, and nitrate in 

raw water. Results from long-term monitoring programs and special field experiments provided an 

excellent database to draw conclusions on the effect of water-quality management measures. A 

detailed description of redox conditions and removal rates of organics is given in Grischek et al. 

(1998, 2000). Table 1 summarizes the effects of different management measures for the RBF site 

at Torgau. 

2.6 

3.2 

4.4 

4.4* 

Observation well with three 

sampling depths; mean DOC 

concentration in mg/L, 1995-97 

Affected by local infiltration 

of surface water 

VIII 

1 km 

IX 

1.5 

old branch 

65

66 

Table 1. Effect of Control Measures on Raw-Water Quality at the RBF Site in Torgau 

(Grischek, 2002) 

Control Measure Bank- Effect on Raw-Water Quality 

Filtrate DOC Trace NO3 – Fe 2+ NH4 + SO4 2– 

Portion 

Organics Mn 2+ 

Continuous operation of wells 

with increased abstraction rate 

Continuous operation of wells 

with decreased abstraction rate 

Periodic operation of selected wells 

Preferential operation of selected 

wells with favorable catchment area 

Operation of every second or third 

well within a group of wells 

Technical measure 

Change of the well filter depth 

➚ 

➚ 

➚ 

➚ 

— — — 

➚ 

➚ ➚➚➚ 

➚ ➚ ➚ 

➚ 

➚ 

➚➚➚ 

➚ 

— — — — — 

— Effect is negligible or only temporary. Concentration increase. Concentration decrease. 

— 

Readily degradable and highly adsorbable organic compounds are attenuated in the biologically 

active riverbed. Long flow paths and long retention times of the bank filtrate in the aquifer allow 

further attenuation of poorly degradable and some polar organics. 

A field test was done to study the effect of an increase in water abstraction resulting in a decrease 

in retention time of bank filtrate in the aquifer. Over a period of 1.5 years, water abstraction from 

five selected wells was increased by about 40 percent. Excluding some slight changes in water 

quality near the bank line, no significant effect due to the decrease in retention time was observed. 

DOC removal and denitrification along the whole flowpath were not affected (Grischek, 2002). 

A continuous operation of selected wells has the advantage of lower concentrations of dissolved 

iron in raw water. High concentrations of dissolved iron were not measured in the bank filtrate, 

but were measured in the landside groundwater. Switching off the abstraction wells results in 

groundwater flow towards the river and the transport of dissolved iron into the aquifer zone 

between the wells and river. When the pumps are switched on again, higher iron concentrations 

are observed in raw water. Furthermore, periodic well operation leads to higher well clogging. 

A groundwater flow and transport model has been used to simulate the change of the well filter 

depth and its effect on groundwater flow, especially groundwater flow from the opposite side of the 

river beneath the riverbed to the wells. Due to the long distance between the wells and bank line, 

the effect was found to be negligible. A change in the filter depth should only be considered if the 

aquifer consists of layers with very different hydraulic conductivities and if the well is located at a 

short distance from the riverbank (the distance is less than aquifer thickness). 

Surprisingly, the most important factor was the selection of abstraction wells according to their 

catchment zone for landside groundwater. Based on the dense net of observation wells, it was possible 

to identify zones with different concentrations of DOC and dissolved iron in landside groundwater. 

Figure 1 shows the range of DOC concentrations in groundwater in the catchment zones behind 

the wells. 

— 

— 

— 

— 

➚ 

— 

➚ ➚➚ 

➚ ➚ 

➚ 

➚ 

➚ ➚➚➚

The mean concentration in the “mixed water” (see Figure 1) was calculated from the concentrations 

determined at different depths in the aquifer. Depth-dependent groundwater sampling was a 

key factor in understanding groundwater flow and quality changes in the catchment zone. The 

operation of Well Groups II to IV results in lower DOC concentration in the raw water compared 

to the operation of Well Groups VIII and IX. Fortunately, low DOC concentrations are associated 

with low dissolved iron concentrations. Thus, the advantage of appropriately selecting abstraction 

wells covers both parameters. Because the water-quality change for bank filtrate was very similar 

in all wells, an improvement of raw-water quality can be achieved mainly by the selection of wells 

abstracting the proportion of landside groundwater with the best quality. At the Torgau 

Waterworks, the preferential operation of these wells has already resulted in cost savings, 

especially for the removal of dissolved iron during the water-treatment process that requires iron 

sludge disposal. 

Summary 

The main aim at all RBF sites along the Elbe River is the attenuation of organic compounds and 

low concentrations of DOC, dissolved iron, and manganese in raw water. Long flow paths and 

retention times promote the attenuation of organics, but were found to have only a relatively small 

effect on iron and manganese concentrations. In general, the continuous pumping of selected 

wells should be preferred over periodic operation. At all sites, mixing ratios of bank filtrate and 

groundwater were found to be of main importance for the concentration of nitrate, sulfate, 

dissolved iron, and manganese in raw water. Due to different groundwater qualities within the 

whole catchment zone, a selection of wells having a catchment zone with good groundwater 

quality offered a significant improvement in raw-water quality. Thus, monitoring systems for RBF 

sites should not only focus on bank filtrate, but should include observation wells landside of the 

abstraction wells. Due to site-specific boundary conditions, a detailed investigation of groundwater 

flow conditions and proportions of bank filtrate in raw water is very important for drafting 

effective water-quality management measures. 

REFERENCES 

Eckert, P., C. Blömer, J. Gotthardt, S. Kamphausen, D. Liebich, and J. Schubert (2000). “Correlation 

between the well field catchment area and transient flow conditions.” Proceedings, International Riverbank 

Filtration Conference, W. Jülich and J. Schubert (eds.), IAWR Rheinthemen 4, 103-113. 

Grischek, T., KM. Hiscock, T. Metschies, P.F. Dennis, and W. Nestler (1998). “Factors affecting denitrification 

during infiltration of river water into a sand and gravel aquifer in Saxony, Germany.” Wat. Res., 

32(2): 450-460. 

Grischek, T., E. Worch, and W. Nestler (2000). “Is bank filtration under anoxic conditions feasible?” 

Proceedings, International Riverbank Filtration Conference, W. Jülich and J. Schubert (eds.), IAWR 

Rheinthemen 4, 57-65. 

Grischek, T. (2002). Zur Bewirtschaftung von Uferfiltratfassungen an der Elbe (Management of riverbank filtration 

sites along the River Elbe), Ph.D. thesis, Dresden University of Technology, Institute for Groundwater 

Management (in German). 

Heinzmann, B. (1998). “Beispiele für integriertes Management von Wasserressourcen in der Region Berlin- 

Brandenburg. (Examples of integrative management of water resources in the Berlin-Brandenburg region).” 

Wasserwirtschaft in urbanen Räumen. Schriftenreihe Wasserforschung, 3: 171-197 (in German). 

Koster, N., U. Willme, and K. Döhmen (1994). “Wasserwirtschaftliche Betriebsanalyse am Beispiel einer 

Wassergewinnungsanlage im Ruhrtal (Analysis of water management at a waterworks in the River Ruhr 

valley).” Wasser Abwasser Praxis, 5: 19-23 (in German). 

67

68 

Luckner, L., and W. Nestler (1982). “Zur Methodik des Aufbaus und der Nutzung von Kontroll- und 

Steuerungsprogrammen von Grundwasserfassungen (On the methodology of structuring and use of control 

programs for groundwater abstraction wells).” Wasserwirtschaft-Wassertechnik, 32(7): 219-223 (in German). 

Nestler, W., W. Walther, F. Jacobs, R. Trettin, and K. Freyer (1998). Wassergewinnung in Talgrundwasserleitern 

der Elbe (Water production in alluvial aquifers along the River Elbe), UFZ-Research Report 7 (in German). 

Schöttler, U., and H. Sommer (1992). “Optimierung einer Wassergewinnungsanlage mit Hilfe der Modellrechnung 

und ihre Auswirkungen auf die Grundwassergüte (Optimisation of a water abstraction system with 

the help of model calculations and the effects on groundwater quality).” Steuerung in der Wasserwirtschaft, 

DFG-Forschungsbericht, VCH-Verlagsges., Weinheim, 178-195 (in German) 

Schubert, J. (1999). “Riverbank filtration — Field studies, modeling, monitoring.” Proceedings, International 

Riverbank Filtration Conference, 4-6 Nov 1999, Louisville, Kentucky, 39-42. 

Sontheimer, H. (1991). Trinkwasser aus dem Rhein? Bericht über ein Verbundforschungsvorhaben zur Sicherheit 

der Trinkwassergewinnung aus Rheinuferfiltrat (Drinking water from the River Rhine? Report on a collaborative 

research project on safety of drinking water production from bank filtrate from the River Rhine), Academia Verlag, 

Sankt Augustin, Germany (in German). 

THOMAS GRISCHEK has 15 years of research experience in the field of riverbank 

filtration. He published eight papers on riverbank filtration in refereed journals as principal 

author and has contributed to more than 10 papers as co-author. His main research 

interests are the interaction of groundwater and surface water, especially riverbank 

filtration and artificial groundwater recharge, groundwater management, and diffusing 

groundwater pollution. Currently, he is Professor of Water Science at the University of 

Applied Sciences Dresden. Prior to this position, Grischek spent 8 years as a Research 

Associate for the Institute for Groundwater Management at the Dresden University of Technology, as well 

as at the University of Applied Sciences Dresden. In addition, he also taught at the Institute for Water 

Chemistry at the Dresden University of Technology. In 2003, he worked as a Referee at the Saxon State 

Agency for Environment and Geology. Grischek studied Water Management in the Department of Water 

Sciences at Dresden University of Technology, where he graduated as a diploma engineer. He also received 

a Ph.D. in Environmental Engineering from the Dresden University of Technology in 2002, where he 

researched the “Management of Riverbank-Filtration Sites along the Elbe River.”

Session 5: Dynamics 

Using Models to Predict Filtrate Quality 

at Riverbank-Filtration Sites – 

What Is the Adequate Level of Modeling? 

Chittaranjan Ray, Ph.D., P.E. 

University of Hawaii at Mañoa 

Honolulu, Hawaii 

Henning Prommer, Ph.D. 

Delft University of Technology 

Delft, The Netherlands, and 

Center for Groundwater Studies 

Perth, Australia 




RBF is an efficient and low-cost treatment technology for drinking-water production that has 

been in use for more than a century in Europe and nearly half a century in the United States. 

There has been a recent surge in the design and construction of RBF facilities in the United States 

due to the potential to receive 0.5- to1.0-log removal credit for protozoa Cryptosporidium filtration 

under the upcoming Enhanced Surface Water Treatment Rule (USEPA, 2002). While most small 

water utilities use RBF as the sole treatment system (with the exception of disinfection), most 

medium to large utilities use RBF as a pretreatment system, which helps process performance in 

the final treatment system. 

The design and operation of RBF systems are facilitated by the use of flow and transport models 

for estimating yield and water quality of the filtrate; however, the level of the modeling effort can 

vary depending upon the utility, data availability, and problems to be solved. While water-quality 

modeling is the primary focus, flow simulation is the first step prior to undertaking transport 

simulations. In this paper, we present various scenarios involved in flow and transport modeling 

at RBF sites and discuss what level of modeling is adequate for what problems. 

Flow Simulations 

Most utilities undertake some sort of flow simulation to estimate yield from RBF wells. These 

simulations are broadly divided into (a) analytical and (b) numerical models. Analytical models 

(Ferris et al., 1962; Hantush, 1959) assume that the river fully penetrates the aquifer and there is 

no additional resistance to flow at the river-aquifer interface. Hantush (1959) also presented a 

simple procedure to handle partial penetration of the river by moving the recharge boundary some 

distance away from the actual location with respect to the pumping well. Other analytical models, 


Chittaranjan Ray, Ph.D., PE 

Associate Professor 

Department of Civil & Environmental Engineering 

University of Hawaii at Mañoa • 2540 Dole Street, 383 Holmes Hall • Honolulu, Hawaii 96822 USA 

Phone: (808) 956-9652 • Fax: (808) 956-5014 • Email: cray@hawaii.edu 

69

70 

as described by Dillon and Ligett (1983) and Wilson (1993), include transient effects and 

clogging. Their application is mostly limited to two-dimensional problems. Three-dimensional 

flow and hydraulic and material heterogeneity cannot be easily handled by these methods. Conrad 

and Beljin (1996) provide a result comparison from the application of analytical and numerical 

models for different sites. 

Nowadays, numerical models are most commonly used to simulate stream-aquifer interaction. 

MODFLOW (Harbaugh et al., 2000) and MODPATH (Pollock, 1994) are some of the commonly 

used models for estimating hydraulic heads and path lines of neutrally buoyant particles. 

MODFLOW can handle material heterogeneity and various boundary conditions. The River and 

Stream/Aquifer Interaction packages are typically used in MODFLOW to examine the interaction 

between surface water and groundwater. In the River package, the hydraulic heads in the aquifer 

and the stage of water in the river are prescribed. Certain block-centered grids serve as the river. 

If the hydraulic head in the aquifer is lower than river stage, then the flow is from the river to the 

aquifer and vice versa. The hydraulic conductivity of riverbed material controls the flow into or 

out of the river. In contrast to the River package, where the hydraulic heads are specified, the flow 

is routed through the channel in the Stream/Aquifer Interaction package using the Manning 

equation, channel geometry, roughness, and other parameters. 

MODPATH is used to conduct advective tracking of neutrally buoyant particles in the flow field using 

forward or reverse particle tracking techniques. For example, reverse particle tracking from the screen 

zone of a well can determine the areas contributing to flow to a well. Similarly, forward tracking of a 

set of particles from a riverbed would indicate if any of the particles will be captured by pumping wells 

located on riverbanks. Path lines can be delineated for steady and transient flow conditions. 

In MODPATH simulations, the hydraulic gradient and well location can play important roles. For 

example, in a stream-aquifer simulation, the path lines for a single well can be seen in Figure 1. 

This figure shows the impact of the placement of one or more wells on the flow field. In the left 

side of this figure, a portion of the flow to the pumping well originates from the river and the rest 

comes from upgradient locations; however, if a second well is installed to the northwest of this well 

(figure to the right), a hydraulic divide is created. As apparent, most of the flow to this new well 

comes from the aquifer. 

In most river-aquifer simulation models, the nearest riverbank or the mid-point of the river is often 

considered as the model boundary. This assumption may hold as long as the river is a significant 

Figure 1. Flow fields for one and two pumping wells located on the bank of the Illinois River.

source of induced infiltration to the well; however, if the stream bottom is constituted with low 

permeability material and the pumping rate is high, a portion of the pumped water could come 

from the other side of the river. Figure 2 shows a case study for the RBF wells of Meissen- 

Siebeneichen, Germany, in which the path lines go past the mid-point of the river and extend to 

the other side. Cross-sections A-A and B-B show that only particles starting in the upper layer of 

the aquifer actually reach the river. All other particles are turned away near the river, but travel 

towards the production borehole. Thus, high nitrate concentrations in the lower layer of the 

aquifer between the river and production borehole can be explained by groundwater flow from the 

opposite side of the river. Further, the impact of riverbed clogging on the proportion of pumped 

river water was investigated using the model. Under site-specific geological conditions, observed 

clogging of the river bed (1 × 10 –5 m per second of the 0.1-m thick clogging layer) causes a 

decrease in RBF by only about 5 percent, as compared to the “no clogging” case. In general, 

geological anisotropy was found to have a stronger effect on the proportion of groundwater flow 

beneath the riverbed than riverbed clogging. Even under conditions where riverbed clogging does 

not occur, groundwater can flow from the opposite side of the river beneath the riverbed towards 

the production boreholes (Grischek et al., 2002). 

Model Grid 

98.0 

98.1 

98.2 

A A 

98.3 

The Meissen-Siebeneichen example demonstrates that if groundwater on the other side of the river 

contained dissolved contaminants, the contaminants could possibly appear in the filtrate. Thus, while 

designing RBF systems, land use on the other bank of the river should be considered and potential 

pollution sources must be identified. The calibration of the flow model heavily depends on the 

accuracy of the hydraulic conductivity of riverbed material. The heterogeneity of riverbed material 

can produce uneven leakage through the riverbed. Further, hydraulic parameters of the riverbed 

material can change depending on the flow status of the river due to scouring and sedimentation 

process. To date, there is no reliable and verifiable method available to estimate riverbed hydraulic 

conductivity of large rivers in situ and to estimate its variability with the flow regime of the river. An 

overview on methods and recent developments is given by Macheleidt et al. (2002). 

Single-Species Transport Modeling 

To estimate the amount of river water entering a well, mass balance methods are generally 

employed. In such methods, the concentrations of a conservative chemical in the river, aquifer, 

and well are used to estimate the mass fractions coming from the river and aquifer: 

C well = xC river +(1 – x)C aquifer 

98.0 

98.1 

98.2 

98.4 

Flow Paths (Plan View) 

98.3 

Adjacent 

Granitic 

Rocks 

98.4 

(Equation 1) 

98.5 

98.5 

Elbe River 

Legend 

Production Well 500 m 

–98– Piezometric Contour 

in Meters Above Mean Sea Level 

98.6 

A Flow Path A 

(Side View) 

Well 

Figure 2. Location map of a RBF site of the Meissen-Siebeneichen Waterworks in Germany (after Grischek et al., 2002). 

71

72 

In Equation 1, the fraction of river water (x) can be calculated if the three concentrations are known. 

Similarly, if x is known from prior investigations, the concentration of a conservative chemical in 

the filtrate may be estimated; however, the equation is only valid for steady-state conditions. In other 

words, concentrations in the river and aquifer do not change with time. Also, it is assumed that 

infiltrating water from the river at a given moment in time is of the same quality as that reaching the 

well and that the quality of the bank filtrate and groundwater does not change over the whole 

thickness of the aquifer. Such assumptions hamper the use of Equation 1 for spill events, even for 

conservative chemicals. Transient effects such as lag time (or travel time) issues are not considered. 

Further, Equation 1 cannot be used for non-conservative contaminants that are subject to 

degradation, sorption, or other reactions. For such contaminants, knowing the concentrations in the 

river and aquifer, as well as the mass fraction of the water derived from the river, one would not be 

able to predict the concentration at the well. For transient problems, the application of numerical 

models is typically required. In those, the advection-dispersion equation is solved with or without 

chemical reactions, depending upon the type of contaminants. The advection-dispersion equation 

can be solved analytically if the flow field is steady and reaction terms are linear. In cases where the 

use of three-dimensional numerical models is beyond the ability of water utilities, simple 

one-dimensional analytical models can be used to estimate chemical concentrations in pumped 

water, especially from shock events or chemical spills. Mälzer et al. (2003) used the analytical 

solution to the one-dimensional advection-dispersion equation to estimate the concentration of a 

chemical reaching a well. In their approach, they approximated the flow field between the river and 

aquifer using five flow tubes. For each tube, they solved the advection-dispersion equation with 

(sorption and first order degradation) to calculate the concentration at Observation Well M1 

(Figure 3); however, the assumption of steady flow through these flow tubes may not be valid for all 

cases. Further, estimating the concentrations for a contaminant at the pumping well may not be easy 

without knowing the amount of groundwater contribution to the well. 

Meters Above Sea Level 

35 

30 

25 

20 

15 

10 

Vertical Section at Wittlaer 

Rhine River 

Aquifer 

For transient three-dimensional simulations, MT3D/MT3DMS (Zheng and Wang, 1999) is 

probably the most common transport simulator that is used in conjunction with MODFLOW. 

MT3D uses the same grid of MODFLOW and solves advection-dispersion equation, along with 

sorption and degradation reactions, to estimate the concentration of a contaminant at a given 

location in the model domain. For estimating the concentration at the pumping well, a flow model 

must be built and calibrated to provide a good description of the velocity field. As mentioned 

earlier, the hydraulic conductivity of bed material at the river-aquifer interface can significantly 

23 m 

27 m 

Figure 3. Vertical section of the aquifer at Wittlaer between the Rhine River 

and Observation Well M1 and a simplified model assumption of five 

parallel tubes (after Mälzer et al., 2003). 

M1 

High 

Low

affect the amount of contaminant entering from the river to the aquifer and, eventually, to the 

pumping well(s). For flow fields affected by pumping wells, the assumption of equilibrium sorption 

(in contrast to kinetically controlled sorption) near the screen zone may not be valid due to fast 

travel rates of water particles. Ray et al. (2002) simulated the impact of riverbed/bank material 

properties on filtrate quality during a flood pass-through event at a RBF facility on the banks of the 

Illinois River. They used equilibrium sorption and first order degradation from literature-reported 

data. Also, in certain simulations, they considered the contaminant to be conservative. This 

assumption enveloped the two extremes. For atrazine, with a highly conductive bank, the 

concentrations in groundwater and the pumped well are slightly attenuated when the effects of 

sorption and degradation are considered (Figure 4A); however, without sorption and degradation, 

the concentrations can be close to that found in river water (Figure 4B). 

Atrazine Concentration, µg/L 


4 

3 

2 

1 

0 

4 

3 

2 

1 

0 

(A) with sorption and decay 

River water 

SP-1 

Caisson 

SP-3 is below detection limit 

0 20 40 60 80 100 

River water 

Days Since Start of Simulation 

(B) without sorption and decay 

SP-3 

SP-1 

Caisson 

0 20 40 60 80 100 


Figure 4. Atrazine transport from the river to the aquifer and the pumping well 

with and without sorption and decay reactions for a highly conductive bed 

and bank (after Ray et al., 2002). 

73

74 

When the hydraulic conductivity of the bed and bank are reduced to moderate values (see sets 

A-A and B-B in Ray et al., 2002), there is a significant attenuation of atrazine at the collector well 

caisson (Figure 5). It is clear that the transient effects call for the use of complex three-dimensional 

models to account for the spatial and temporal variability of streamlines. Based on hydraulic head 

measurements and water-quality observations, the calibration of flow and transport models is 

essential for an accurate description and prediction of the fate of individual or multiple chemicals. 

The current version of the standard MT3DMS code (Zheng and Wang, 1999) accounts for the 

transport of multiple species and a small range of basic reactions; however, for cases where 

chemical species are assumed to interact in a more complex manner, MT3DMS-based packages 

such as RT3D (Clement, 1997) or PHT3D (Prommer et al., 2003a), which account for a greater 

variety of biogeochemical processes, are available. 


4 

3 

2 

1 

0 

River water 

Caisson (set B-B) 

Caisson (set A-A) 

Multi-Species and Multi-Component Transport Modeling 

No reaction – filtrate 

With reaction – filtrate 

River water 

0 20 40 60 80 100 


Figure 5. Atrazine transport from the river to the aquifer and the pumping well for a case 

with low permeability riverbed/bank (after Ray et al., 2002). 

In those scenarios where the reaction progress of a dissolved chemical strongly depends on 

the concentration of one or more other dissolved species, reactive multi-species models can 

typically provide a better process description. This is particularly important if models are used in a 

predictive mode. Furthermore, if water-quality changes are additionally affected by water-sediment 

interactions, such as mineral dissolution/precipitation and/or ion-exchange reactions, a reactive 

multi-component transport model might need to be applied to explain specific field observations. 

Multi-component models typically use an extensive reaction database. Until recently, those 

databases were (in most cases) confined to the definition of thermodynamic equilibrium reactions, 

which made those models only applicable to systems where (all) reactions proceed relatively fast in 

relation to groundwater flow velocity (local equilibrium assumption); however, in most of the 

recently published models, a wide range of different kinetic reactions and processes can be defined. 

Models with those capabilities can be used to study complex process interactions that lead to non-

intuitive system behavior. In the case of RBF, such examples might be: 

• The reductive dissolution of iron oxides by DOC and the resulting release of sorbed 

heavy metals. 

• The assessment of pesticide mobility during flood events. 

Denitrification, degradation of pesticides, and dissolution of minerals are closely linked to DOC, 

a key component of river water whose concentration can vary depending on the season and flow 

events. Because of the reaeration process, river water is oxygenated compared to groundwater. 

During the induced infiltration process, oxygen-rich river water enters the aquifer, as does DOC. 

On its travel path, DOC from river water is oxidized and either mineralizes completely or is 

transformed to intermediates through bacterial catalysis, together with the organic carbon that is 

perhaps naturally abundant in the aquifer as sediment-bound organic matter. Oxygen in 

the invading water is used as an electron acceptor in the process. Normally, there is sufficient 

carbon available for microbial use; however, oxygen can become in short supply along the flow 

path. Once microbes consume the oxygen, an anoxic zone develops where the nitrate of the 

infiltrating river water and groundwater is used as a substitute electron acceptor. This leads to the 

reduction of nitrate along the flow path. Once nitrate is also depleted, thermodynamically less 

favorable oxidized iron and manganese minerals and/or sulfate might act as alternative electron 

acceptors. Note that the simulation of the oxidation of one or multiple organic substrates using a 

sequence of electron acceptors is routinely applied in the field of bioremediation modeling, mainly 

where the transport and natural attenuation of oxidizable organic contaminants is simulated 

(Barry et al., 2002). 

The simultaneous simulation of RBF-typical denitrification and mineral dissolution reactions can 

be handled by a number of existing codes (for examples, see Table 1). One such code is EASY- 

LEACHER (Stuyfzand and Lüers, 2000), which was used to simulate reactions along a transect of 

the Torgau RBF site on the Elbe River in Germany. EASY-LEACHER is a two-dimensional 

reactive transport code in EXCEL spreadsheet, combining chemical principles with empirical 

rules in an expert system. The code was found useful to attain a first estimate of water quality in 

the production well for a RBF site where the operation of wells is started and for an easy 

calculation of different boundary conditions. In contrast to EASY LEACHER, which uses a 

collection of (one-dimensional) flow tubes to account for the transport of chemicals, the 

FEREACT model (see Tebes-Stevens et al., 1998; Tebes-Stevens and Valocchi, 2000) is a 

Table 1. Examples of Reactive Multi-Component Transport Models 

Model Reference 

CRUNCH Steefel (2001) 

EASY-LEACHER Stuyfzand and Lüers (2000) 

FEREACT Tebes-Stevens et al. (1998) 

PHT3D Prommer et al. (2003a) 

PHAST Parkhurst et al. (1995) 

MIN3P Mayer (1999) 

TBC Schäfer et al. (1998) 

HBGC123D Salvage and Yeh (1998) 

75

76 

two-dimensional, finite element-based transport model that can simulate biodegradation and 

geochemical reactions along, for instance, a vertical transect of the river-aquifer interface. Although 

the model cannot handle the true three-dimensional (and perhaps transient) flow dynamics 

experienced at a RBF site, it accounts for all typical geochemical and microbial reactive processes. 

An example for a fully three-dimensional model is the MODFLOW/MT3DMS-based code 

PHT3D (Prommer, 2002, Prommer et al., 2003a). It combines the previously mentioned 

MT3DMS (Zheng and Wang, 1999) with PHREEQC-2 (Parkhurst and Appelo, 1999), whereby 

the former solves the advection-dispersion equation and the latter accounts for all geochemical 

reactions. Prommer et al. (2003b) recently applied the model to simulate the transport and 

reactive processes that affect the fate of atrazine near a RBF scheme during a flood event. 

For that modeling study, the hydrogeological setting and hydrological characteristics described by 

Ray et al. (2002) were used to demonstrate the potential influence of a microbial lag effect and 

the redox-dependency of atrazine degradation on abstraction water quality during the flood event. 

They considered kinetically controlled mineralization of DOC, dissolution of sediment-bound 

organic matter, growth and decay of atrazine degraders, and microbially mediated atrazine 

degradation. During the flood event, DOC from the river water modifies the redox patterns in the 

aquifer along the flow path to the well screens. The simulations demonstrated that with the flood, 

the concentration of atrazine reaches a peak in a similar pattern to that found in floodwater and 

the concentration in groundwater remains high; however, once atrazine degraders are active, the 

concentration drops significantly (Figure 6, Case 1). On the other hand, if the DOC of the river 

water is somewhat increased, this leads to the formation of an anoxic zone, which promotes 

denitrification, but inhibits atrazine degradation (Figure 6, Case 2). 

Head (m) 

142 

140 

138 

136 

134 

Water level 

Concluding Remarks 

River 

Obs. Pt. 

132 

0 50 

Time (days) 

100 

C (mol/l) 

C (mol/l) 

0.8 

0.6 

0.4 

0.2 

x 104 

5 

4 

3 

2 

1 

x 107 

1 

Oxygen 

reactive case 1 


0 

0 50 100 

Atrazine 

conservative 



0 

0 50 

Time (days) 

100 

0 

0 50 

Time (days) 

100 

It is apparent that the level of modeling can range from a very simple mass balance to very 

complex studies that involve transient flow and biogeochemical reactions. The objective of the 

modeling study, as well as the type, quantity, and quality of data, will dictate what level of 

sophistication is needed. For example, if yield determination is the primary purpose, MODFLOW 

simulations will mostly be adequate. Preliminary investigations to delineate the sources of water 

1.5 

1 

0.5 

x Nitrate 

104 

8 

6 

4 

2 

0 

0 50 100 

x 108 

2 



Atrazine degraders 



Figure 6. Schematic representation of the 3-D model, hydraulic heads, and selected simulated concentrations 

of oxygen, nitrate, atrazine, and atrazine degraders.

entering the well can be carried out with MODPATH or comparable particle tracking tools; 

however, the accuracy of the simulation strongly depends on how well the true values of hydraulic 

conductivity of riverbed/bank materials are represented in the models. Especially for transient 

simulations, accurate estimates of riverbed and bank hydraulic conductivity are critical. For 

transport simulations, simple mixing models (as used by many utilities) may not provide good 

estimates of contaminant concentrations at RBF wells due to lag times involved, transient flow, 

and heterogeneity issues. Models using MT3D for simulating the transport of single or multiple 

contaminants may be adequate for specific problems (for example, worst-case estimates); however, 

the assumption of equilibrium reactions near well screens may not be valid. On the other hand, 

for kinetic reactions, the availability of appropriate rate parameters might not always be 

warranted. Finally, biogeochemical models such as PHT3D are most comprehensive and are ideal 

to study both steady and transient effects; however, the input data sets for such models can be 

extensive, and field data sets that underpin the simulations may not always be readily available. 

For such complex models, input parameters are rarely found in literature and are very much sitespecific. 

On the other hand, the determination of such parameters at the laboratory-scale requires 

both significant time and effort and, additionally, the usefulness of the results for field-scale 

simulations might still not be warranted due to scale-issues. Furthermore, the application of such 

models typically requires users with advanced modeling skills and experience in the areas of 

geochemistry and (groundwater) hydrology. After all, the complexity found in some of the models 

is only a reflection of the complexity and diversity found in nature and natural materials. To find 

and make use of adequate simplifications should be part of the development of the (site-specific) 

conceptual model. Each complex numerical model can also be used as a “simpler” model by fixing 

those parameters that were excluded in the corresponding simple model. Also, it must be noted 

that the parameters used in simpler models are by far not less site-specific than those of more 

complex models. 

A typical example of the “how much complexity is adequate” question is DOC, since it consists 

of thousands of single organic compounds with different biodegradation behavior. The question is 

how many different fractions (concerning biodegradability and sorption behavior) should be 

identified and used in the model to be adequate for the selected level of model discretization, 

reactions included, and kinetic parameters. Based on the application of different models for 

chosen field-sites and qualified sensitivity analyses (benchmarking), a guideline should be 

developed to propose the right type of model for a specific task. The development of reduced 

models, such as the spreadsheet program EASY-LEACHER (Stuyfzand and Lüers, 2000), and the 

improvement of multi-species reaction models should be carried out simultaneously. 

Furthermore, for detailed modeling, the hydraulic component must be able to handle scour, 

deposition, high water levels, flooding, etc. For detailed transport and reaction modeling, varied 

properties of the riverbed material and aquifer material (e.g., organic carbon content, enzymatic 

activity) should be considered. 

Finally, we must mention that besides the widely known models mentioned in Table 1, there are 

many other models that have been applied successfully for the simulation of processes at RBF sites 

for the specific evaluation purposed. 

77

78 

REFERENCES 

Barry, D.A., H. Prommer, C.T. Miller, P. Engesgaard, and C. Zheng (2002). “Modeling the fate of oxidisable 

organic contaminants in groundwater.” Adv. Water Resour., 25: 945-983. 

Clement, T.P. (1997). RT3D - A Modular Computer Code for Simulating Reactive Multi-species Transport in 3- 

Dimensional Groundwater Aquifers, Battelle Pacific Northwest National Laboratory Research Report, PNNL- 

SA-28967. 

Conrad, L.P., and M.S. Beljin (1996). “Evaluation of an induced infiltration model as applied to glacial 

aquifer systems.” Wat. Resour. Bull., 32(6): 1,209-1,220. 

Dillon, P.J., and J.A. Liggett (1983). “An ephemeral stream-aquifer interaction model.” Wat. Resour. Res., 

19(3): 621-626. 

Ferris, J.G., D.B. Knowles, R.H. Brown, and R.W. Stallmann (1962). Theory of Aquifer Tests, U.S. Geological 

Survey Water Supply Paper 1536-E, Government Printing Office, Washington, D.C. 

Grischek, T., D. Schoenheinz, and W. Nestler (2002). “Unexpected groundwater flow beneath the River Elbe at the 

bank filtration site Meissen, Germany.” Water resources and environment research, ICWRER 2002, Vol. 1: 116-120. 

Hantush, M.S. (1959). “Analysis of data from pumping wells near a river.” Jour. Geophys. Res., 64(11): 1,921-1,931. 

Harbaugh, A.W., E.R. Banta, M.C. Hill, and M.G. McDonald (2000). MODFLOW-2000, the U.S. Geological 

Survey modular ground-water model — User guide to modularization concepts and the Ground-Water Flow Process, 

U.S. Geological Survey Open-File Report 00-92, 121 p. 

Macheleidt, W., W. Nestler, and T. Grischek (2002). “Determination of hydraulic boundary conditions for 

the interaction between surface water and groundwater.” Sustainable groundwater development, Geological 

Society, London, Special Publ. 193: 235-243. 

Mälzer, H., J. Schubert, R. Gimbel, and C. Ray (2003). “Effectiveness of riverbank filtration sites to mitigate 

shock loads.” Riverbank Filtration: Improving Source Water Quality, Kluwer Academic Publishers, Dordrecht, 

The Netherlands. 

Mayer, K.U. (1999). A numerical model for multicomponent reactive transport in variably saturated porous media, 

Ph.D. thesis, University of Waterloo, Waterloo, Ontario, Canada. 

Parkhurst, D.L., P. Engesgaard, and K.L. Kipp (1995). “Coupling the geochemical model PHREEQC with a 

3D multi-component solute transport model.” Fifth Annual V.M. Goldschmidt Conference, Penn State 

University, University Park Pennsylvania, USA, May 1995. 

Parkhurst, D.L., and C.A.J. Appelo (1999). User’s guide to PHREEQC - A computer program for speciation, 

reaction-path, 1D-transport, and inverse geochemical calculations: Technical Report 99-4259, U.S. Geological 

Survey Water-Resources Investigations Report. 

Pollock, D.W. (1994). User’s Guide for MODPATH/MODPATH-PLOT, Version 3: A particle tracking postprocessing 

package for MODFLOW, the U.S. Geological Survey finite-difference ground-water flow model, 

U.S. Geological Survey Open-File Report 94-464. 

Prommer, H. (2002). PHT3D: A Reactive Multicomponent Transport Model for Saturated Media, User’s Manual 

Version 1.0, Contaminated Land Assessment and Remediation Research Centre, The University of Edinburgh, UK, 

http://www.pht3d.org. 

Prommer, H., D.A. Barry, and C. Zheng (2003a). “PHT3D - A MODFLOW/MT3DMS based reactive multicomponent 

transport model.” Ground Water, 42(2): 247-257. 

Prommer, H., J. Greskowiak, P.J. Stuyfzand, and C. Ray (2003b). “Geochemical transport modeling of water 

quality changes during managed artificial recharge.” MODFLOW and More 2003: Understanding through 

Modeling, Proceedings of the International Ground Water Modeling Conference, Golden, Colorado USA, 

16-19 September 2003.

Ray, C., T.W. Soong, Y.Q. Lian, and G.S. Roadcap (2002). “Effect of flood-induced chemical load on filtrate 

quality at bank filtration sites.” J. Hydrol., 266: 235-258. 

Salvage, K.M., and G.T. Yeh (1998). “Development and application of a numerical model of kinetic and 

equilibrium microbiological and geochemical reactions (BIOKEMOD).” J. Hydrol., 209(1-4): 27-52. 

Schäfer, D., W. Schäfer, and W. Kinzelbach (1998). “Simulation of processes related to biodegradation of 

aquifers 1. structure of the 3D transport model.” J. Contam. Hydrol., 31(1-2): 167-186. 

Steefel, C.I. (2001). GIMRT, Version 1.2: Software for Modeling Multicomponent, Multidimensional Reactive 

Transport. Users Guide, Technical Report UCRL-MA-143182, Lawrence Livermore National Laboratory, 

Livermore, California, 2001. 

Stuyfzand, P.J., and F. Lüers (2000). “Modelling the quality changes upon artificial recharge and bank 

infiltration.” Principles and user`s guide of EASY-LEACHER 4.5, KIWA-report SWI 99.199. 

Tebes-Stevens, C.L., and A.J. Valocchi (2000). “Calculation of reaction parameter sensitivity coefficients in 

multicomponent subsurface transport models.” Adv. Water Resour., 23: 591-611. 

Tebes-Stevens, C.L., A.J. Valocchi, J.M. VanBriesen, and B.E. Rittmann (1998). “Multicomponent transport 

with coupled geochemical and microbiological reactions: Model description and example simulations.” 

J. Hydrol., 209: 8-26. 

United States Environmental Protection Agency (2002). Long-Term 1 Enhanced Surface Water Treatment 

Rule, Final Rule. Federal Register, 67:9:1812 (January 14, 2002). 

van Breukelen, B.M., C.A.J. Appelo, and T.N. Oltshoorn (1998). “Hydrogeochemical transport modeling of 

24 years of Rhine water infiltration in the dunes of the Amsterdam water supply.” J. Hydrol., 209: 281-296. 

Wilson, J.L. (1993). “Induced infiltration in aquifers with ambient flow.” Wat. Resour. Res., 29(10): 3,503-3,512. 

Zheng, C., and P.P. Wang (1999). MT3DMS: A modular three-dimensional multispecies model for simulation of 

advection, dispersion and chemical reactions of contaminants in groundwater systems, Documentation and User’s 

Guide, Contract Report SERDP-99-1, U.S. Army Engineer Research and Development Center, Vicksburg, MS. 

CHITTARANJAN RAY is an Associate Professor in the Department of Civil & 

Environmental Engineering and an Associate Researcher with the Water Resources 

Research Center at the University of Hawaii at Mañoa. His current research interests 

include riverbank filtration, pesticides in drinking-water wells, and the flow and transport 

of pathogens and chemicals in saturated/unsaturated media. Over the past 2 years, he has 

edited two books on riverbank filtration and written a monograph on pesticides in 

domestic wells. Among his honors, he is a recipient of the Fulbright faculty scholarship for 

conducting riverbank filtration-related research in Nepal, India, and Bangladesh. Prior to joining the 

University, Ray worked as a staff engineer with the groundwater consulting firm of Geraghty & Miller, Inc. 

(now called Arcadis Geraghty & Miller) and in the Groundwater Section of Illinois State Water Survey, 

conducting various groundwater quantity and quality investigations, including bank filtration. Ray received 

a B.S. in Agricultural Engineering from Orissa University of Agriculture and Technology in India, an M.S. 

in Agricultural Engineering from the University of Manitoba in Canada, an M.S. in Civil Engineering from 

Texas Tech University, and Ph.D. Civil and Environmental Engineering from the University of Illinois at 

Urbana-Champaign. 

79



and Its Effects on Riverbank-Filtration Sites 

Dipl.-Ing. Matthias Krueger 

Fernwasserversorgung Elbaue-Ostharz GmbH 

Torgau, Germany 

Dipl.-Ing. Ingbert Nitzsche 


Torgau, Germany 

Introduction 

To historians, Torgau, Germany, is mainly known as the former residence of the Saxon Electors 

and the cultural capital of sixteenth-century Saxony (Figure 1). A short time later, Martin Luther’s 

work gave the town on the Elbe the nickname, “Nursemaid of the Reformation.” Finally, many 

people know Torgau as the historic site where the defeat of Nazi Germany was symbolized by the 

meeting of Soviet and American troops on the Elbe River on April 25, 1945. 

Figure 1. Air view of Torgau, Germany, 2 days before the flood peak. 

Torgau is the headquarters of the Fernwasserversorgung Elbaue-Ostharz GmbH, one of the largest 

drinking-water supply companies in Germany. The company was founded about 50 years ago and 

manages one waterworks at a reservoir in the Harz Mountains, five bank-filtration waterworks in the 

Elbe River Basin near Torgau, and a 700-kilometer long distribution network. In 2002, the 

drinking-water production was 76-cubic meters. About 3.5-million people were supplied with 

drinking water of high quality. 


Dipl.-Ing. Matthias Krueger 

Head of Laboratory 


Naundorferstraße 46 • 04860 Torgau, Germany 

Phone: +49 3421 757511 • Fax: +49 3421 757522 • Email: matthias.krueger@fwv-torgau.de 

81

82 

Riverbank Filtration Near Torgau 

In this system, raw-water abstraction via bank filtration in the Elbe River Basin is a key element. 

The production boreholes are located in an alluvial sand and gravel aquifer with a thickness of 

40 to 60 m, covered by a 2- to 5-m thick layer of meadow loam. The meadow loam provides an 

important protection against pollution and infiltration during flooding. Due to erosive conditions, 

there is only low clogging of the riverbed. The good hydraulic connection between the Elbe River 

and the adjacent aquifer ensures stable water abstraction during low flow periods. 

Figure 2 shows the biggest waterworks, Torgau-Ost, which has a maximum capacity of 

120,000 m 3 /d. Raw water is abstracted from 42 production boreholes on the western bank of the 

Elbe River. At a distance of about 300 m between the production boreholes and the riverbank, 

the flow time of the bank filtrate is between 60 and 200-plus days. 

Figure 2. Torgau-Ost Waterworks and the Elbe River. 

The produced water has a good quality due to the natural purification ability of the aquifer. Two 

monitoring cross-sections positioned along assumed flowpaths to the production boreholes have 

been installed to control borehole operation and to guarantee flexibility in response to environmental 

impacts, both seasonal and long-term. These monitoring cross-sections include up to five 

observation stations between the production boreholes and the river. Each observation station 

consists of three to five monitoring points at varying depths. 

Raw water is purified by aeration and deacidification, pre-purification by 20 tube sedimentation 

basins, and fine purification by 16 open sand filters. Iron, manganese, and carbon dioxide are the 

main constituents of raw water that require treatment. These constituents are eliminated by 

adding hydrated lime and potassium permanganate. Prior to pumping into the public-water supply 

network, the water is further treated with small amounts of chlorine and chlorine dioxide to 

prevent bacteriological deterioration during the long transportation process to the consumer. 


In Torgau, the mean discharge of the Elbe River is about 330 cubic meters per second (m 3 /s). In 

August 2002, a so-called “Flood of the Century” occurred. The event was caused by a “Vb” (extreme) 

weather situation. Precipitation reached 300 millimeters per day, with 25 to 30 millimeters per hour 

in the Ore Mountains. This intensity had never been measured before in this region. In Dresden, 

the Elbe River exceeded the 1845 flood level of 8.77 m, hitting a record of 9.40 m on August 17, 

2002. The towns of Meissen, Torgau, Wittenberg, and Dessau were also partially flooded.

In Torgau, the water level of the Elbe River increased within 10 days by about 8 m up to 9.47 m 

on August 19, a level that has never been observed before. The peak discharge of the Elbe River 

in Torgau was estimated to have been 4,295 m 3 /s. The City of Torgau had to be evacuated 

completely. Technical staff, soldiers, and helpers built anti-flood barricades with thousands of 

sandbags. 

To ensure a safe drinking-water production, the water company formed an emergency task force 

on August 13 to organize necessary measures. 

The Effects of the Flood 

Most bank-filtration waterworks near Torgau are located at elevated levels at a sufficient distance 

to the river and are normally not affected by flooding. Nearly all production boreholes were flooded, 

but still functioned due to their special construction. A problem arose with the power-supply system 

for the boreholes. Transformer stations are located behind the dikes. But, due to the high water 

level and risk of dikes breaking, the stations had to be extra protected; therefore, the stations were 

housed-in using sheet piling and equipped with pumps for dewatering and additional emergency 

power generators (Figure 3). 

Figure 3. Protected transformer station behind a dike. 

As a result of these measures, the power supply was maintained at most locations. Only two small 

waterworks in the north of Torgau had to be abandoned due to water influx at transformer stations. 

Besides the aspect of water quantity, there is water quality. As a result of former bank-filtration 

research programs at the Torgau site, there is an excellent knowledge of flow processes and 

water-quality changes along the flowpath. Based on this knowledge, no significant change in 

raw-water quality was expected, mainly because the flow time of the bank filtrate ranges between 

60 and 300-plus days. For control purposes, the continuous monitoring of raw-water quality was 

intensified during the flood, and a special groundwater-monitoring program for the cross-sections 

was set up after the flood. 

All results of measurements at the outflow of the waterworks proved that drinking-water quality 

was not at risk at any time. The quality of the production water met the German standards for 

83

84 

drinking-water quality. Drinking-water disinfection was done by 0.2 mg/L chlorine dioxide (ClO 2) 

and 0.6 mg/L chlorine (Cl 2). There was no problem with bacterial contamination. 

But what about raw-water quality? The flood caused a strong increase in turbidity, organic carbon 

concentration, and number of microorganisms in river water. The increase in DOC from 5 mg/L 

to more than 10 mg/L in river water did not affect the DOC concentration in the raw water 

(Figure 4). The increase in DOC in river water was mainly caused by an increase of the 

biodegradable fraction of DOC, which is consumed along the flowpath of the bank filtrate. Thus, 

biodegradation and mixing in the aquifer prevented an increase in DOC concentration in the raw 

water and an increase in DBPs. 

40 

35 

30 

25 

20 

15 

10 

Increases in concentrations of organic trace compounds such as polynuclear aromatic hydrocarbon, 

chlor-organics, and pesticides were not observed in Elbe River water during the flood. Despite 

many reported inputs of contaminants, the huge discharge caused an effective dilution and 

minimized the risk. After the flood, concentrations were found to be around the mean annual 

values. Despite these facts, measurements of the sum of adsorbable organic halogenated 

compounds (AOX) were used to check for the potential contamination of river water and raw 

water. The AOX concentration in the raw water remained at a low level of less than 30 µg/L, 

giving no indication of contamination (see Figure 4). Measured DOC and AOX concentrations 

in bank-filtrate samples were found to be within long-term concentrations ranges. 

Increased heavy metal concentrations in the river water also did not affect raw-water quality. 

Conclusions 

5 

0 

Elbe River 

4/29/02 

Elbe River 

8/20/02 

Borehole 22 

9/4/02 

Raw water 

10/7/02 

AOX (µg/l) 

DOC (mg/l) 

Figure 4. DOC and AOX concentrations in Elbe River water and raw water 

The 100-year flood of the River Elbe in August 2002 became the greatest challenge for 

Fernwasserversorgung Elbaue-Ostharz GmbH in its history. Maintaining the power supply for 

pumps in the production boreholes was the most important factor, whereas the quality of 

abstracted bank filtrate was never at risk due to the design of bank-filtration sites. Based on the 

measures taken and management of efficient distribution network, the water company ensured a

stable water supply for millions of people during the flood without restrictions to quantity and 

quality. The available technology for drinking-water treatment was successful in producing 

high-quality drinking water. 

Since 1990, MATTHIAS KRUEGER — a process engineer and chemist — has worked for 

the Fernwasserversorgung Elbaue-Ostharz GmbH, a water company that provides drinking 

water for over 3.5-million people in Germany. Early on, he was responsible for water 

treatment technology, then became Head of Laboratory in 2001. Under his leadership, the 

Laboratory has recently been involved in large research projects on physical and chemical 

processes during bank filtration. Prior to joining Fernwasserversorgung Elbaue-Ostharz 

GmbH, Krueger was an Engineer at Galvanotechnic in Leipzig, Germany, for 7 years. His 

areas of interest include flow path, residence times, and redox conditions in the behavior of pollutants during 

riverbank filtration. Krueger received a diploma (Dipl.-Ing.) in Chemistry from the Technical University 

Ilmenau and completed post-graduated studies in analytics and spectroscopy at Leipzig University. 

85


Temporal Changes of Natural Attenuation Processes 

During Bank Filtration 

Paul Eckert, Ph.D. 



Rudolf Irmscher, Ph.D. 



RBF is a well-proven natural purification step in water supply. Sustainable water management 

should be based specifically on natural purification methods, as declared by the International 

Association of Waterworks in the Rhine Catchment Area, in their 2003 memorandum. To 

achieve this aim, a profound knowledge of the purification capacity of bank filtration is essential. 

At the Düsseldorf Waterworks in Germany, the influence of long-term — as well as periodic — 

changes of both hydraulics and river-water quality on natural attenuation processes were 

investigated. 

The improvement of Rhine River water quality over the last 30 years enabled the Düsseldorf 

Waterworks to reduce their technical treatment expenses. Temperature variations throughout the 

year and flood events significantly influenced the purification capacity of bank filtration. This 

reinforces the need for flexible technical treatment methods capable of adapting to changing rawwater 

quality. 

Even though the complete replacement of subsequent technical treatment steps might be seen as 

an unreachable vision, the substantial knowledge we have acquired on the purification capacity 

of bank filtration enables the design of tailor-made treatment methods. 

Site Description and Treatment Concept 

The City of Düsseldorf is situated in northwestern Germany, in the lower Rhine Valley (Figure 1). 

The Düsseldorf Waterworks supply 600,000 inhabitants with treated bank filtrate. A multiprotective 

barrier concept ensures the constant production of high-quality drinking water (Figure 2). 

Natural attenuation processes during bank filtration form the first and most efficient protective 

barrier. The subsequent protective barrier is raw-water treatment, including ozonation, biological 

active filtration, and active carbon adsorption. 

The Rhine River has a length of 1,320 kilometers and a catchment area of 185,000 square kilometers; 

it is the third biggest river and the largest source of drinking water in Europe. The mean discharge 

of the Rhine at Düsseldorf is 2,200 m 3 /s, while the Waterworks use less than 2 m 3 /s. During flood 

events, discharge increases up to 9,900 m 3 /s. 


Paul Eckert, Ph.D. 

Head of the Water Management Department 


Abt. Wasserwirtschaft • Höherweg 100 • 40233 Düsseldorf, Germany 

Phone: +0211/8218359 • Fax: +0211/821778359 • Email: peckert@swd-ag.de 

87

88 

Protection Zones 

II 

IIIa 

IIIb 

Local Frontier 

0 1 2 3 4 5km 

Rhine 

Krefeld 

The production wells, situated at a distance of 50 to 350 m of the riverbank, discharge water from 

a sandy and gravel aquifer. Depending on the hydraulic situation, the residence time of bank 

filtrate in the aquifer varies between weeks to several months. 

From a water-resources perspective, bank filtration is characterized by an improvement in water 

quality (Kühn and Müller, 2000). The most important effects of bank filtration include: 

• Removal of particles and turbidity. 

Duisburg 

Neuss 

Figure 1. Geographic location of the Düsseldorf Waterworks. 

Corg 

Inactivation 

of virusis and 

pathogena 

CO 2 

Staad Waterworks 

Düsseldorf 

Lörick Waterworks 

Flehe Waterworks 

Auf dem Grind Well Field 

Ozonation 

NBG GmbH 

Figure 2. The multi-protection barrier concept for drinking-water production. 

Mettmann 

Mettmann 

Rhine 

N 

Biological active filtration 

Activated carbon adsorption 

upper layer 

lower layer

• Equalization of fluctuating concentrations in river water. 

• Removal of biodegradable compounds. 

• Removal of bacteria, viruses, and parasites. 

The efficiency of these natural attenuation processes is influenced by river-water quality and the 

hydraulic situation. Continuous and periodic changes of water quality must be considered together 

with fluctuating river-water levels. 

Temporal Changes Affecting Bank Filtration 

During the last 30 years, Rhine River water quality has improved significantly. Many actions to 

reduce nutrients and pollutants were necessary. These measures comply with the best available 

technology in production, as well as wastewater treatment along the Rhine. The return of salmon 

in the year 2000 is a visible sign of the success of this program. 

These quality improvements were accompanied by a decrease of ammonia and DOC in river 

water. Oxygen concentrations reached saturation. The higher oxidation capacity of bank filtrate 

is linked to more efficient natural attenuation processes within the aquifer. This enabled the 

Waterworks to reduce treatment expenses. 

Natural attenuation processes are affected periodically by flood events. The water level increases 

up to 5 m, which leads to a higher influx of river water into the aquifer. Thus, oxygen and organic 

carbon flux also increase by a factor of 3 to 4. The aerobic microbes adapted to the lower flux are 

not able to degrade the entire organic carbon during flood events. Another decrease of natural 

attenuation processes becomes evident in the removal of bacteria. While mostly raw water already 

fulfills the European Drinking Water Standard, higher colony counts are observed in the 

production wells following flood events (Irmscher and Teermann, 2002; Schubert, 2002). These 

findings explain the need of subsequent purification processes: the so-called “second protective 

barrier” (see Figure 2). 

Another effect on natural attenuation processes during bank filtration is caused by changing water 

temperatures throughout the year. As the temperature of the river rises quickly in the spring and 

summer, the temperature of the water in the aquifer also rises, but not as quickly due to a certain 

time delay. Microbial activity related to temperature leads, therefore, to a higher degradation rate 

of organic carbon. The organic carbon is removed more efficiently; however, it must be considered 

that, under aerobic conditions, the bank filtrate becomes more aggressive versus calcite. A flexible 

treatment step must be adaptable to these changing conditions. 

Conclusions 

To reduce technical treatment expenses, river water of high quality, together with a profound 

knowledge of the natural attenuation processes within the aquifer, is essential. At the Rhine River, 

successful efforts have improved water quality significantly. More efficient natural attenuation 

processes accompanied this improvement during bank filtration, which enabled the Waterworks 

to reduce expenses in water treatment. 

Temporal changes of river-water quality and hydraulics still influence natural attenuation processes 

during bank filtration. They have to be well-understood to design adequate treatment steps and 

to define specific target values on river-water quality. The multi-protective barrier concept, 

including both natural and technical purification, is still necessary to ensure drinking water of 

continuously high standards. 

89

90 

REFERENCES 

Irmscher, R., and I. Teermann (2002). “Riverbank filtration for drinking water supply – A proven method, 

perfect to face today’s challenges.” Water Science and Technology, 2(5-6): 1-8. 

Kühn, W., and U. Müller (2000). “Riverbank filtration.” Journal AWWA, 92(12): 60-69. 

Schubert, J. (2002). “Water-Quality improvement with riverbank filtration at Düsseldorf Waterworks in 

Germany.” Riverbank Filtration: Improving Source-Water Quality, C. Ray, G. Melin, and R.B. Linsky (eds.), 


PAUL ECKERT has more than 10 years of professional experience in water-management 

projects. He was involved in the project for bank filtration (risk assessment of shock loads) 

funded by the Federal Ministry for Research and Technology (BMFT). Currently, as Head 

of the Water Management Department, he is responsible for protection zone management 

at Düsseldorf Waterworks, where special investigations are performed in the field of bank 

filtration and groundwater remediation. A hydrogeochemist, he specializes in 

hydraulic/hydrochemical modeling, the design of field studies, and the assessment of 

groundwater-quality problems. Extended experience includes the application of a geographic information 

system database for groundwater. Eckert received both an M.S. (Diplom) and Ph.D. from the University of 

Bochum in Germany, where he researched the efficacy of enhanced natural attenuation processes in a BTEXcontaminated 

aquifer with nitrate.


An Update of the City of Guelph’s Response 

to Regulation 459/00: Effective Natural In Situ 

Filtration of Several Groundwater Under the Direct 

Influence of Surface-Water Supplies 

Dennis E. Mutti, P.E. 

Associated Engineering Limited 

Toronto, Ontario, Canada 

Peter Busatto 

City of Guelph Waterworks Division 

Guelph, Ontario, Canada 

Douglas H. Stendahl 

City of Guelph Waterworks Division 

Guelph, Ontario, Canada 

Caroline Korn, P.E. 



Elia Edwards, P.E. 



Associated Environmental Limited developed a protocol for determining GWUDI status to assist 

the City of Guelph in Ontario, Canada, in assessing treatment requirements for their municipal 

water supply. This protocol is based on protocols established by the United States Environmental 

Protection Agency (USEPA) and further defined in American Water Works Association Research 

Foundation Project #605. The protocol entails three stages of assessment. Stage 1 is a review of 

existing information to characterize the well as a true groundwater or GWUDI. This data includes 

well construction and maintenance records, sanitary condition of the well, hydrogeological data 

including time-of-travel estimates (well-head delineation), geological characteristics, and waterquality 

data. Stage 2 is initiated if there is insufficient data to make a characterization of the nature 

of the well in Stage 1 and consists of a period of data collection that will allow the characterization 

to be completed. Wells that are characterized as GWUDI are then subject to a Stage 3 assessment, 

which consists of a data collection period followed by an assessment of the level of natural or in situ 

filtration that is occurring. If sufficient evidence of in situ filtration is shown in the Stage 3 

assessment, then a case can be made for waiving the chemically assisted filtration component of the 

minimum treatment requirement for GWUDI in the Province of Ontario. The minimum treatment 

requirements in the Province of Ontario for GWUDI sources with effective in situ filtration is 3-log 

inactivation for Giardia cysts and 4-log inactivation for viruses from disinfection alone. 


Dennis Mutti, P.E. 

Senior Water Treatment and Supply Engineer 


525-21 Four Seasons Place • Toronto, Ontario M9B 6J8 Canada 

Phone: (416) 622-9502 • Fax: (416) 622-6249 • Email: muttid@.ae.ca 

91

92 

The City of Guelph’s supply system consists of a network of 21 groundwater wells and a spring 

collector system, augmented by artificial recharge. The Stage 1 assessment was completed for all 

of the City’s wells as a part of the provincially mandated Engineers’ Report. Two bedrock wells 

(Burke Well and Downey Well) and an overburden well system (Carter Wells 1 and 2) were 

identified as requiring a Stage 2 assessment due to insufficient existing data. The collector system 

(Arkell Spring and Glen Collector) was identified as GWUDI by definition and required a Stage 3 

assessment. The City of Guelph decided to expedite the process and collect data required for a 

Stage 3 assessment for the three wells, as well as for the surface-water recharge and collector system 

in case the wells were determined to be GWUDI. 

The results of the GWUDI assessment were presented at a stakeholders meeting with key 

provincial regulators. Consensus was reached as follows: 

• Burke Well and Downey Well are representative of true groundwater supplies and, as such, 

must provide disinfection as mandated by the Drinking Water Protection Regulation 

(2.0-log virus inactivation). 

• The treatment requirement to provide chemically assisted filtration for the Carter Wells, 

Arkell Recharge System, and Glen Collector System should be waived as effective 

natural in situ filtration is provided through the subsurface. As such, the treatment 

requirements for Giardia cysts (3-log inactivation) and viruses (4-log inactivation) may 

be provided by disinfection only. 

The City of Guelph is currently negotiating treatment requirements, as well as the final wording 

in its Consolidated Certificate of Approval, with the Ontario Ministry of the Environment. 

DENNIS MUTTI has 13 years of progressive experience as a Project Manager and Process 

Engineer, the last 5 of these years with Associated Engineering, which provides environmental 

engineering consulting services to the Ontario, Canada, market. His focus has been on 

process evaluation and development, and the design, implementation, and optimization of 

water supply and treatment systems. Mutti has worked on all types of water-supply and 

treatment projects, including water-supply master plans, water treatment plant designs and 

upgrades, groundwater supplies, automation and Supervisory Control and Data Acquisition 

(SCADA) system implementation, and start-up and commissioning. He received both a B.S. in Chemical 

Engineering and an M.S. in Civil Engineering from the University of Waterloo.


On Bank Filtration and Reactive Transport Modeling 

Dr.-Ing. Ekkehard Holzbecher 

Humboldt University 


Prof.-Dr. Gunnar Nützmann 

Humboldt University 


Modeling can enhance the understanding of the influence of hydraulic, transport, and biogeochemical 

processes for the development of bank filtration as an applied technology. Because a 

coupled three-dimensional approach, including all biogeochemical transformations, speciation, 

and kinetics, is not feasible nowadays in applied projects, a simpler procedure is proposed. In 

the first step, flow is calculated using analytical solutions or numerical models in two- or 

three-dimension. In the second step, reactive transport is simulated in one-dimension along 

flowpaths. How these two steps are performed in a project depends on the chosen software for flow 

and reactive transport. A simple approach is demonstrated that uses analytical solutions for flow and 

the pH redox equilibrium equation (PHREEQC) for reactive transport. This approach is appropriate 

for understanding processes in porous media in the direct vicinity of a surface-water body. 

Flow Modeling 

The simulation of reactive transport at a field site is always based on a flow model. Flow models 

can account for: 

• Inhomogeneities. 

• Anisotropies. 

• Site-specific design of the well system. 

• Irregular (usually) setting of the boundaries. 

Rather complex two-dimensional or three-dimensional flow patterns emerge. 

Although the capability of software has increased in conjunction with the power of hardware, 

there are still limiting factors for solving general three-dimensional reactive transport models. 

Transport models require a small spatial resolution (e.g., a high number of nodes or blocks) in 

which dependent variables are computed. Biogeochemistry requires a small temporal resolution to 

capture the changes of some variables in response to changes in the system. 

Numerical errors and their propagation when solving huge linear and/or nonlinear systems are 

hard to predict. If some well-known constraints (grid Péclet-number criterion, Courant-number 

and Neumann-number criteria) are not fulfilled, numerical errors increase dramatically. The 


Dr.-Ing. Ekkehard Holzbecher 

Senior Scientist 

Humboldt University Berlin 

Institute of Freshwater Ecology and Inland Fisheries • Müggelseedamm 310 • 12587 Berlin, Germany 

Phone: +0049-30-64181 667 • Fax: +0049-30-64181 663 • E-Mail: holzbecher@igb-berlin.de 

93

94 

resulting numerical dispersion or oscillations disturb results in that the results cannot be used for 

comparisons with observations. 

To couple the result of a flow model with a general geochemical approach, as given by the 

PHREEQC code, the flow system must be reduced to one-dimension. Flowpaths can be taken from a 

two- or three-dimensional model and be used for a one-dimensional reactive transport simulation. 

The procedure is applicable for bank-filtration systems, as concentration gradients in transverse 

directions can be assumed to be much smaller than in longitudinal directions. 

In the first step towards such a simulation, flowpaths are taken from analytical solutions for 

idealized well gallery systems. MATLAB (2002) software is applied for the analytical solution. 

Systems with different numbers of wells, different distances from the surface-water body, and 

different pumping rates can be investigated. It is also possible to introduce a groundwater base 

flow. Moreover, different types of boundary conditions at the bank can be considered. 

Figures 1 and 2 show the results of a simulation for two wells. The calculations with analytical 

solutions, as well as the graphical representation, were made using MATLAB. 

Figure 1. Hydraulic potential φ (m 3 /s) for two wells with different pumping rates and distances from a 

surface-water body near a gaining stream. 

Some basic results can be obtained for a single well near a gaining stream. There is a critical 

pumping rate, Q crit , below which only groundwater is pumped. By increasing the pumping rate 

above Q crit , the share of surface water (filtrate) increases. An equal share between groundwater 

and surface water is given for: 

Q = 6.1878 •Q crit 

(Equation 1)

Figure 2. Streamlines for two wells with different pumping rates and distances from a surface-water body near 

a gaining stream. 

We intend to further explore the use of analytical solutions. A numerical algorithm, which 

calculates flowpaths and travel times from the surface-water body to the wells, is currently being 

tested. If combined with geochemical computations, the approach can provide a highly important 

tool for the design of well galleries. 

Travel times along flowpaths are obtained by numerical integration. 

Reactive Transport Modeling 

With the outlined strategy, the well-established PHREEQC code (new version: PHREEQC2) can 

be applied for reactive transport modeling to reduce the number of spatial dimensions. The 

PHREEQE code (which is the origin of PHREEQC) was mainly a geochemical speciation code 

and originally not intended to be combined with a transport model. Velocity is not an input to be 

specified. Instead, lengths (∆x), time step (∆t), cells (number of blocks), and shifts (number of time 

steps) must be given as parameters (Parkhurst, 1995). Nevertheless, the simulated velocity v 

results from the input parameters: 

ν = ∆x / ∆t 

(Equation 2) 

The so-called “mixing cell approach” for advection, which uses the operator-splitting technique, is 

combined with the finite difference method for diffusion and proves to be competitive with advanced 

numerical techniques. Errors from the discretization advection term are successfully suppressed. 

PHREEQC can also be used for variable velocity along flowpaths. The time step in Equation 2 is 

fixed, but grid spacing varies locally, taking velocity changes into account. Small velocities result 

in small ∆x and high velocities in large ∆x. 

95

96 

As an alternative to PHREEQC for certain tasks, a new model based on MATLAB is currently in 

development. The model is being tested in a number of benchmark studies. Figure 3 shows the 

results of the MATLAB model for a test of redox sequences, as compared with PHREEQC output. 

Concentration [mol/L] 

5.00E-03 

4.50E-03 

4.00E-03 

3.50E-03 

3.00E-03 

2.50E-03 

2.00E-03 

1.50E-03 

1.00E-03 

5.00E-04 

CH20 MATLAB 

N(0) PHREEQE 

S-2 MATLAB 

02 PHREEQC 

CH20 PHREEQC 

N(-3) MATLAB 

S(-2) PHREEQC 

pH MATLAB 

N(5) MATLAB 

N(-3) PHREEQC 

HCO3 MATLAB 

pH PHREEQC 

N(3) PHREEQC 

S(6) MATLAB 

HCO3 PHREEQC 

pe MATLAB 

Figure 3. Reactive transport (redox) test case: Redox zoning after a 1-year simulation timeframe. 

N(0) MATLAB 

S(6) PHREEQC 

O2 MATLAB 

pe PHREEQC 

0.00E-00 

–6.00 

0.00 10.00 20.00 30.00 40.00 50.00 

x [m] 

60.00 70.00 80.00 90.00 100.00 

In the MATLAB model, valence electron balances are obtained using an adequate conceptualization 

of operational valence electron balancing. Redox equilibrium modeling, including kinetics 

for organic carbon biodegradation and operational valence electron balancing to the equilibrium 

module, yields automatically to a correct choice of electron acceptor. 

For non-redox equilibrium systems, which incorporate proton and component mass balances and 

mass action equations, the operational valence electron balance and the electron (e – ) as formal 

species entering mass action equations — derived from redox half reactions — are considered. 

Since free electrons are generally not observed in aqueous solutions (Thorstenson, 1984), 

operation valence electron balancing has to be carried out by adapting the share of different 

valence states of heterovalent components (in a sense, valence electron “bookkeeping”) (Appelo 

and Postma, 1996). 

In inorganic redox systems, the sum of mobile operational valence electrons is an additional 

transport species. In the presence of biodegradation redox reactions (which supply additional 

carbonate species to the system), the actual operation electron balance has to be corrected by the 

carbonate species arising from biodegradation, as pointed out by Brun and Engesgaard (2002). 

A test case including hydrodynamic transport to the redox system was set up using MATLAB. At 

the inlet of the model column, a solution that is rich in mobile organic matter infiltrates into an 

aquifer void of organic matter, but contains oxygen, nitrate, and sulphate as electron acceptors. 

Unlike Redox Test Case 1, ammonium (N[–3]) is included in the equilibrium reaction framework. 

During infiltration, a steady-state redox zone is simulated similar to the redox zone observed at 

16.00 

14.00 

12.00 

10.00 

8.00 

6.00 

4.00 

2.00 

0.00 

–2.00 

–4.00 

pH. pE [–]

well galleries near the Unterhavel (Berlin). The distance from the inlet at which the redoxcline 

(a large jump in the redox state) is expected depends on: 

• Hydrodynamic parameters in relation to the first-order organic matter decay constant. 

• Initial redox state (operational valence electron balance). 

• Initial electron-acceptor concentrations. 

• Availability of reactive organic matter. 

These are the main sensitivity parameters that determine the extent of redox zoning. 

Figure 3 shows the simulated steady-state species and pH/oxidation-reduction (redox) potential (pE) 

distribution along the model transect. The simulation reached steady-state concentrations after a 

150-day simulation time. The pE profile shows the typical variation caused by aerobic respiration 

(pE buffering at +15), denitrification (pE buffering at +12 to +14), and sulphate reduction 

(pE buffering at –3). The profiles of oxygen (O 2) concentrations, nitrate (N[5]), and sulphate 

(S[6]) concentrations also are figured so that the changes of pE are explained straightforwardly. 

Because ammonium (N[–3]) is incorporated into the speciation, elemental nitrogen (N[0]) will 

not be the end step of the nitrogen species, but will be reduced to ammonium in conjunction with 

the sulphate-reduction step. 

Outlook 

The outlined coupling strategy for flow and reactive transport is intended to be applied on real 

sites. Flow simulation will be done using MODFLOW (Harbaugh et al., 2000) or FEFLOW (2002) 

for site-specific models and by analytical solutions for conceptual models concerning well-gallery 

design. For reactive transport, there is the choice between PHREEQC and MATLAB. 

REFERENCES 

Appelo, C.A., and D. Postma (1996). Geochemistry, groundwater and pollution, Balkema, Rotterdam. 

Brun, A., and P. Engesgaard (2002). “Modeling of transport and biogeochemical processes in pollution 

plumes: literature review and model development.” Journ. of Hydrology, 256: 211-227. 

FEFLOW (2002). Finite element, subsurface flow & transport simulation system, WASY, Berlin, Germany. 

Harbaugh, A.W., E.R. Banta, M.C. Hill, and M.G. McDonald (2000). MODFLOW-2000, the U.S. Geological 

Survey modular ground-water model — User guide to modularization concepts and the Ground-Water Flow Process, 

U.S. Geological Survey Open-File Report 00-92, 121p. 

MATLAB (2002). MATLAB the language of technical computing, The MathWorks, Inc., Natick, MA. 

Parkhurst, D.L. (1995). PHREEQC: A computer program for speciation, reaction-path, advective transport, and 

inverse geochemical calculations, U.S. Geological Survey, Water Res. Invest. Report 95-4227, Lakewood, 143p. 

Thorstenson, D.C. (1984). The concept of electron activity and its relation to redox potentials in aqueous 

geochemical systems, Open-File Report 84-072, U.S. Geological Survey Denver, Colorado. 

97

98 

Mathematician EKKEHARD HOLZBECHER has worked the field of groundwater 

science and applications since 1984. Most of his projects are related to numerical 

modeling. Past projects include investigating the safety of nuclear waste repositories in 

Germany, saltwater intrusion in the Nile-Delta in Egypt, geothermal flow in Japan, and 

abandoned industrial sites in Germany. Currently, he is investigating various problems 

regarding eco-hydrology. Since 2002, Holzbecher has been a Senior Scientist with 

Humboldt University in Berlin, where he teaches groundwater and transport modeling 

courses, and participates in the Natural and Artificial Systems for Recharge and Infiltration project. He was 

also involved in the international Groundwater Hydrology Modeling Strategies for Performance Assessment of 

Nuclear Waste Disposal (HYDROCOIN) project and is a member of the United Nations Educational, 

Scientific and Cultural Organization (UNESCO) working group for the Development and Calibration of 

Coupled Hydrological/Atmospheric Models. Holzbecher received a Ph.D. from the Civil Engineering 

Department at the Technical University Berlin and the German degree of Habilitation (qualification for a 

tenure professorship) at the Faculty of Geosciences at the Free University Berlin, where he is now a 

Privatdozent.

Dinner Presentation 

Hydraulic Sensitivities and Reduction Potential 

Correlated with the Distance Between 

the Riverbank and Production Well 

Bernhard Wett, Ph.D. 

University of Innsbruck 

Innsbruck, Austria 

Objectives 

Beginning with a detailed case study of a RBF system at the alpine river, Enns, in Austria, some 

aspects of the question, “In what ways do specified hydraulic parameters affect water quality (e.g., 

what is the appropriate well distance from the riverbank)?” are elaborated. In addition to 

monitoring data, numerical sensitivity analyses will be applied to display interactions between 

well distances, filtrate production, and reduction processes. 

Methodology and Site Description 

The investigated RBF well is situated about 50 m from the bank of the oligotrophic alpine river, 

Enns, at the beginning of the 5-kilometer long reservoir of the HPP Garsten power plant (Figure 1). 

At this particular location, the river stage (as determined by a dam) is 302.0 m above sea level and 

showed only minor elevations when discharge varied from about 70 to 540 m 3 /s during a 1-year 

measurement period. Since the riverbed is cut into the dense flysch zone, river water infiltrates 

almost exclusively through the bank and not the bottom. Between the river and well, aquifer 

thickness is about 5 m, and the total thickness of the gravel layer is 15 m (Ingerle et al., 1999; 

Wett et al., 2002). 

Organic loading is very low (the DOC concentration varies between 1 and 2 mg/L), and river 

water is saturated with oxygen (greater than 10 mg/L). Groundwater quality reflects the trends of 

land use and the intensity of agriculture along the river. The further downstream of the river, the 

higher the nitrate and pesticides concentrations are in groundwater. In the region of the filtration 

site, groundwater shows nitrate concentrations of about 50 mg/L. The enrichment of groundwater 

by river filtrate (less than 5 mg/L nitrate) offers a solution to obtain nitrate concentrations in 

agreement with drinking-water standards. 

A high-grade steel box with windows for visual observation and video recording of riverbed clogging 

was installed in the riverbank. One set of five probes was installed at the upstream side of the box in 

natural sediment (multi-level Probe F at depths between 0.1 and 0.9 m beneath the riverbed surface), 

and two sets of five probes (D and E) were installed downstream in specified filter sand (0 to 4 mm) 

to measure hydraulic heads and obtain water samples. Water levels in the probes correspond with the 

levels of graduated glass pipes within the steel box. An analysis of hydraulic head data and relative 

variations of head differences allows for filter velocity calculations. Higher hydraulic conductivity in 


Bernhard Wett, Ph.D. 

Scientific Researcher and Lecturer 

Institute of Environmental Engineering 

University of Innsbruck • Technikerstr. 13 • A-6020 Innsbruck, Austria 

Phone: +43 512 507 6926 • Email: Bernhard.Wett@uibk.ac.at 

99

100 

Figure 1. Schematic overview and cross-section of the RBF site on the Enns River, a tributary of 

the Danube River. 

the filter sand causes a significantly higher filter velocity than in natural sediment (the mean filter 

velocity is 1.99 × 10 –5 m per second and 0.83 × 10 –5 m per second, respectively). 

Results 

In general, microbial metabolism and degradation processes require both electron donators and 

acceptors. Depending on the site and its inherent boundary conditions, an oligotrophic river 

would have carbon as a limiting factor. The lack of electron donators can be compensated for by 

long hydraulic retention in organic-rich sediment layers at high water temperatures. A long 

retention time (represented by low infiltration rates in Figure 2) in the most biologically active 

zone near the water-sediment interface drives the balance between the availability of electron 

donators and acceptors towards higher oxygen consumption. The arrow indicating the influence 

of hydraulic parameters refers to the distance d from the river to the production well, hydraulic 

Figure 2. Limitations of microbial metabolism in the riverbank and their balancing factors.

conductivity k (especially of the riverbed), aquifer thickness H, and production rate Q BF. 

Additionally, an increase in temperature causes a decrease in water viscosity and, therefore, creates 

higher hydraulic conductivity (this is a 50-percent increase in the presented case study). 

It is a known fact that particulate organic matter deposited in the riverbed represents an important 

carbon pool (Bretschko and Moser, 1993). The metabolized mass of particulate organic carbon can 

be estimated from the respiration of the sediment community to identify the dominant carbon 

source. The difference between total hyporeic respiration and DOC respiration, as depicted in 

Figure 3, describes a substantial gap in the mass balance between carbon fluxes in and out of the 

riverbed. In the case of increased DOC loads of the filtrate, the microbial community can adapt 

as long as electron acceptors are available (e.g., basin maturation for soil-aquifer treatment of 

percolated secondary effluent occurs within a few months) (Quanrud et al., 2003). Reactive 

transport models can describe the relatively quick development from electron donator limitation 

towards electron acceptor limitation (Lensing et al., 1994). 

DOC immobilization / microbial C respiration 

(µ mol C per L) 

DOC immobilization C respiration 

Station C Station D 

Station F Station E 

month (1997/1998) 

(non-DOC)-C respiration 

Figure 3. Seasonal variation of DOC immobilization (∆DOC), total hyporheic carbon respiration (HCR), 

and the difference between ∆DOC and HCR in the riparian zone (0.9 to 1.0 m from the sedimentwater 

interface, with Stations C and F in natural sediment and Stations D and E in specified filter 

sand) (Brugger et al., 2001). 

Figure 4 demonstrates influences on microbial reduction processes both by temperature and flow 

velocity: the oxygen and nitrate of the electron acceptors show the highest peaks during the winter 

season (the oxygen concentration approaches saturation at water temperatures of 2-degrees Celsius 

in February) and drop to low points during the summer season (8.5-mg/L oxygen [O 2] in Station E 

and 4.5-mg/L O 2 in Station F in July, when water temperature reaches 15-degrees Celsius). This 

period of increased oxygen depletion corresponds with concentration peaks of iron and manganese. 

The mean oxygen depletion during the summer season in the riverbed achieves 4.0-mg/L O 2 at 

Bore F and 2.8-mg/L O 2 at Bore D. These respiration rates include a portion for DOC immobilization 

of 25 and 19 percent, respectively (see Figure 3). Hence, both DOC immobilization and 

(non-DOC)-C respiration 

(µ mol C per L) 

101

102 

Figure 4. Courses of oxygen, nitrate, iron, and manganese concentrations at Station E in riverbed zones with 

low hydraulic conductivity and at Station F with higher conductivity during the 1-year monitoring 

period (0.9 to 1.0 m from the sediment-water interface). 

total hyporheic carbon respiration are significantly higher in Infiltration Area F, where the filter 

velocity is 58 percent lower than at Bore D. These results suggest that monitored degradation 

processes are limited by available retention time and that the hydrolyses of particulate organic 

matter is the time-limiting process step. 

The spatial distribution of oxygen, as presented in Figure 5, clearly shows the influence of 

retention time on reduction processes. During the migration time in the first meter of the flowpath 

in natural sediment, the oxygen concentration drops to a minimum level and the subsequent 

recovery is attributed to mixing with oxygen-enriched water. 

Figure 5. Oxygen concentration profiles along the flowpath from the river to the well, 

starting from four different infiltration areas (Stations D and E in natural 

sediment, July 27, 1998). 

In the following section, a calibrated model of the presented site serves as the initial situation (plot 

and cross-section in Figure 6) of a systematic parameter investigation. The linear sensitivity 

analysis uses a function, δ v,p , to quantify the sensitivity of the model response to a unit change in

parameter value: 

Figure 6. Initial conditions of numerical sensitivity analyses (MODFLOW software). 

dv/ v 

δv,p = where p = parameter and v = output variable (model response). 

dp/ p 

The larger the value of the function, the more significant the specific parameter is for model 

behavior. The applicability of this form is limited to linear cause-and-effect relationships or small 

parameter variations. The results in Table 1 outline the major influence of riverbed clogging 

k A/k BF on filtrate production Q BF and well distance d on the total migration time HRT and 

maximum infiltration rate v max. 

Table 1. Comparison of Parameter Sensitivities to Hydraulic Properties of the Considered RBF Site 

d v,p k A / k BF k A Q Prod. d H 

HRT [d] 0.24 0.33 –1.01 1.89 1.20 

Q BF [%] –0.25 –0.15 0.03 –0.18 0.17 

v max [m/h] –0.44 –0.44 0.89 –0.76 –0.89 

HRT = Hydraulic retention time. Q BF = Filtrate portion. v max = Maximum infiltration rate. 

103

104 

Conclusions 

Biologically mediated degradation processes mainly occur at the first meter of the flowpath from 

the river to the well. Hydraulic retention in this layer and infiltration rates, respectively, are 

crucial parameters of additional particulate carbon respiration. Infiltration velocity, in turn, 

depends on the site-specific hydraulic setting, especially on production rate, well distance, and 

aquifer thickness. The appropriate selection of these parameters, therefore, should aim towards 

achieving preferable hyporheic conditions. 


Basic results presented in this paper have been achieved by close interdisciplinary cooperation in the 

course of a research project coordinated (Ch. Hasenleithner) and funded by the Ennskraft Company. 

REFERENCES 

Brettschko, G., and H. Moser (1993). “Transport and retention of matter in riparian ecotones.” Hydrobiologia, 

251: 95-102. 

Brugger, A., B. Wett, I. Kolar, B. Reitner, and G.J. Herndl (2001). “Immobilization and bacterial utilization 

of dissolved organic carbon entering the riparian zone of the alpine Enns River, Austria.” Aquatic Microbial 

Ecology, 24(2): 129-142. 

Ingerle, K., A.P. Blaschke, A. Brugger, C. Hasenleithner, G.J. Herndl, H. Jarosch, I. Kolar, H.J. Lensing, S. 

Pöschl, N. Quéric, B. Reitner, F. Schöller, R. Sommer, and B. Wett (1999). Forschungsprojekt Uferfiltrat 

(Research project bank filtration), Research Initiative Verbund, Vienna, 60, 43-78. 

Lensing, H.J., M. Vogt, and B. Herrling (1994). “Modeling of biologically meditated redox processes in the 

subsurface.” J. of Hydrology, 159: 125-143. 

Quanrud, D.M., J. Hafer, M.M. Karpiscak, J. Zhang, K.E. Lansey, and R.G. Arnold (2003). “Fate of organics 

during soil-aquifer treatment: sustainability of removals in the field.” Wat. Res., 37: 3,401-3,411. 

Wett, B., H. Jarosch, and K. Ingerle (2002). “Flood induced infiltration affecting a bank filtrate well at the 

River Enns, Austria.” J. of Hydrology, 266(3-4): 222-234. 

BERNHARD WETT has been a Scientific Researcher at the Department of Environmental 

Engineering at the University of Innsbruck in Austria for 9 years. His recent scientific 

work has focused on the separate biological treatment of rejection water — a major 

interaction between the wastewater and the sludge lane of a treatment plant. With regard 

to riverbank filtration and its relation to flooding events, he has conducted an 

interdisciplinary project together with various companies and universities over the past 

several years. The central question of these investigations was the vulnerability of 

riverbank-filtration waterworks, in regard to both hydraulic and water-quality aspects. Wett participates in 

working groups of national water associations (Wastewater Technology Association, Austrian Water-and 

Waste Management Association) and in the EU-COST-624 Action. He lectures in various fields of 

environmental engineering, including numerical methods of groundwater flow modeling. Wett received an 

M.S. in Civil Engineering and a Ph.D. in Engineering Science-Environmental Engineering from the 

University of Innsbruck.

Keynote Presentation 

Riverbank Filtration: The European Experience 

Prof. Dr.-Ing. Martin Jekel 






Introduction 

The first historical citation of a small bank filtration system for a village dates back to 1798 in a 

German text entitled, About Water. It describes a construction procedure for artificial bank 

filtration into a dug well, about 2 m from a turbid creek. The intermediate space is filled with sand 

to remove pollution during gravity flow into the well. The sand is replaced twice a year to 

guarantee sufficient yield. Thus, we interpret this treatment as a combination of RBF and slow 

sand filtration, two techniques still widely used and known as effective physical and biological 

filters with many similarities. 

In Europe, more than 130 years of experience exist in the O&M of large bank-filtration schemes. 

The oldest operating RBF waterworks were founded in the 1870s in Germany and The Netherlands 

during the introduction of centralized water supply in urban areas. The most important sites are 

situated along the Danube, Rhine, and Elbe rivers, and the lakes in the City of Berlin area. A large 

variety of schemes has been designed and operated according to site-specific conditions. Many 

research projects have been conducted within the last 20 years to study attenuation processes 

during bank filtration. 

Relevance of Riverbank Filtration in Europe 

Groundwater derived from infiltrating river water provides 45 percent of drinking-water supplies in 

Hungary, 16 percent in Germany, 5 percent in The Netherlands, and 50 percent in the Slovak 

Republic. In Germany, the City of Berlin depends on bank filtration for about 60 percent of its 

drinking water. The City of Düsseldorf, situated on the River Rhine in Germany, is entirely supplied 

with drinking water derived from bank filtration. Budapest, the capital of Hungary, also depends 

100 percent on bank filtration. Waterworks in many other cities (e.g., Cologne, Dresden, Zurich, 

Lindau, and Maribor) rely on bank filtrate as an important resource. In Finland, 217 waterworks use 

bank-filtration techniques as part of their water-treatment process (Kivimäki, 1995). 

Besides known bank-filtration sites, there are many groundwater works where bank filtration is 

not planned and is unintentional, especially within periods of high surface-water levels during 

floods. 


Prof. Dr.-Ing. Martin Jekel 

Professor, Department of Water Quality Control 


Sekr. KF 4 • Strasse des 17. Juni 135 • 10623 Berlin, Germany 

Phone: +(30) 314 - 25058 • Fax: +(30) 314 - 23313 / 23850 • Email: jekel@itu201.ut.TU-Berlin.DE 

105

106 

Geohydraulic Characteristics 

Grischek et al. (2002) provide an overview on the geohydraulic characteristics of different 

bank-filtration sites in Europe. Typical aquifers used for bank filtration consist of alluvial sand and 

gravel deposits and have a hydraulic conductivity higher than 1 × 10 –4 m per second. The 

thickness of the exploited alluvial aquifers covers a wide range. Along the Danube River in the 

Slovak Republic and Elbe River in Germany, aquifer thickness exceeds 50 m. Along the Lot River 

in France and Neckar River in Germany, boreholes are operated in shallow aquifers of about 5-m 

thickness. The distance between production wells and the bank of the river or lake is, in general, 

more than 50 m. Excluding some sites in The Netherlands and along the Danube River in the 

Slovak Republic, travel times are mostly between 20 and 300 days. 

In some countries, a travel time minimum of 50 days is proposed because an accepted rule of 

thumb assumes that about 50 days are sufficient to obtain pathogen-free water. But, recent findings 

underline that other factors, such as surface area (flowpath length), riverbed properties, and redox 

conditions, play an important role in removing pathogens. 

At most sites, vertical wells are in operation. At sites with a high aquifer thickness (for example, 

Kalinkovo in the Slovak Republic), vertical wells have screen lengths of 40 m, allowing high 

abstraction rates. Older systems often include siphon pipes connected to vertical well galleries 

with low abstraction rates (for example, Karany in the Czech Republic and Dresden-Tolkewitz in 

Germany). Horizontal wells are used at sites with high abstraction rates, such as in Düsseldorf, 

Germany, and Budapest, Hungary. 

Riverbeds at RBF sites are normally cut into an underlying sand and gravel aquifer, resulting in 

direct hydraulic contact of the river and aquifer. At many RBF sites, erosive conditions in the river 

limit the formation of a siltation layer. Detailed research to understand the specific mechanisms 

controlling clogging has been undertaken on the Rhine River (Schubert, 2001), Enns River 

(Ingerle et al., 1999), and Elbe River (Heeger, 1987). At most rivers, infiltration occurs in the 

areas between the bank adjacent to the wells and the middle of the river, sometimes over the 

whole width of the river. At the Rhine River, partial clogging of the riverbed limits infiltration 

near the bank adjacent to the wells. The riverbed infiltration zone is fairly clean, but other areas 

are coated with a layer of about 1-millimeter thickness. Sand ripples also develop. Under rapidly 

changing river levels, the riverbed is cleaned (Schubert, 2000). The other case is observed at 

impounded river lakes in Berlin, where infiltration occurs mainly via the bank because thick 

organic mud layers at the bottom of lakes have low hydraulic conductivity. 

Surface-Water Quality 

In general, organic carbon concentrations (TOC) in river waters used for RBF are between 1 and 6 mg/L. 

The lakes of Berlin are higher in TOC, up to 10 mg/L, due to natural fulvic acids, effluent organic 

matter, and algal growth. In Finland, where NOM concentrations contribute to TOC values of 

4 to 15 mg/L, problems are associated with natural humic substances and DBP formation. 

Temperature variations in most rivers range from zero to 25-degrees Celsius, and the pH is between 

6 and 8. 

Since the late 1950s, the water quality of large rivers in Europe began to deteriorate. High 

wastewater input threatened the use of bank filtrate. Furthermore, spectacular spills (for example, 

the Sandoz accident on November 1, 1986 [Sontheimer, 1991]), underlined the need for 

sanitation measures and pollution control. The activities of waterworks and water-industry 

associations, authorities and industries, transborder programs, and the closure of industries all

esulted in a significant improvement of river-water quality (e.g., the Rhine River since 1980 to 

1985 and the Elbe River since 1990). Historic water-quality records show a strong decrease in the 

concentrations of phosphate, polynuclear aromatic hydrocarbon compounds, pesticides such as 

atrazine and mecoprop, and biodegradable organic carbon. 

Raw-Water Quality 

A summary evaluation of elimination processes at bank-filtration sites in The Netherlands showed 

an effective, sustainable, and redox-independent elimination of polycyclic aromatic hydrocarbons, 

polychlorinated biphenyls, chloroorganic pesticides, bromoform, dichloroaniline, nitrobenzene, 

nitrotoluene, and chlornitrobenzene (Stuyfzand, 1998). Anoxic conditions cause an effective 

elimination of pesticides, such as atrazine, diurone, and simazine, which are less (or not) degraded 

under aerobic conditions. 

During bank filtration at the Rhine River, the concentrations of DOC, AOX, and sulphur 

compounds in Rhine water are reduced by more than 50 percent during a mean retention time of 

the infiltrate in the aquifer of 6 to 8 weeks (Brauch and Kuehn, 1988). Concentrations of 

malodorous compounds such as geosmin and monoterpenes decrease through degradation 

processes during bank filtration by 80 to 99 percent (Juettner, 1995). During bank filtration at 

Schlachtensee, Wannsee, and Tegeler See lakes in Berlin, Germany, algae-born terpenes causing 

malodor of lake waters were not found in bank filtrate (Chorus et al., 1992). Complex studies on 

biogeochemical reactions are reported by Bourg and Bertin (1993), Sontheimer (1991), and 

Nestler et al. (1998). 

The long-term monitoring of water quality has demonstrated the effective removal of a range of 

contaminants, including ammonium and polar compounds. Results from monitoring programs at 

sites on the Rhine, Elbe, and Danube rivers and at Tegel Lake have been recently summarized by 

Kuehn and Mueller (2000), Brauch et al. (2000), Grischek (2002), Mucha et al. (2002), and Fritz 

(2002), respectively. 

Drinking-Water Treatment 

In the 1960s and 1970s, one of the major problems was bad odor of drinking water derived from 

bank filtrate. The main ways to solve this problem were a reducing the organic load of the river 

water and using GAC filtration as a post-treatment step (Sontheimer, 1980). 

At some sites, advanced technologies such as ozone treatment, biological filtration, or GAC 

adsorption were established to treat the pumped infiltrate (e.g., along the Rhine River). At other 

sites, simple treatment technologies, such as pH-adjustment, aeration with sand filtration for iron 

and manganese removal, or disinfection by chlorine, are sufficient for meeting today’s drinkingwater 

standards (e.g., along the Elbe River). The necessary treatment steps are still in discussion, 

especially due to recent findings of persistent organic trace compounds in bank filtrate. 

In Germany, different tests have been developed to classify compounds into compound classes that 

are not removed during aquifer passage and not removed by common drinking-water treatment 

(Sontheimer, 1991; Mueller et al., 1993). This concept has been used to identify compounds that 

are problematic in terms of drinking-water quality and to initiate source control measures for 

reducing the input of these mobile substances into receiving rivers. 

107

108 

Problems 

At several sites (e.g., Budapest, Hungary and Dresden, Germany, or at the lower Rhine River), the 

quality of backside groundwater beneath the river floodplain, especially regarding the nitrate 

concentration, is becoming a greater problem than bank-filtrate quality. 

Problems with persistent organic compounds (e.g., pharmaceuticals) are reported for bank 

filtration in Berlin, Germany, where the percentage of surface waters comprised of municipal 

sewage effluents is relatively high (Ziegler et al., 2000, 2002). Recent work has shown that 

pharmaceuticals, with sources including feed additives, animal drugs, and human medical care, 

can be persistent in the aquatic environment. Trace-level analysis shows the presence of drug 

residues such as clofibric acid (a metabolite of a blood lipid regulator in medical care) in municipal 

sewage effluents and in receiving surface waters. As polar compounds, clofibric acid and several 

other pharmaceuticals, such as diclofenac, propyphenazone, carbamazepine, and primidone, are 

highly mobile (Heberer, 2002). Findings of pharmaceutical residues in drinking water are not 

desirable from a hygienic point of view; nevertheless, from today’s knowledge, these low 

concentrations do not have any toxicological impacts on human health. A limited removal of the 

polar drug residues is possible using activated carbon filtration. In summary, polar-peristent 

synthetic organic contaminants such as pharmaceuticals and industrial compounds require greater 

attention to assess the possibility, if any, of attenuation during bank filtration. 

This topic is one of the major issues in a large-scale research program of the new Berlin Centre of 

Competence on Water, called NASRI, which commenced in May 2002. Research activities include 

several field-site investigations with bank filtration and recharge, as well as semi-technical and 

laboratory studies on hydrogeology, modeling, algal toxins, viruses and bacteria, natural and 

anthropogenic organic compounds, and the clogging processes. Results will be presented in this 

survey, as well as in other presentations out of the NASRI-group at this conference. 

Conclusions 

For the future, bank filtration is viewed as a promising approach to increase limited groundwater 

resources and to provide a sustainable pretreatment step with multiple-barrier functions regarding 

chemical and microbial parameters. 

A new period in the exchange of worldwide experience started in November 1999 with the 

Inernational Riverbank Filtration Conference in Louisville, Kentucky (United States), followed by a 

workshop on the Attenuation of Groundwater Pollution by Bank Filtration in June 2000 in Dresden, 

Germany, an International Riverbank Filtration Conference in November 2000 in Düsseldorf, Germany, 

and a NATO Advanced Research Workshop on Riverbank Filtration on Understanding Contaminant 

Biogeochemistry and Pathogen Removal, which was held in September 2001 in Tihany, Hungary. 

REFERENCES 

Bourg, A.C.M., and C. Bertin (1993). “Biogeochemical processes during the infiltration of river water into 

an alluvial aquifer.” Environ. Sci. Technol., 27(4): 661-666. 

Brauch, H.-J., and W. Kuehn (1988). “Organische Spurenstoffe im Rhein und bei der Trinkwasseraufbereitung.” 

gwf Wasser Abwasser, 129(3): 37-44 (in German). 

Brauch, H.-J., U. Mueller, and W. Kuehn (2000). “Experiences with riverbank filtration in Germany.” 

Proceedings, International Riverbank Filtration Conference, IAWR, 4: 33-39. 

Chorus, I., G. Klein, J. Fastner, and W. Rotard (1992). “Off-flavors in surface waters - How efficient is bank 

filtration for their abatement in drinking water?” Wat. Sci. Technol., 25(2): 251-258.

Fritz, B. (2002). Uferfiltration unter verschiedenen wasserwirtschaftlichen, hydrogeologischen und hydraulischen 

Bedingungen. (Investigations of river bank filtration influenced by different hydrogeological, hydraulic and water 

management systems), PhD thesis, Free University of Berlin (in German). 

Grischek, T. (2002). Zur Bewirtschaftung von Uferfiltratfassungen an der Elbe. (Management of bank filtration 

sites along the River Elbe), PhD thesis, Dresden University of Technology (in German). 

Grischek, T., D. Schoenheinz, E. Worch, and K. Hiscock (2002). “Bank filtration in Europe - An overview 

of aquifer conditions and hydraulic controls.” Management of aquifer recharge for sustainability, P. Dillon (ed.), 

Swets & Zeitlinger, Balkema, Lisse, 485-488. 

Heberer, T. (2002). “Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment: 

a review of recent date.” Toxicology Letters, 131: 5-17. 

Heeger, D. (1987). Investigations on clogging of river beds, Unpublished PhD-thesis (in German). 

Ingerle, K., A.P. Blaschke, A. Brugger, C. Hasenleithner, G.J. Herndl, H. Jarosch, I. Kolar, H.J. Lensing, S. 

Pöschl, N. Queric, B. Reitner, F. Schoeller, R. Sommer, and B. Wett (1999). Forschungsprojekt Uferfiltrat 

(Research Project Bank Filtration), Research Initiative Verbund, Vienna 60 (in German). 

Kivimäki, A.L. (1995). “Production of artificially recharged groundwater using bank filtration” Publication of 

the Water and Environment Administration 573, Helsinki, Finland (in Finnish). 

Kuehn, W., and U. Mueller (2000). “Riverbank filtration: An overview.” Journal AWWA, 92, (12): 60-69. 

Mucha, I., D. Rodak, Z. Hlavaty, and L. Bansky (2002). “Groundwater quality processes after bank 

infiltration from the Danube at Cunkovo.” Riverbank Filtration: Understanding Contaminant Biogeochemistry 

and Pathogen Removal, C. Ray (ed.), Kluwer Academic Publishers, The Netherlands, 177-219. 

Mueller, U., B. Wricke, and H. Sontheimer (1993). “Wasserwerks- und trinkwasserrelevante Substanzen in 

der Elbe.” Vom Wasser 81, 371-386 (in German). 

Nestler, W., W. Walther, F. Jacobs, R. Trettin, and K. Freyer (1998). “Water production in alluvial aquifers 

along the River Elbe.” UFZ-Research Report 7, 203 p. (in German). 

Schubert, J. (2001). “How does it work? Field studies on riverbank filtration.” Proceedings, International 

Riverbank Filtration Conference, 2-4 Nov. 2000, Düsseldorf, Germany, IAWR-Rheinthemen 4, 41-55. 

Schubert, J. (2000). “Entfernung von Schwebstoffen und Mikroorganismen sowie Verminderung der 

Mutagenität bei der Uferfiltration (Removal of suspended matter and microorganisms and reduction of 

mutagenity during bank filtration).” gwf Wasser Abwasser, 14(1/4): 218-225 (in German). 

Sontheimer, H. (1980). “Experience with riverbank filtration along the Rhine River.” Journal AWWA, 

72: 386-390. 

Sontheimer, H. (1991). Drinking water from the River Rhine? Academia Verlag, Sankt Augustin (in German). 

Stuyfzand, P.J. (1998). “Fate of pollutants during artificial recharge and bank filtration in the Netherlands.” 

Artificial Recharge of Groundwater, Balkema, Rotterdam, 119-125. 

Ziegler, D., C. Hartig, S. Wischnack, and M. Jekel (2000). “Behaviour of dissolved organic compounds and 

pharmaceuticals during lake bank filtration in Berlin.” Proceedings, International Riverbank Filtration 

Conference, IAWR, 4: 151-160. 

Ziegler, D., C. Hartig, S. Wischnack, and M. Jekel (2002). “Organic substances in partly closed water cycles.” 

Management of aquifer recharge for sustainability, P. Dillon (ed.), Swets & Zeitlinger, Balkema, Lisse, 161-167. 

109

110 

Water chemist and treatment engineer MARTIN JEKEL has about 28 years of experience 

in the field of water and wastewater treatment, as well as in water-quality analysis. Since 

1988, he has been a full Professor for water-quality control at the Institute for 

Environmental Engineering of the Technical University of Berlin. His research interests 

include the development of the “Muelheim Process” for drinking-water treatment, basic 

and applied studies on the preoxidation mechanisms in coagulation of water treatment, 

advanced treatment processes for indirect potable water reuse, a new adsorption technique 

with Granular Ferric Hydroxide for solving the worldwide problem of arsenic in groundwater, and the 

analysis and fate of new organic trace organics, such as iodinated X-ray contrast agents. During the last 

5 years, he has conducted several studies on the processes of organics removal in lake bank filtration in the 

Berlin area, and he is Scientific Coordinator of Natural and Artificial Systems for Recharge and Infiltration, 

one of the largest bank-filtration and recharge research programs worldwide. Jekel received a diploma 

(M.Sc.) in Chemistry, a Ph.D. in Chemical Engineering, and the German degree of Habilitation 

(qualification for a tenure professorship) at the University of Karlsruhe, Germany.

Session 6: Microorganisms 

Using Microscopic Particulate Analysis 

for Riverbank Filtration 

Jennifer L. Clancy, Ph.D. 

Clancy Environmental Consultants, Inc. 

St. Albans, Vermont 

William D. Gollnitz 


Cincinnati, Ohio 

Microscopic Particulate Analysis has been used since the mid-1980s as a tool to examine the 

nature of biological particulates in water. This methodology is used for two purposes: 

• To examine groundwater for the presence of biological surface-water indicators. These 

species may indicate that the groundwater source is GWUDI and, hence, subject to the 

requirements of the Surface Water Treatment Rule. 

• To determine filtration efficiency by comparing the types of particulates in raw and 

finished waters. The log removal of particulates is calculated to determine the efficiency 

of the filtration process. Both full-scale and pilot processes can be evaluated using 

Microscopic Particulate Analysis. 

Microscopic Particulate Analysis samples are collected by filtering a relatively large volume of 

water through a cartridge filter. The sample is generally several hundred gallons, but volume can 

be adjusted depending on the type of sample. The filter is shipped overnight on ice to the laboratory. 

The filter is cut apart and the fibers washed or backwashed directly (Envirochek capsule) to 

remove particulate matter. The particulates are concentrated into a pellet, a portion of which may 

be subjected to buoyant density gradient centrifugation to separate biological particulates from 

heavier debris. This material is examined in two ways: 

• Direct microscopic observation using brightfield and phase or differential interference 

contrast to identify, enumerate, and classify biological particulates. 

• A portion of the sample may be stained using fluorescent monoclonal antibodies specific 

to Giardia cysts and Cryptosporidium oocysts. The sample is examined using epifluorescence 

microscopy to detect the presence of these parasites. 

The data are reported as the numbers and types of biological particulates. Categories include, but 

are not limited to, algae, diatoms, rotifers, crustaceans, insects, protozoa, plant debris, Giardia cysts, 

and Cryptosporidium oocysts. The log reduction of each bio-indicator category is determined by 

comparing the numbers of particulates in raw and finished water samples. A filtration performance 

rating is determined, and the significance of any additional data or observations is explained. 

Other information that can be obtained includes the presence of alum/polymer floc, carbon fines, 


Jennifer L. Clancy, Ph.D. 

President 

Clancy Environmental Consultants, Inc. 

P.O. Box 314 • St. Albans, Vermont 05478 USA 

Phone: (802) 527-2460 • Fax: (802) 524-3909 • Email: jclancy@clancyenv.com 

111

112 

and other material in the finished water. The overall microscopic quality of the water can be readily 

assessed using Microscopic Particulate Analysis. If a surface-water sample has been analyzed 

simultaneously, a comparison of particulates in surface-water and groundwater samples can be made. 

The rationale in examining groundwater using Microscopic Particulate Analysis is that if 

surface-water indicators are observed, then the source may be at risk for contamination by Giardia 

cysts and Cryptosporidium oocysts. Groundwaters that are at risk of surface-water influence are 

considered to be surface waters under the Surface Water Treatment Rule and are required to 

comply with filtration and disinfection requirements. An obvious extension of this technique is to 

examine water samples in systems using RBF. Similar to conducting a filtration plant efficiency 

study, a river or source sample can be compared to samples from the groundwater collection 

device, and a determination of the level of in situ natural filtration occurring through the aquifer 

can be made. This idea of natural filtration to improve water quality is not new. The concept of 

using the aquifer as a natural filtration device was termed “RBF” over 100 years ago, although the 

process has been used for centuries. RBF is a standard surface-water treatment practice in Europe. 

In 1997, the authors first proposed considering natural filtration when performing GWUDI 

evaluations of collection devices located in porous media aquifers. In a long-term study of 

groundwater collection devices in the North Platte Alluvial Aquifer in Casper Wyoming, the 

authors found that natural filtration through the aquifer produced finished water of superior 

quality to that produced by the water filtration plant, both using the North Platte River as a 

source. Surface-water indicators (algae and diatoms) were found in some of the groundwater 

collection devices, and always at significantly lower concentrations than in surface water. Even 

when algae and diatoms were present, cysts and oocysts were never observed in any of the 

groundwater samples. The reduction of algae and diatoms through the aquifer ranged from 4.2 to 

greater than 5 log, while the removal range in the plant, which consistently met all compliance 

requirements, ranged from 0.3 to 1.5 log. 

The key to providing adequate RBF is the aquifer type. Natural filtration through porous media 

aquifers, particularly those with a matrix in the sand- and gravel-size range, is very effective for 

particulate removal. Natural filtration is similar to engineered filtration in that we expect to see a 

reduction of particulates through the process; the end result is not necessarily the observation of 

no surface-water indicators, but adequate removal so as to minimize the risks to public health. In 

the Surface Water Treatment Rule, this is defined as 2-log removal of Giardia cysts through 

filtration, and in the Long Term 2 Enhanced Surface Water Treatment Rule (LT2ESWTR), 3-log 

Cryptosporidium removal through some combination of processes, including RBF removal credit. 

This paper examines the filtration efficiencies of six surface water treatment plants and six systems 

using in situ natural filtration or RBF. Each system was monitored regularly using Microscopic 

Particulate Analysis for a minimum of 1 to 11 years. Samples were collected in pairs (raw and 

treated surface water or river water and groundwater collection device) and analyzed by a single 

group of analysts, minimizing some of the inherent variability noted between laboratories and 

analysts. Sample sites are located in the United States and Canada, and the selected sites represent 

a broad geographic region (California to Connecticut). Over 1,400 samples were analyzed in this 

comparison. 

As noted in previous studies, surface-water samples showed the largest diversity and concentrations 

of microorganisms. Algae and diatoms occurred in the highest concentrations, ranging from l0 2 to 

greater than 10 9 per 100 gallons. The reduction of indicators through engineered filtration ranged 

from no reduction to greater than 6 log 10. Log 10 reduction of biological indicators varied within 

treatment plants, and poorer filtration performance noted using Microscopic Particulate Analysis

was independent of finished water turbidity. All plants in the study were considered well-operated 

and met the requirements of the Surface Water Treatment Rule. In some cases, the levels of algae 

and diatoms in finished water were actually higher than that observed in raw water. This may be 

due to sampling anomalies, growth of algae in the filters, poor filtration performance, or vagaries 

associated with the Microscopic Particulate Analysis method itself. 

For RBF groundwater collection devices, log 10 reduction of algae and diatoms ranged from about 

4 log 10 to greater than 7 log 10. Reductions across the aquifer through natural filtration were far 

more consistent, and the minimum level of reduction was always greater than that required for 

filtration performance under the Surface Water Treatment Rule (2-log 10 Giardia and 3-log 10 

Cryptosporidium). No Giardia cysts or Cryptosporidium oocysts were noted in any of the RBF 

groundwater collection device samples. 

While all samples met turbidity requirements, the biological water quality measured using 

Microscopic Particulate Analysis was far superior in the RBF groundwater collection device 

samples than that noted in the treatment plant effluents. This same observation has been noted 

when comparing surface-water treatment plant effluents and groundwater samples in previous 

studies. RBF is a more effective process than engineered filtration for removing biological particles 

from surface-water sources. 

JENNIFER CLANCY is a microbiologist with 25 years of experience in environmental 

microbiology, focusing on water. She is currently President of Clancy Environmental 

Consultants, Inc., which provides microbiological testing and consulting services to the 

water and wastewater industries. Clancy has spent years studying the issue of groundwater 

under the direct influence of surface water in systems in the United States and Canada. 

Prior to forming Clancy Environmental Consultants, Inc., in 1994, she was the Director 

of Water Quality at the Erie County Water Authority in Buffalo, New York, and was 

responsible for administration of the Water Quality Department. Clancy received a B.S. in Microbiology 

from Cornell University, an M.S. in Microbiology and Biochemistry from the University of Vermont, a Ph.D. 

in Microbiology and Immunology from McGill University, and an M.S. in Environmental Law from Vermont 

Law School. 

113

114


Transport and Removal of Cryptosporidium Oocysts 

in Subsurface Porous Media 

Menachem Elimelech, Ph.D. 

Yale University 

New Haven, Connecticut 

Garrett Miller 



Zachary Kuznar 



Cryptosporidium parvum contamination is considered one of the most important water-supply 

problems today. Disinfectants commonly used by water treatment plants are ineffective at reducing 

Cryptosporidium parvum risk, and deep-bed filtration represents one of the main barriers to oocyst 

contamination. RBF, which is gaining popularity in the United States, presents an alternative 

method of pretreating water to remove oocysts. Very little research has been done so far to 

understand the mechanisms controlling the transport and filtration behavior of Cryptosporidium in 

saturated porous media under chemical and physical conditions relevant to RBF. The objective of 

this study was to elucidate the role of electrostatic double layer interactions in the attachment and 

transport of Cryptosporidium in flow through saturated porous media. Well-controlled column 

experiments were carried out using ultra-clean quartz sand and Cryptosporidium parvum oocysts. 

Complimentary experiments using a stagnation point flow system have been conducted under 

identical chemical and hydrodynamic conditions. Initial bacterial cell deposition rates are compared 

with column breakthrough curves, and the results are used to highlight the controlling mechanisms 

of Cryptosporidium parvum adhesion and transport. 

MENACHEM ELIMELECH is the Llewellyn West Jones Professor of Environmental 

Engineering and Director of the Environmental Engineering Program at Yale University. 

His research interests center on problems involving physicochemical and colloidal 

processes in engineered and natural systems, and he has worked on the dynamics of colloid 

transport and deposition in porous media, transport and fate of microbial particles (viruses, 

bacteria, and Cryptosporidium) in the subsurface environment, and contaminant removal 

by membrane processes. Elimelech is the author of over 90 journal publications and is the 

principal author of Particle Deposition and Aggregation (1995). Among his recent honors, he was the recipient 

of the Association of Environmental Engineering and Science Professors (AEESP) Outstanding Paper Award 

in 2002 and the AEESP Doctoral Dissertation Award, with his graduate student Eric M.V. Hoek, in 2002. 

He also has served on the Editorial Advisory Boards of Environmental Science & Technology, Environmental 

Engineering Science, Desalination, and Journal of Colloid and Interface Science. Elimelech received both a B.S. in 

Soil and Water Sciences and an M.S. in Environmental Science and Technology from the Hebrew 

University in Jerusalem and a Ph.D. in Environmental Engineering from The Johns Hopkins University. 


Menachem Elimelech, Ph.D. 

Llewellyn West Jones Professor of Environmental Engineering 

Department of Chemical Engineering • Environmental Engineering Program 

Yale University • P.O. Box 208286 • New Haven, Connecticut 06520-8286 USA 

Phone: (203) 432-2789 • Fax: (203) 432-2881 • Email: menachem.elimelech@yale.edu 

115

116


Laboratory and Field Strategies for Assessing 

Pathogen Removal by Riverbank Filtration 

Monica B. Emelko, Ph.D. 

University of Waterloo 

Waterloo, Ontario, Canada 

Mark T. Watling 



Martin M. Côté 



Introduction 

RBF systems can significantly reduce the concentrations of many surface-water pollutants 

(Shubert, 2000; Wang et al., 2000; Kuehn and Mueller, 2000); however, predicting and 

quantifying those reductions is difficult. An understanding of subsurface pathogen transport, 

particularly Cryptosporidium and viruses, is critical for utilities faced with potential GWUDI of 

surface-water wells. In the United States, the LT2ESWTR prescribes Cryptosporidium removal 

credits (0, 0.5, or 1.0 log) based on aquifer grain-size distribution and the distance between the 

well and riverbed (USEPA, 2001). Regulations in Ontario, Canada, require utilities to 

demonstrate in situ filtration performance using particle counts (less than 100 particles greater 

than or equal to 10 micrometers per milliliter) and qualitative assessments of potential temporal 

changes in well effluent particle counts and raw-water quality (Ontario Ministry of the 

Environment, 2001). While these approaches represent an attempt at demonstrating the efficacy 

of RBF as a treatment technology for reducing pathogen concentrations in drinking-water sources, 

it is generally acknowledged that they do not provide adequate assessments of pathogen removal 

by RBF. 

Assessments of pathogen removal by RBF are difficult. Although column studies may approximate 

some RBF performance, they are of limited value because they often fail to adequately represent 

temporal changes in regional groundwater conditions or microbiological activity, geologic 

heterogeneity, and physical scale (filtration length) of the true system. While full-scale 

investigations are possible and have been conducted, they are costly and plagued by analytical 

limitations such as low indigenous pathogen concentrations (that preclude representative 

sampling) and, in the case of Cryptosporidium, unreliable analytical methods with low and highly 

variable recoveries (Nieminski et al., 1995; Clancy et al., 1999). Full-scale investigations are 

further complicated by the use of more readily analyzed, but unproven, surrogates (which is 


Monica B. Emelko, Ph.D. 

Assistant Professor 

Department of Civil Engineering 

University of Waterloo • Waterloo, Ontario N2L 3G1 Canada 

Phone: (519) 888-4567 ext. 2208 • Fax: (519) 888-6197 • Email: mbemelko@uwaterloo.ca 

117

118 

necessitated because of the aforementioned analytical limitations). As a result, the outcomes of 

full-scale performance or surrogate validation studies are difficult to interpret and compare, 

hindering the development of widely accepted protocols for assessing pathogen removal by RBF. 

Objectives 

In this paper, we highlight some recent laboratory and analytical advances that are useful for 

improving assessments of pathogen removal by RBF. The specific objectives of this work include: 

• Investigating the impact of column orientation (and the associated impacts of 

gravitational force) on pathogen removal in laboratory studies investigating RBF 

performance. 

• Quantitatively demonstrating the impact of the actual number of discrete particles 

(e.g., pathogens or surrogates) counted from a sample on the uncertainty of inferences 

drawn from experimental and monitoring data. 

• Proposing pathogen or surrogate count targets so that reasonably certain conclusions can 

be drawn from experimental or monitoring data. 

Methodology 

Column Studies 

It has been suggested that column orientation (horizontal versus vertical versus angled) may be an 

important parameter in mimicking subsurface transport and filtration due to the impacts of 

sedimentation (Harvey, 1997). To investigate the impact of the gravitational force on pathogen 

removal during column studies, four 10-cm filter columns with 6-cm internal diameters were 

packed with sieved aquifer material and were operated at conditions typical of those that may be 

encountered during RBF (flow of ~1 milliliter per minute). Two of the columns were operated in 

a horizontal mode and two of the columns were operated in an upflow vertical mode. Preliminary 

investigations were conducted at water-quality conditions that do not favor pathogen removal 

(i.e., low ionic strength). 

After the columns were flushed with degassed water for several hours, the influent reservoir — 

which fed the four columns simultaneously — was spiked with both Cryptosporidium oocysts and 

oocyst-sized polystyrene microspheres. The influent oocyst and microsphere concentrations were 

determined from influent reservoir samples. Effluent samples were collected at intervals of 

0.25-pore volumes until at least 6-pore volumes of influent had passed through the columns. 

Microsphere removals from samples collected during the passage of 0.3- to 5.2-pore volumes 

through the columns are discussed herein. 

Polycarbonate membranes (25-millimeter, 0.40-micrometer nominal porosity) were used to filter 

the samples. Samples were filtered directly on a manifold with a vacuum of ~5-inches of mercury. 

Microsphere enumeration was performed at 100× and 400× magnification (Nikon Labophot 2A, 

Nikon Canada Inc., Toronto, Ontario, Canada). The filtered sample volumes were selected to 

yield between 10 and 1,000 microspheres per membrane. 

Data Reliability and Interpretation 

The occurrence of waterborne pathogens such as Cryptosporidium is difficult to measure precisely 

because they are present in varied concentrations, often at concentrations so low that detection 

is difficult (Lisle and Rose, 1995). Not surprisingly, current methods for quantifying

Cryptosporidium are unreliable, laborious, and expensive. Analytical recoveries are often low and 

highly variable (Nieminski et al., 1995; Clancy et al., 1999). 

Nahrstedt and Gimbel (1996) proposed a model for the process of estimating the concentration 

of Cryptosporidium oocysts in a body of water; it addresses uncertainty associated with sampling and 

enumeration. The authors used: 

• A Poisson distribution to model true sample counts (representative sampling). 

• A binomial distribution to model the recovered fraction of oocysts (random analytical error). 

• A Beta distribution to describe non-constant analytical recovery. 

The authors provided only limited information on how to apply their model to obtain 

concentration estimates and to quantify the uncertainty in these estimates. 

To obtain both concentration and removal estimates and to quantify the associated uncertainties, 

Emelko (2001) presented a Bayesian analysis of the Nahrstedt and Gimbel (1996) model and 

utilized the Gibbs sampler to provide easy access to a wide variety of estimated quantities 

(e.g., point estimates, probability [confidence] intervals, probability density functions, etc.). In 

brief, this approach results in a joint posterior probability density function for the true 

concentration of pathogens in the water body (c), the probability of pathogen recovery from the 

water sample (p), and the number of pathogens in the water sample (N) given the observed 

pathogen count (X). The uncertainty of recovery is accounted for by the Beta distribution’s shape 

parameters (a and b). Replicate sampling is incorporated, if applicable (n replicate samples). The 

posterior probability density function, after simplification and suppressing constant multipliers, is: 

Df [(c,p,Ni|Xi),i = 1,…, n] ∝ c –1+ Σ N i×e –c+Σ V i×p a–1+Σ X i×(1–p) b–1+Σ(N i –X i ) ×Π 

Since the parameter of interest is the true concentration c, the remaining (nuisance) parameters 

may be integrated out (Box and Tiao, 1973). This operation finds the mathematical expectation 

of c over all values of the nuisance parameters. The posterior distribution given by Equation 1 is 

very difficult, as it stands, to integrate or to use directly to make inferences about c. The Gibbs 

sampler, a Monte Carlo method that produces an indefinitely long Markov chain of vectors of 

parameter values (Geman and Geman, 1984), can handle these problems with relative ease. 

Applicable to all discrete particles, this technique was used in the present investigation; details 

regarding the development of this approach were provided in Emelko (2001). 

The impact of the actual number of pathogens or surrogates counted from a sample on uncertainty 

(confidence interval for pathogen concentration) is demonstrated herein using a concentration of 

one pathogen per 100 liters. Counts between one and 1,000 pathogens (and the associated sample 

volumes that would yield a concentration of one pathogen per 100 liters) were utilized. The recovery 

data and associated Beta parameters (a = 28.12, b = 8.43) used in this assessment correspond to oocyst 

recoveries ranging from 69 to 85 percent and average 77 percent, as reported by Emelko (2001). 

Results 

Column Studies 

n 

i=1 

n 

i=1 

Preliminary investigations of Cryptosporidium oocyst-sized polystyrene microsphere removal by 

horizontal and upflow vertical columns are summarized in a box and whisker plot (Figure 1). This 

figure illustrates that a consistent level of microsphere removal was achieved during the period 

n 

i=1 

n 

i=1 

n 

i=1 

Vi N i 

(N i –N i) 

(1) 

119

120 

Microsphere Removal (log 10 ) 

5 

4 

3 

2 

1 

0 

Vertical #1 Vertical #2 Horizontal #1 Horizontal #2 

Figure 1. Cryptosporidium-sized microsphere removal by 10-centimeter columns operated in upflow vertical 

and horizontal orientations. 

that 0.3- to 5.2-pore volumes of water passed through the columns. In addition, the data suggest 

slightly higher removal of oocyst-sized microspheres by the horizontal columns. Moreover, 

excellent reproducibility between duplicate columns was achieved; this is also evident in the 

breakthrough curves (not shown). It should be noted that some microsphere loss occurred in the 

sample line prior to reaching the columns. The overall microsphere removals are, therefore, 

somewhat high; however, given that the columns received the same influent water (losses 

occurred before the feed lines were split), the relative relationship between the columns should be 

unaffected. Continued experiments will be useful in determining whether these differences in 

microsphere removal as a function of column orientation are statistically significant. 

Data Reliability and Interpretation 

n=8 n=8 n=8 n=8 

Column 

The impact of the actual number of pathogens or surrogates counted from a sample on uncertainty 

(confidence interval for pathogen concentration) was demonstrated using a concentration of 

one pathogen per 100 liters (Figure 2). Figure 2 illustrates that the uncertainty associated with 

pathogen concentration data can be considerably reduced by increasing the number of pathogens 

that are counted in a sample; naturally, this can be achieved by increasing the processed sample 

volume. Similar to the present analysis, Emelko and Reilly (2002) demonstrated the same impact 

on uncertainty when examining confidence intervals for pathogen removal (as opposed to 

concentration) data. To avoid confidence intervals that extend beyond an order of magnitude, 

counts of approximately 10 or more pathogens (preferably, approximately 50 pathogens) are 

required from an individual sample. Only limited improvements in uncertainty are observed at 

counts greater than 50 pathogens per sample. 

It is important to note that the outcomes in Figure 2 are associated with the recovery profile of 

one analytical method. Somewhat different outcomes with respect to the range of the confidence 

intervals would be expected with different analytical methods (and their associated recovery 

profiles); however, the same general trend of increased counts resulting in decreased uncertainty

Pathogen Concentration (pathogens/100 Liters) 

10. 

1. 

0.1 

0.01 

0.001 

Figure 2. Ninety-five percent confidence intervals obtained for pathogen concentrations using the 

Cryptosporidium recovery profile of Emelko (2001). 

would be expected. This type of statistical analysis for already reported data can be somewhat 

difficult as adequate recovery data (actual counts) are often lacking from the reported literature, 

underscoring the need for guidance on how to adequately report pathogen recovery information. 

The analysis in Figure 2 does not address the issues of replicate sampling or pooled data. Emelko 

and Reilly (2002) briefly discussed this issue and demonstrated that the largest improvements 

associated with decreasing data uncertainty were associated with increased total counts, regardless 

of the number of samples from which they were derived. The impact of pathogen recovery profiles 

on this relationship remains to be fully demonstrated. 

Conclusions 

1 2 5 10 50 100 500 1000 

Pathogen Count in Sample 

Several preliminary conclusions have resulted from this work. They include the following: 

• Pathogen removal appears to be slightly higher in horizontal laboratory columns as 

compared to upflow vertical columns. As more data are collected, the statistical significance 

of this outcome must be determined. 

• Column studies that are used to investigate RBF should be conducted using a column 

orientation that is as representative of the site as possible; in most cases, this orientation 

should be at or near horizontal. 

• To minimize the uncertainty associated with pathogen data so that it ranges over less 

than an order of magnitude, total counts of between 10 and 50 pathogens are required. 

This can be achieved via increased sample volume processing or increased replication. 

121

122 

REFERENCES 

Box, G.E., and G.C. Tiao (1973). Bayesian Inference in Statistical Analysis, Addison Wesley Publishing 

Company, Reading, MA. 

Clancy, J.L., Z. Bukhari, R.M. McCuin, Z. Matheson, and C.R. Fricker (1999). “USEPA Method 1622.” 

Journal AWWA, 91(9): 60-68. 

Emelko, M.B. (2001). Removal of Cryptosporidium parvum by granular media filtration, Ph.D. Dissertation, 

University of Waterloo, Waterloo, Ontario, Canada. 

Emelko, M.B., and P.M. Reilly (2002). “Reporting and Regulating Cryptosporidium Concentrations and 

Removals.” Proceedings, AWWA WQTC, AWWA, Denver, CO. 

Geman, S., and D. Geman (1984). “Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration 

of Images.” IEEE Trans. Pattn. Anal. Bach. Intell., 6: 721. 

Harvey, R.W. (1997). “In Situ Laboratory Methods to Study Subsurface Microbial Transport.” Manual of 

Environmental Microbiology, C.J. Hurst, G.R. Knudsen, M.J. McInerney, L.D. Stetzenbach, and M.V. Walter 

(eds.), ASM Press, Washington, D.C., p. 586-589. 

Kuehn, W., and U. Mueller (2000). “Riverbank filtration: An overview.” Journal AWWA, 82(12): 60-69. 

Lisle, J.T., and J.B. Rose (1995). “Cryptosporidium Contamination of Water in the U.S.A. and U.K.: A Mini 

Review.” Jour. Water SRT–Aqua, 44(3): 103. 

Nahrstedt, A., and R. Gimbel (1996). “A Statistical Method for Determining the Reliability of the Analytical 

Results in the Detection of Cryptosporidium and Giardia in Water.” Jour. Water SRT – Aqua, 45(3): 101. 

Nieminski, E.C., F.W. Schaefer, and J.E. Ongerth (1995). “Comparison of two Methods for Detection of 

Giardia Cysts and Cryptosporidium Oocysts in Water.” Appl. and Envir. Microbiol., 61(5): 1714. 

Ontario Ministry of the Environment (MOE) (2001). Terms of Reference, Hydrogeological Study to Examine 

Groundwater Sources Potentially Under Direct Influence of Surface Water, Ontario Ministry of the Environment. 

Shubert, J. (2000). “How does it work? Field Studies on Riverbank Filtration.” Proceedings, International 

Riverbank Filtration Conference, Düsseldorf, 2-4 November 2000, Internationale Arbeitsgemeinschaft der 

Wasserwerke im Rheineinsugsgebiet (IAWR), Düsseldorf. 

USEPA (2001). National Primary Drinking Water Regulations: Long Term 2 Enhanced Surface Water Treatment 

Rule, 40 CFR Parts 9, 141 and 142, United States Environmental Protection Agency. 

Wang, J.Z., R. Song, and S.A. Hubbs (2000). “Particle removal through riverbank filtration process.” 

Proceedings, International Riverbank Filtration Conference, Düsseldorf, 2-4 November 2000, Internationale 

Arbeitsgemeinschaft der Wasserwerke im Rheineinsugsgebiet (IAWR), Düsseldorf. 

MONICA EMELKO has research expertise in microbial pathogen and surrogate transport 

and removal by porous media, developing analytical methods for enumerating 

microorganisms during porous media investigations, and developing statistical tools for 

describing analytical uncertainty. She was awarded the American Water Works 

Association Academic Achievement Award for her doctoral dissertation, and has over 

25 publications on pathogen transport and filtration in porous media systems. Emelko has 

been an advisor in the development of the Long-Term 2 Enhanced Surface Water 

Treatment Rule and serves on several American Water Works Association Research Foundation advisory 

committees for projects focused on characterizing filter effluents and filtration performance; she is also the 

Vice-Chair of the Particulate Contaminants Research Committee for the American Water Works Association. 

At present, she is an Assistant Professor in the Department of Civil Engineering at the University of 

Waterloo in Canada. Emelko received B.S degrees in Chemical Engineering and Environmental Engineering 

from the Massachusetts Institute of Technology, an M.S. in Civil Engineering from the University of 

California, Los Angeles, and a Ph.D. in Civil Engineering from the University of Waterloo.


Fate of Disinfection Byproduct Precursors 

and Microorganisms During Riverbank Filtration 

W. Joshua Weiss 






William P. Ball 



Charles R. O’Melia, Ph.D. 



Kellogg J. Schwab, Ph.D. 

Bloomberg School of Public Health, The Johns Hopkins University 


Binh T. Le 

Bloomberg School of Public Health, The Johns Hopkins University 


Ramon Aboytes, D.V.M., Ph.D. 

American Water, Belleville Laboratory 

Belleville, Illinois 

Experience with RBF in Europe and more recently in the United States has demonstrated 

significant improvements in raw-water quality, including the removal of NOM, biodegradable 

compounds, pesticides, microbes, and other water-quality contaminants and compensation for 

shock loads of chemical contaminants (Kuehn and Mueller, 2000; Ray et al., 2002a and 2002b; 

Tufenkji et al., 2002; Weiss et al., 2003a and 2003b). Because of these potential improvements, 

regulators and utilities in the United States have recently looked more strongly at RBF as a means 

for providing high-quality sources for drinking water; however, little data are available to compare 

the performance of RBF with that of conventional drinking-water treatment processes more 

commonly used in the United States (e.g., coagulation, flocculation, sedimentation) from identical 

river-water sources, especially with regard to the removal of organic DBP precursor material. In 

addition, little is known about the extent to which RBF may serve to reliably remove Giardia, 

Cryptosporidium, and other pathogens (e.g., bacteria, viruses) from river water. In particular, data 

are needed on the transport of microbial pathogens through riverbank systems relative to that of 

more easily measured indicator parameters, such as particles and coliform bacteria. 


W. Joshua Weiss 

Ph.D. Student/Research Assistant 

Department of Geography and Environmental Engineering 

The Johns Hopkins University • 3400 N. Charles Street • Baltimore, Maryland 21218 USA 

Phone: (410) 516-6220 • Fax: (410) 516-8996 • Email: jweiss@jhu.edu 

123

124 

In the above context, research was conducted to document water-quality benefits during RBF at 

three major river sources in the Midwestern United States (the Ohio River at Jeffersonville, 

Indiana; Wabash River at Terre Haute, Indiana; and Missouri River at Parkville, Missouri), 

specifically with regard to reducing DBP precursor organic matter and microbial pathogens. 

Specific objectives were to: 

1. Evaluate the merits of RBF for removing organic DBP precursor material. 

2. Evaluate whether RBF can improve finished drinking-water quality by removing and/or 

altering NOM in a manner that is not otherwise accomplished through conventional 

processes of drinking-water treatment (e.g., coagulation, flocculation, sedimentation). 

3. Evaluate changes in the character of NOM upon ground passage from the river to wells. 

4. Evaluate the merits of RBF for removing pathogenic microorganisms. 

5. Compare the transport of microbial pathogens with that of some potential surrogate or 

indicator parameters (e.g., particles, turbidity, coliforms, aerobic and anaerobic spores, 

diatoms, bacteriophage). 

To address Objectives 1, 2, and 3, samples of river source waters and bank-filtered well waters from 

the three study sites were analyzed for a range of water-quality parameters, including: 

• TOC. 

• DOC. 

• UV-absorbance at 254-nm (UV254). • Biodegradable dissolved organic carbon. 

• Biological AOC. 

• Inorganic species. 

• DBP formation potential. 

In the second year of the project, river waters were subjected to a bench-scale conventional 

treatment train consisting of coagulation, flocculation, sedimentation, glass-fiber filtration, and 

ozonation. The treated river waters were compared with the bank-filtered waters in terms of TOC, 

DOC, UV 254, and DBP formation potential. In the third and fourth years of the project, NOM 

from the river and well waters was characterized using the XAD-8 resin adsorption fractionation 

method (Leenheer, 1981; Thurman and Malcolm, 1981). XAD-8 adsorbing (hydrophobic) and 

non-adsorbing (hydrophilic) fractions of the river and well waters were compared with respect to 

DOC, UV 254, and DBP formation potential to determine whether RBF alters the character of the 

source-water NOM upon ground passage and, if so, which fractions are preferentially removed. 

The ongoing research to address Objectives 4 and 5 consists of: 

• Field studies at the three study sites to document actual changes in microorganism 

concentrations upon subsurface travel between the rivers and wells. 

• Parallel laboratory column studies with riverbank aquifer media to provide insights into 

process mechanisms so that reliable treatment credits for pathogen removal can be 

established and the suitability of using indicator parameters for pathogens that are 

difficult to measure in these systems can be determined. 

The results of the DBP-precursor study demonstrate the effectiveness of RBF at removing organic 

precursors to potentially carcinogenic DBPs. When compared to a bench-scale conventional

treatment train optimized for turbidity removal, RBF performed as well as treatment at one of the 

sites and substantially better than treatment at the other two sites in terms of removing organic 

carbon and DBP-precursor material. Removals of TOC and DOC upon RBF at the three sites 

generally ranged from 30 to 70 percent, compared to 20- to 50-percent removals upon bench-scale 

treatment of river waters. Reductions in precursor material for a variety of DBPs (THMs, HAAs, 

haloacetonitriles, haloketones, chloral hydrate, and chloropicrin) upon RBF ranged from 

50 to 100 percent using both the formation potential (FP) and uniform formation conditions 

(UFC) tests (Standard Methods, 1998; Summers et al., 1996), while reductions upon bench-scale 

treatment were generally in the range of 40 to 80 percent. The substantially higher reductions of 

the DBP precursors relative to those of TOC and DOC indicate a preferential reduction upon 

ground passage in the NOM that reacts with chlorine to form DBPs. 

Upon both bench-scale conventional treatment and RBF, a shift was observed in DBP formation 

from the chlorinated to the more brominated species due to the removal of DOC relative to 

bromide upon treatment or RBF. Brominated THMs have greater toxicity than chloroform, so the 

shift from chlorinated to brominated DBP species means that the reduction in risk is not directly 

proportional to the reduction in DBP formation. Risk calculations demonstrated the ability of 

RBF in all cases to reduce the theoretical excess cancer risk due to THMs formed upon 

chlorination, and with substantially better performance than the bench-scale treatment train. 

The results of the NOM characterization study indicate that RBF appears to be equally capable of 

removing material of different character. The different removal mechanisms in the subsurface 

(e.g., sorption, biodegradation, filtration) combine to provide similar removal of the operationally 

defined hydrophilic and hydrophobic fractions of organic material upon ground passage. Thus, the 

observed reductions in DBP formation upon RBF were largely the result of a decrease in the NOM 

concentration rather than a consistent change in NOM character. 

Field monitoring of a number of microorganisms on a tri-monthly basis between 1999 and 2000 

(Table 1) indicated that RBF may also serve as a significant barrier for removing microbial 

contaminants, including human pathogens. The monitoring data demonstrated: 

• Greater than 3-log removal of Clostridium spores. 

• Greater than 2.5-log removal of E. coli C bacteriophage (somatic phage host). 

• Greater than1.9-log removal of E. coli Famp bacteriophage (male-specific host). 

Log removals were calculated using “average” concentrations determined for each organism as the 

sum of the counts over all sampling rounds divided by the sum of the volumes sampled over all 

sampling rounds (Parkhurst and Stern, 1998); non-detects were treated as being zero (in the event 

of no detects over all sampling rounds, the average concentration is reported as being less than 

one count divided by the sum of the volumes sampled over all sampling rounds). Assuming that 

these indicator organisms can be used as surrogates for Giardia cysts and human enteric viruses, 

RBF at the three study sites met or surpassed the performance requirements in the United States 

for conventional coagulation, sedimentation, and filtration (i.e., 2.5-log removal for Giardia cysts 

and 2.0-log removal of viruses). More recent monthly field-monitoring results (Table 2) indicate 

greater than 0.8-log reduction of Bacillus, greater than 3-log reduction of Clostridium, greater than 

1.8-log reduction of bacteriophage MS-2, and greater than 3.7-log reduction of bacteriophage 

φX174 concentrations in bank-filtered waters relative to river waters. Cryptosporidium and Giardia 

were occasionally detected in river waters and were below the detection limits during all monthly 

sampling events in well waters. 

125

126 

Table 1. Results from 1999 to 2000 Tri-Monthly Field Monitoring 

(Seven Total Sampling Rounds) a,b 

Clostridium E. coli C 

Bacteriophage 

c E. coli F-ampc Average counts/ 

100 mL 

Indiana American Water at Jeffersonville, Indiana 

pfu /100 mL pfu /100 mL 

Ohio River 122 49 12 

Well 9 3.2] 2.8] 0.09 [2.1] 

Well 2 3.2] 2.8] 2.2] 

Indiana American Water at Terre Haute, Indiana 

Wabash River 183 147 13 

Collector Well 0.07 [3.4] 3.3] 2.3] 

Well 3 3.4] 3.3] 2.3] 

Missouri American Water at Parkville, Missouri 

Missouri River 143 31 6 

Well 4 3.3] 2.6] 1.9] 

Well 5 3.3] 2.6] 1.9] 

a 

Concentrations calculated as Σ (counts for all sampling rounds)/Σ (volume sampled for all sampling rounds); see text. 

b 

Log removals shown in brackets. 

c 

E. coli C and E. coli F-amp are the host organisms. 

pfu = Plaque-forming unit. 

Because of low and variable concentrations of pathogens such as Cryptosporidium in river waters, 

it is difficult to evaluate log removals for such parameters. Laboratory-scale column studies with 

riverbank media are currently underway to compare the behavior of pathogens (viruses Poliovirus 

and Feline Calicivirus; bacteria E. coli; and protozoans Giardia and Cryptosporidium) with that of 

a number of potential surrogate parameters. The potential surrogate or indicator parameters being 

studied include: 

• Bacteriophage MS-2 and PRD-1 (for human viruses). 

• Particles (latex and native river particles). 

• Turbidity. 

• Aerobic and anaerobic spores. 

Column studies are being conducted under a variety of physical/chemical conditions, with pH, 

ionic strength, multi-valent cation concentration (Ca 2+ ), NOM concentration, and flow rate as the 

variables, with laboratory-prepared water and river water collected from full-scale RBF systems. 

Column experiments indicate that bacteriophage can travel through riverbank sediment columns 

to a greater extent than Poliovirus, suggesting that bacteriophage may be useful as conservative 

indicators of human viruses. Continuing studies will further explore this relationship under a 

variety of physical/chemical conditions. Similar studies will explore the relationship between the 

transport of particle counts, turbidity, and aerobic and anaerobic spores with that of E.coli, 

Cryptosporidium, and Giardia.


We gratefully acknowledge the support of the USEPA Office of Research and Development 

(Project CR-826337; Thomas F. Speth, Project Officer), USEPA Science to Achieve Results 

(STAR) program (Grant #R82901101-0; Angela Page, Project Officer), and the utility 

subsidiaries of American Water. 

REFERENCES 

Table 2. Results from January to December 2002 Monthly Field Monitoring 

(12 Total Sampling Rounds) a,b 

Bacillus Clostridium Total E. coli Bacterio- Bacterio- Crypto- Giardia 

(cfu/L) (cfu/L) Coliforms (MPN/L) phage phage sporidium (cysts/L) 

(MPN/L) MS-2 PhiX174 (oocysts/L) 

(pfu/L) (pfu/L) 

Indiana-American Water Company at Jeffersonville, Indiana 

Ohio River 8.9 × 10 4 3.8 × 10 2 9.2 × 10 4 2.6 × 10 3 6 × 10 1 1.7 × 10 3 2.5 × 10 –2 3 × 10 –2 

Well 9 2.5 × 103 0.5] 

Well 2 1.3 × 104

128 

Ray, C., G. Melin, and R.B. Linsky, editors (2002b). Riverbank Filtration: Improving Source-Water Quality, 


Standard Methods for the Examination of Water and Wastewater, Twentieth Edition (1998). APHA, AWWA, 

& WEF, Washington. 

Summers, R.S., S.M. Hooper, H.M. Shukairy, G. Solarik, and D.M. Owen (1996). “Assessing DBP Yield: 

Uniform Formation Conditions.” Journal AWWA, 88(6): 80. 

Thurman, E.M., and R.L. Malcolm (1981). “Preparative Isolation of Aquatic Humic Substances.” Envir. Sci. 

& Technol., 15(4): 463. 

Tufenkji, N., J.N. Ryan, and M. Elimelech (2002). “The Promise of Bank Filtration.” Envir. Sci. & Technol., 

36(21): 423A. 

Weiss, W.J., E.J. Bouwer, W.P. Ball, C.R. O’Melia, M.W. LeChevallier, H. Arora, and T.F. Speth (2003a). 

“Riverbank Filtration: Fate of Disinfection By-product Precursors and Selected Microorganisms.” Journal 

AWWA, in publication (expected October, 2003). 

Weiss, W.J., E.J. Bouwer, W.P. Ball, C.R. O’Melia, H. Arora, and T.F. Speth (2003b). “Riverbank Filtration: 

Comparison with Bench-Scale Conventional Treatment for NOM and DBP Precursor Reductions.” Journal 

AWWA, in publication (expected December, 2003). 

JOSH WEISS is a Research Assistant in the Department of Geography and Environmental 

Engineering at The Johns Hopkins University, where he is currently investigating waterquality 

improvements during riverbank filtration at three Midwestern drinking-water 

utilities. He expects to receive a Ph.D. in 2004 after the completion of his thesis, 

“Reduction in Disinfection Byproduct Precursors and Microorganisms During Riverbank 

Filtration.” Weiss is the co-author of seven publications — including a chapter in Riverbank 

Filtration: Improving Source-Water Quality (2002) — and several conference proceedings, 

and was the 2000 recipient of the Chesapeake Section American Water Works Association Student Paper 

Award. Prior to attending Johns Hopkins, he researched the geochemical impact of proposed developments 

on a Florida barrier island for the Georgia Institute of Technology and investigated the bioremediation of sites 

contaminated by nitroaromatic compounds for Argonne National Laboratory. Weiss received a B.S. in Civil 

Engineering from the Georgia Institute of Technology and a M.S. in Environmental Engineering from The 

Johns Hopkins University.


Assessment of the Microbial Removal Capabilities 

of Riverbank Filtration 

Vasiliki Partinoudi 

New England Water Treatment Technology Assistance Center 

Department of Civil Engineering, University of New Hampshire 

Durham, New Hampshire 

M. Robin Collins, Ph.D., P.E. 




Aaron B. Margolin, Ph.D. 




Larry K. Brannaka, Ph.D., P.E. 




Introduction 

Riverbank filtrate includes both groundwater and river water that has percolated through the 

riverbank or riverbed to an extraction well. One of the primary objectives of this study was to 

assess the microbial removal capabilities of RBF independent of any groundwater dilution (i.e., a 

worse-case scenario). This study monitored total coliforms, E. coli, and aerobic spore-forming 

bacteria along with other water-quality parameters over a 12-month period in the following 

locations: 

• Pembroke, New Hampshire. 

• Milford, New Hampshire. 

• Jackson, New Hampshire. 

• Louisville, Kentucky. 

• Cedar Rapids, Iowa. 

This study also monitored the removal of male-specific coliphage achieved by RBF in Louisville, 

Kentucky, and Cedar Rapids, Iowa. Samples were collected at both of these sites for the detection 

of enteric viruses. 


Vasiliki Partinoudi 

Graduate Student 

Environmental Technology Building • Department of Civil Engineering • Environmental Research Group 

University of New Hampshire • 25 Colovos Road • Durham, New Hampshire 03824 USA 

Phone: (603) 862-1197 • Fax: (603) 862-3957 • Email: Vp4@cisunix.unh.edu 

129

130 

Objectives 

The overall study objectives were to: 

• Quantify the contributions of river water and groundwater to RBF-extraction water. 

• Assess RBF as a viable treatment and pretreatment option. 

• Compare RBF to slow sand filtration in regards to particulate, organic precursors, and 

microbiological removal capabilities expressed in terms of log-removal credits. 

The focus of this paper will be on microbial removals achieved by RBF. 

Characterization of Sampling Sites 

Five RBF sites (three in New Hampshire, one in Kentucky, and one in Iowa) were chosen that 

possessed different characteristics, in terms of the degree of hydraulic connection between the 

river and RBF extraction well, pumping rates, well construction criteria, geology, water source and 

quality, and distance of the RBF extraction well from the river. The sites were chosen to enable 

the researchers to assess the abilities of RBF technology in different settings. The study sites in 

Louisville, Kentucky, and Cedar Rapids, Iowa, were chosen because these particular sites have 

been characterized in detail in previous studies. 

Pembroke, New Hampshire: The well site lies within a stratified-drift aquifer located within the 

Soucook River Valley. The eastern aquifer boundary is delineated by contact with glacial till. The 

aquifer consists of layers of fine to course sand of varying thickness grading with fine gravel to 

boulders in some locations. In the area of the RBF extraction well, aquifer thickness ranges from 

9.7 to 18.9 m, with a saturated thickness of 7.6 to 17.4 m. The distance between the RBF 

extraction well and river is approximately 55 m. The RBF extraction well has a diameter of 

30.5 centimeters and is 17-m deep. The subsurface material at that depth was dense silt and very 

fine sand. 

Milford, New Hampshire: The Milford State Fish Hatchery is located along the Souhegan River 

near its confluence with Purgatory Brook. The Souhegan River Valley contains rich deposits of 

glacial sand, gravel, and silt. The well is a 61-centimeter diameter gravel-pack well, constructed 

to a depth of 19.8 m. The screen is 50.8 centimeters in diameter and the length is 4.6 m. The 

aquifer at the well site consists mainly of sand and gravel. The distance between the river and the 

well is 23 m. The well is pumped continuously and delivers 1.6 MGD. 

Jackson, New Hampshire: The Jackson Water Precinct infiltration gallery was constructed in the 

early 1980s and is located on the banks of the Ellis River. Five infiltration galleries are connected 

to the river to provide adequate flow to the RBF extraction well. The five infiltration gallery 

intakes are each 6.1-m long, 1.2-m deep, and 1.2-m wide, and are located underneath the riverbed. 

They were placed 1.2-m apart, thus covering a total area of 10.8 m. A 12.2-m long polyvinyl 

chloride pipe connects each gallery to an intake pipe leading to the 20.3-centimeter diameter, 

7.3-m deep RBF extraction well equipped with a submersible pump. A new infiltration gallery is 

now under construction. 

Louisville, Kentucky: The construction of the collector well was completed in March 1999. The 

Louisville Water Company has extensively characterized this site. The RBF extraction well is on 

the bank of the Ohio River, about 30.5 m from the river. The depth of the well is 12.2-m below 

the river bottom and the pumping rate is 75.7-million liters per day. The well has seven horizontal 

laterals, which are each 73.1-m long and constructed of 30.5-centimeter stainless steel, with well

screens along the entire lateral length. Of the seven laterals, L4 is the lateral underneath the river 

separated by at least 40 ft of aquifer material; L1 is the furthest (perpendicular) to the river; and 

L2 is parallel to the river. 

Cedar Rapids, Iowa: The Cedar River is a meandering steam that has cut into the bedrock 

surface, exposing steep valley walls in the study area (Schulmeyer, 1995), where a total of 

53 vertical and horizontal wells have been installed. Seminole Valley Park Well Number 22 is 

located on the banks of the Cedar River, and is 19.5 m from the crest of the riverbank. The RBF 

extraction well was drilled in 1991 to a total depth of 17.4-m below grade. The borehole diameter 

was 10.7 centimeters to depth, and a 7.6-centimeter screen and casing were installed. 

Microbial Analyses 

Total Coliforms and E. coli: The IDEXX Quanti-Tray/2000 method was used for this analysis. 

Aerobic Spore-Forming Bacteria: Procedures were followed as outlined by Ballester and Margolin (2000). 

Virus Indicators: Male-specific and somatic coliphage viruses were monitored intensively for 

2 weeks in December 2002 at Louisville, Kentucky, and Cedar Rapids, Iowa, using a single agar 

overlay (Method 1602, 1999) and a two-step enrichment method. Antibiotic-resistant strains 

E. coli F-amp (resistant to Streptomycin and Ampicillin) and E .coli CN-13 (resistant to Nalidixic 

Acid) were used as hosts for F-specific coliphage and somatic coliphage, respectively. Analyses 

followed the method as outlined by Ballester and Margolin (2000). 

Viruses: The intensive coliphage monitoring was followed by the collection of samples to detect 

human enteric viruses (Adenovirus type 40 and 41, Astrovirus, Poliovirus, Coxsackie virus, 

Rotavirus, and Echovirus) using positive microwound filters. The virus samples were analyzed 

using the Integrated Cell Culture-Reverse Transcription-Nested Polymerase Chain Reaction 

(ICC-RT-nPCR) method, due to its high specificity and sensitivity (Chapron et al., 2000). 

Problems with Assessing RBF Removal Capabilities 

It is essential to understand the extent and magnitude of river-aquifer interaction to address waterquality 

and supply issues, as well as to ensure the health of ecosystems (Winter et al., 1998). There 

are many questions to be answered as to how RBF works, how to establish travel time, and how to 

assess the mixing ratio of river water to groundwater in an RBF extraction well. These questions 

need to be answered to determine the removal efficiency of the RBF process independently of 

groundwater dilution. 

What Is the Exact Travel Time from the River to the Well? 

Table 1 provides a summary of travel time determinations for river water to reach the RBF 

extraction well and the selected method used for each site in this study. 

How Much Removal Is Due to Filtration and How Much Is Due to Dilution with Groundwater? 

A variety of water-quality parameters (Table 2) were analyzed to assess the contribution of river 

and groundwater to RBF-extracted water. Knowledge of the mixing ratio would be necessary to 

determine removals achieved solely by subsurface filtration. The degree of mixing depends on 

factors such as hydraulic head gradient in the aquifer, aquifer properties, and hydrostatic pressure 

in the river, as well as the degree of connectivity between the river and parts of the aquifer. 

131

132 

The contribution of river water and groundwater to each of the RBF extraction wells was 

evaluated based on a different parameter at each site due to different types of information available 

at each site. The average source contributions to the RBF-extracted water during this study period 

is summarized in Table 3. 

Results and Discussion 

Table 1. Travel Times from the River to the RBF Extraction Well 

Sampling Site Travel Time Evaluation of Travel Time 

(from river 

to RBF well) 

Pembroke, 5 days Darcy’s Law in terms of seepage velocity 

New Hampshire 

Milford, 1 day Darcy’s Law in terms of seepage velocity 

New Hampshire 

Jackson, 1 day Infiltration gallery. Estimated from information based 

New Hampshire on the construction details of the gallery 

Louisville, 1 day Estimated from information provided by the 

Kentucky Louisville Water Company (initial pumping test data) 

Cedar Rapids, 5 days Estimated from information provided by the 

Iowa City of Cedar Rapids (Schulmeyer, 1995) 

Table 2. Sampling Parameters Used to Assess the Mixing Ratio in the RBF Extraction Well 

Sulfate Color Hardness 

(true) 

Chloride UV 254 absorbance Alkalinity 

pH Particle counts Redox potential 

Temperature Selected radioisotopes (222Ra) Conductivity 

RBF reduces DOC by as much as 82 percent and reduces temperature spikes by more than 

46 percent in the summer and 96 to 97 percent in the winter months. Turbidity was reduced to 

well below 1 ntu (80- to 86-percent removal). The reductions were computed from field values 

presented in Table 4. 

Typical river-water concentrations ranged between: 

• Below detection limit to 24,192 colony-forming units (cfu)/100 milliliters for total coliforms. 

• Below detection limit to 1,031 cfu/100 milliliters for E. coli. 

• Eighty-four to 145,000 cfu/100 milliliters for aerobic spore-forming bacteria (see Table 4). 

All three of these microbial concentrations were below detection limit (less than 1 cfu/100 milliliters) 

in RBF-extraction well water, even after eliminating the “dilution” effects with groundwater.

Table 3. Average Distribution of River Water and Groundwater to RBF-Extracted Water 

Percent of Percent of Parameter 

River Water Groundwater Upon Which 

in RBF Wells in RBF Wells the Ratio Is Based 

Pembroke, 40.7 ± 3.7 59.3 ± 3.7 Conductivity 

New Hampshire 

Milford, 40.8 ± 6.4 59.2 ± 6.4 Sulfate 

New Hampshire 

Jackson, 100 0 Infiltration gallery receiving only 

New Hampshire river water (due to an impermeable 

barrier that prevents groundwater 

from entering the well) 

Louisville, 78.1 ± 4.4 21.9 ± 4.4 Hardness 

Kentucky 

Cedar Rapids, 70 30 Based on groundwater flow modeling 

Iowa 

Parameter/ 

Site 

Pembroke, 

New Hampshire 

Table 4. Typical Ranges of Selected Analytes of Interest 

Milford, 

New Hampshire 

Jackson, 

New Hampshire 

Louisville, 

Kentucky 


Date 8/01 to 11/03 11/01 to 11/02 5/02 to 11/02 9/01 to 12/02 9/02 to 1/03 

River RBF GW River RBF GW River RBF River RBF GW River RBF GW 

pH 6.3–7.3 6.1–6.6 5.6–6.2 6.3–7.6 5.8–6.6 5.9–6.7 6.3–7.3 6.0–7.5 NA NA NA 8.2–8.5 7.6–7.7 7.3–7.4 

DOC (mg/L) 1.7–6.6 0.3–1.2 0.3–0.7 1.6–6.4 0.5–1.3 0.3-0.8 2.4–2.7 1.7–1.8 3.0–5.0 1.7–2.0 1.3–1.8 2.8–4.3 1.5–2.5 0.1–0.3 

Temperature (°C) 0.3–25.7 8.5–13.8 8.3–11.8 0.4–26.1 7.3–16.2 6.7-15.3 10–21.1 9.5–11 3.6–29.7 13.3–26.415.5–16.6 0.4–20.6 8–22.9 9.4–13.1 

Conductivity 42–150 193–269 294–395 119–214 107–152 113-194 23–83 52–69 NA NA NA 454–666 473–624 468–658 

(µS/cm) 

Total 141– BDL BDL 423– BDL BDL 52–1,356 16–160 10– BDL BDL 75– BDL BDL 

Coliforms 1,399 2,431 24,192 2,572 

(cfu/100 mL) 

E. coli 4–108 BDL BDL 6–119 BDL BDL 2.0–20 2.0–10 2–1,031 BDL BDL 23–56 BDL BDL 

(cfu/100 mL) 

Aerobic 84– BDL BDL 124-1,093 BDL BDL BDL BDL 3,500– BDL BDL 39–217 BDL BDL 

Spore-Forming 1,997 145,000 

Bacteria 

(cfu/100 mL) 

NA = Not available. µS/cm = MicroSemens per centimeter. GW = Groundwater. BDL = Below detection limit. 

Typical concentrations of aerobic spore-forming bacteria, total coliforms, E. coli, and male-specific 

coliphage in river water and RBF-extracted water can be seen in Table 5. The total removal is 

calculated by subtracting the concentration of an analyte in the river from that in the RBF 

extraction well. Total coliforms and E. coli were reduced on average by at least 2.6- and 0.8-log removal 

credits, respectively. These values were considered artificially low due to low numbers observed in 

river water. RBF achieved a 4.4-log credit reduction of total coliforms on May 20, 2002, in 

133

134 

Louisville, Kentucky. Aerobic spore-forming bacteria were removed by at least 1.9-log credit in 

Pembroke, New Hampshire, and up to 5.2-log credits in Louisville, Kentucky. These values are in 

line with those reported for Louisville, Kentucky, by Wang et al. (2002). 

The male-specific coliphages ranged between 3,453 and 4,622 plaque-forming units (pfu)/100 mL 

in river water. The concentration of the male-specific coliphages was reduced at least 80 percent 

by riverbank passage at all the study sites, as is indicated in the reduction summary presented in 

Table 5. Although there was a high concentration of male-specific coliphages in the river water 

and RBF-extracted water, the virus samples collected in Cedar Rapids in December 2003 and 

Louisville in March 2003 were negative (ICC-nPCR method) for the viruses of interest. 

Conclusions 

At each study site, the water quality of the source water, aquifer material, distance from the river 

to the RBF extraction well, travel time, and mixing ratio between river and groundwater were 

evaluated and found to differ from site to site. Travel times and mixing ratios were evaluated by 

different methods based on the information available at each site. RBF was found to be a sitespecific 

process and should be treated as such. Based on differences between sites in source-water 

quality, aquifer material, and other parameters, it will be difficult to develop guidelines applicable 

to all RBF sites. 

In summary, the sites evaluated in this study indicated the conservative effectiveness of RBF in 

removing bacteria and virus indicators (any groundwater dilution with RBF extract should 

contribute to even lower microbial concentrations). Aerobic spore-forming bacteria, total 

coliforms, E. coli, and male-specific bacteriophage were removed by at least an average of 1.9, 1.0, 

0.3 and 0.2 logs, respectively. Based on this study, RBF shows potential to be a viable pretreatment 

and treatment process and warrants additional study. 


• USEPA for funding this project through the New England Water Treatment Technology 

Assistance Center at the University of New Hampshire. 

• Jackson, New Hampshire Waterworks. 

• Nicola A. Ballester and Justin H. Fontaine of the University of New Hampshire. 

• Milford, New Hampshire Fish Hatchery personnel. 

• Cedar Rapids Water Department, Iowa. 

• Pembroke, New Hampshire Waterworks. 

• Louisville Water Company, Kentucky. 

• Melisa A. Smith of the University of New Hampshire.

Location Number 

of 

Sampling 

Events [1] 

Table 5. Average Reductions Achieved by RBF 

River [2] RBF [2] Background 

Well [2] 

Total 

Log 

Removal 

Average 

Percent 

Removal 

Achieved by 

Subsurface 

Filtration 

Average 

Percent 

Removal 

Achieved by 

Groundwater 

Dilution 

Aerobic Spore Forming Bacteria (cfu/100 mL) 

Pembroke, 19 502 BDL BDL >1.9 100% 0% 

New Hampshire ± 11 

Milford, 13 673 BDL BDL >2.1 100% 0% 


Jackson, 

New Hampshire 

3 32 ± 1 BDL NA BDL 100% 0% 

Louisville, 11 33, 404 BDL BDL >3.5 100% 0% 

Kentucky ± 5,672 

Cedar Rapids, 

Iowa 

5 101 ± 8 BDL BDL >2.6 100% 0% 

Total Coliforms (cfu /100 mL) 

Pembroke, 19 521 BDL BDL >2.1 100% 0% 


Milford, 13 1,091 BDL BDL >2.6 100% 0% 


Jackson, 3 504 85 NA >0.5 100% 0% 

New Hampshire ± 42 ± 12 

Louisville, 11 3,921 BDL BDL >1 100% 0% 

Kentucky ± 182 

Cedar Rapids, 5 1,391 BDL BDL >1.4 100% 0% 

Iowa 

E. coli (cfu /100 mL) 

± 25 

Pembroke, 

New Hampshire 

19 27 ± 4 BDL BDL >0.6 100% 0% 

Milford, New 

Hampshire 

13 55 ± 10 BDL BDL >0.8 100% 0% 

Jackson, 

New Hampshire 

3 30 ± 1 6.5 ± 1 NA >0.4 100% 0% 

Louisville, 

Kentucky 

11 5 ± 2 BDL BDL >0.3 100% 0% 

Cedar Rapids, 

Iowa 

5 16 ± 1 BDL BDL >0.7 100% 0% 

Virus Indicators (Male-Specific Coliphage) (pfu/100 mL) 

Louisville, 4 4,342 3,703 3,402 >0.2 80.2% 19.80% 

Kentucky ± 24 ± 22 ± 18 

Cedar Rapids, 5 3,438 753 BDL >0.7 100% 0% 

Iowa ± 21 ± 9 

Where: [1] = One sampling event includes the collection of a river water, groundwater, and RBF-extracted water sample. 

The RBF-extracted water sample was sampled with travel time taken into consideration. 

Where: [2] = This value shows the average concentration of the microorganism of interest throughout the study ± analytical error. 

BDL = Below detection limit. NA = Not available. 

135

136 

REFERENCES 

Ballester, N.A., and A.B. Margolin (2000). Detection of Bacillus Spores in Environmental Samples, University 

of New Hampshire Waterborne Disease Laboratory SOP Manual. 

Chapron, C.D., N.A. Ballester, J.H. Fontaine, C.N. Frades, and A.B. Margolin (2000). “Detection of 

Astroviruses, Enteroviruses, and Adenovirus types 40 and 41 in Surface Waters Collected and Evaluated by 

the Information Collection Rule and an Integrated Cell Culture-Nested PCR Procedure.” Applied and 

Environmental Microbiology, 66(6): 2,520-2,525. 

Method 1602 (1999). Male-specific (F+) and Somatic Coliphages in Water by Single Agar Layer Procedure. 

United States Environmental Protection Agency-821-R-00-00X. Draft: July 1999. 

Schulmeyer P.M. (1995). Effect of the Cedar River on the quality of the groundwater supply for Cedar Rapids, 

Iowa, U.S. Geological Survey Water Resources Investigations Report 94-4211. 

Wang J.Z., S.A. Hubbs, and R. Song (2002). Evaluation of riverbank filtration as a drinking water treatment 

process, American Water Works Association Research Foundation and American Water Works Association. 

Winter T.C., J.W. Harvey, O.L. Franke, and W.M. Alley (1998). Groundwater and surface water – A single 

resource, U.S. Geological Survey Circ. 1139, 79 pp. 

VASO PARTINOUDI is a graduate student at the University of New Hampshire who is 

working on her Master’s thesis, “Riverbank Filtration as a Viable Treatment and 

Pretreatment Process.” She has worked as a Research Assistant at the New England Water 

Treatment Technology Assistance Center (WTTAC) at the University of New Hampshire 

and has participated in WTTAC projects, performing various tasks such as installing and 

maintaining water monitoring equipment and pilot slow sand filters, and sampling and 

performing various water-quality tests in the laboratory and in the field. She has presented 

her thesis-related work at the New England Water Works Association conference in 2002 (paper titled, 

“Riverbank Filtration as a Viable Treatment and Pretreatment Method”) and in 2003 at the American 

Geophysical Union conference in Nice, France (poster presentation titled, “Assessment of the microbial 

removal capabilities of Riverbank Filtration”). Partinoudi has received a B.S. in Civil Engineering from the 

University of Brighton, United Kingdom, and an M.S. in European Construction Engineering from the 

University of Coventry, United Kingdom.

Session 7: Organics Removal 

Riverbank Filtration: A Very Efficient Treatment 

Process for the Removal of Organic Contaminants? 

Dr.-Ing. Heinz-Jürgen Brauch 

DVGW-Technologiezentrum Wasser 

Karlsruhe, Germany 

Dr. rer. nat. Frank Sacher 



Prof. Dr. Wolfgang Kühn 



Introduction 

In Germany, RBF has been used for more than 100 years as a natural treatment process for the 

production of drinking water (Sontheimer, 1980 and 1991; Kühn and Müller, 2000; Sacher and 

Brauch, 2002). Nowadays, the major raw-water resource for drinking-water supplies in Germany 

is groundwater (about 64 percent), whereas bank-filtrated (or infiltrated) water is about 16 percent 

(Sacher and Brauch, 2002; Brauch et al., 2001). Compared to this, the direct abstraction of river 

water is of minor importance (less than 1 percent). In many cases (and mostly along larger rivers), 

a clear distinction between bank-filtrated water and groundwater is difficult, and the raw water 

used for drinking-water production is bank-filtrated water blended with groundwater. 

Bank-filtrated water as a source of raw water provides high-quality river water, and its subsoil passage 

guarantees an efficient and lasting removal of suspended matter, microorganisms, and organic 

micropollutants. Hence, the occurrence and fate of organic contaminants in river water and bankfiltrated 

water are of great concern for water suppliers worldwide using surface water or artificial 

groundwater as a drinking-water resource. Due to the huge number of possible contaminants in river 

water, a necessary restriction has to be made on organic substances that may be relevant to drinkingwater 

production (Sacher and Brauch, 2002 and 1999; Sacher et al., 2001a). These target 

compounds are characterized by criteria like microbial biodegradability, adsorbability onto activated 

carbon and onto soil material, behavior versus oxidation agents, bioaccumulation, and groundwater 

mobility, as well as specific data about production and consumption quantities. Toxicological data, if 

available, are also important, but are not the main criterion. 

Methodology 

Within the last few years, the behavior of selected hydrophilic organic micropollutants during 

RBF was studied in waterworks on the lower Rhine River, as well as by laboratory-scale 


Dr.-Ing. Heinz-Jürgen Brauch 

Head of the Analytical Department 

DVGW-Technologiezentrum Wasser (TZW) 

Karlsruher Strasse 84 • D-76139 Karlsruhe, Germany 

Phone: +49(0)721/9678-150 • Fax: +49(0)721/9678-104 • Email: brauch@tzw.de 

137

138 

experiments. Most of the hydrophilic compounds under investigation are industrial chemicals 

with high production volumes and a broad range of applications; therefore, they are likely to enter 

river water if they are not totally removed in industrial or municipal wastewater treatment plants. 

In this paper, data on their fate during RBF will be presented, whereby the results of both lab-scale 

experiments and long-time measurements of river water and bank-filtrated water will be given. 

For the simulation of microbial degradation during RBF, a so-called “test filter” is used. This 

closed-loop apparatus was developed and used by Sontheimer and Völker (1987) for the 

characterization of fractions of surrogate parameters from wastewater effluents. In the last few 

years, the method was adjusted and optimized to study the biodegradation of single compounds at 

concentration levels relevant for the environment (Karrenbrock et al., 1999; Knepper at al., 

1999). Details of the experimental set-up are given elsewhere (Karrenbrock et al., 1999). 

Several waterworks along the lower Rhine River were selected to measure the behavior of organic 

micropollutants during RBF. All use bank-filtrated water from the Rhine River as raw water for 

drinking-water production (Schubert, 2000; Denecke, 1997; Brauch et al., 2000), whereby their 

wells are situated in a distance of 30 to 50 m from the Rhine River. In all cases, the raw water is a 

mixture between bank-filtrated water and groundwater, whereby the groundwater fraction ranges 

between 5 and 40 percent (i.e., the raw water is predominantly bank-filtrated water from the 

Rhine River). Regular measurements were performed over a time period of several years to attain 

reliable and significant data on the removal of organic compounds during underground passage 

(Brauch et al., 2000). 

Results 

Complexing Agents 

Aminopolycarbonic acids like nitrilotriacetic acid (NTA), EDTA, or diethylenetrinitrilopentaacetic 

acid (DTPA) are used as chelating agents in detergents and industrial cleaners, as well as in the 

photo, textile, and pulp- and paper-making industries. Due to their widespread use, NTA and 

EDTA are permanently found in the Rhine River in concentration levels of more than 1 µg/L 

(ARW and AWBR annual reports; Sacher et al., 1998). Besides these compounds, other 

complexing agents like ß-alaninediacetic acid (ADA) or 1.3-propylenedinitrilotetraacetic acid 

(PDTA) are used for special applications or as substitutes for EDTA. To test the biodegradation of 

these synthetic compounds, test-filter experiments were performed in which water from the Rhine 

River at Karlsruhe was spiked with NTA, EDTA, DTPA, PDTA, and ADA at a concentration 

level of 10-µg/L each. These experiments showed that the concentration of NTA decreased quite 

rapidly, indicating a fast microbial degradation of this complexing agent. The concentration of 

ADA decreased much slower and, after 35 days, about 30 percent of the initial concentration was 

still present. The concentrations of EDTA, PDTA, and DTPA seemed to be more or less constant, 

indicating that these complexing agents are persistent under the conditions of the test-filter 

experiment. 

Looking at the concentrations of NTA, EDTA, and DTPA in the Rhine River and in the raw 

water of the waterworks, which consists of at least 90-percent bank-filtrated water from the Rhine 

River, it is clear that in correspondence to the results of the test-filter experiments, NTA is nearly 

totally removed during RBF and is only sporadically found in raw water. On the other hand, 

EDTA proved to be recalcitrant and was present in the raw water under investigation. DTPA was 

found only once in raw water, but concentrations in the Rhine River are quite near to the limit 

of determination (which was 2 µg/L until 1996 and is currently 1 µg/L), and mixing with some

uncontaminated groundwater could lead to an apparent removal of DTPA; therefore, based on 

environmental data, no definite statement can be given on the behavior of DTPA during RBF. 

ADA is only sporadically found in the Rhine River, and PDTA could not be detected in the Rhine 

River (but is often found in the Neckar River, for instance). 

Aromatic Sulfonates 

Aromatic sulfonates are the corresponding bases to sulfonic acids and, due to their permanent 

negative charge, are highly soluble in water. Benzenesulfonates are mainly used as intermediates 

during the manufacture of azo dyestuffs, optical brighteners, ion-exchange resins, plasticizers, and 

pharmaceuticals. Amino- and hydroxynaphthalene-sulfonates and anthraquinonesulfonates are 

important building blocks for azo dyestuffs. A major source of sulfonated stilbenes is the production 

of fluorescent whitening agents for laundry products and paper. Naphthalenesulfonates and their 

condensates with formaldehyde are large-scale products that have widespread applications, 

including paper chemicals, superplasticisers for concrete, textile auxiliaries, and synthetic leather 

tanning agents (Redin et al., 1999). 

Test-filter experiments with 2-naphthalene-sulfonate and 1,5-naphthalenedisulfonate showed 

that the behavior of the 2-naphthalene-sulfonates is different. Whereas 2-naphthalene-sulfonate 

degraded very fast and, after 2 days, could not be detected in water, 1,5-naphthalenedisulfonate 

proved to be persistent and no change in concentration could be observed even after 30 days. The 

behavior of 1,5-naphthalenedisulfonate is characteristic for many two- or threefold sulfonated 

naphthalene compounds. Naphthalene-sulfonates with two sulfo groups in alpha position seem to 

be especially persistent (Lange et al., 1995). 

As could be expected from the results of the test-filter experiments, 1,5-naphthalenedisulfonate, 

which is almost always present in the Rhine River, is not removed during RBF and was always 

found in the raw water of the waterworks under investigation. The same behavior was found for 

2-amino-1,5- and 2-amino-4,8-naphthalene-disulfonate, for 1,3,5- and 1,3,6-naphthalenetrisulfonate, 

for cis-4,4’-dinitro-2,2’-stilbene-disulfonate, and for 8,8’-methylenebis-2-naphthalenesulfonate 

(Lange et al., 1995). 

Pharmaceutical Compounds 

Drugs were produced, prescribed, and used in quantities up to some hundred tons per year (in 

Germany). Due to an incomplete elimination in wastewater treatment plants, residues of 

pharmaceutical products have recently been found in surface waters and groundwaters (Ternes, 

1998; Heberer, 2002; Sacher et al., 2001b). In the Rhine River, compounds like diclofenac (an 

antirheumatic and analgesic), carbamazepine (an antiepileptic that is also used as antidepressant), 

and clofibric acid and bezafibrate (two lipid-regulating agents) are most often found with 

concentrations in the 10- to 100-nanograms per liter range. To study their behavior during RBF, 

test-filter experiments were performed, yielding that only bezafibrate is biodegradable under the 

conditions of the test-filter experiment. The concentration of carbamazepine decreases as a 

function of time, but even after 30 days, it is present in the test-filter system. The concentrations 

of diclofenac and clofibric acid are more or less constant, indicating that the respective 

compounds are not easily biodegradable. Monitoring data on the behavior of diclofenac and 

carbamazepine during RBF under environmental conditions show that that diclofenac was never 

detected in the raw water under investigation, although it was nearly always found in the Rhine 

River. Qualitatively, the same results were found for bezafibrate and clofibric acid (even if clofibric 

acid was only found a few times in the Rhine River). For bezafibrate, this finding is in accordance 

with the results of the test-filter experiments. For diclofenac and clofibric acid, there is a 

139

140 

contradiction to the results of the test-filter experiment, indicating that these two compounds are 

not biodegradable during underground passage. A satisfying explanation for the more effective 

elimination of diclofenac and clofibric acid under environmental conditions cannot be given. 

Furthermore, it can be seen from the monitoring data that carbamazepine is always present in the 

raw waters of all waterworks taking bank-filtrated water from the Rhine River, confirming the 

poor biodegradability of this compound found in the test-filter experiment. 

Conclusions 

Many organic micropollutants present in the Rhine River (and other river waters) are eliminated 

during RBF. Test-filter experiments prove that this elimination is mainly due to a microbial 

degradation of the compounds. Until now, less is known about metabolites and their behavior in 

the environment and during treatment processes in waterworks. Only few compounds are not 

totally removed during RBF and enter the raw waters used for drinking-water production. For the 

waterworks under investigation at the Rhine River, the compounds found in raw water could be 

totally removed by subsequent treatment steps like ozonation or GAC filtration, in most cases. 

Hence, the use of bank-filtrated water instead of river water is advisable if industrial or municipal 

wastewaters may affect river-water quality. The use of bank-filtrated water, however, cannot 

replace further treatment steps. 

REFERENCES 

Annual reports of Arbeitsgemeinschaft Rhein-Wasserwerke e.V. (ARW) and Arbeitsgemeinschaft Wasserwerke 

Bodensee-Rhein (AWBR). 

Brauch, H.-J., F. Sacher, E. Denecke, and T. Tacke (2000). “Wirksamkeit der Uferfiltration für die Entfernung 

von polaren organischen Spurenstoffen.” gwf-Wasser/Abwasser, 141: 226-234. 

Brauch, H.-J., U. Müller, and W. Kühn (2001). “Experiences with riverbank filtration in Germany.” 

Proceedings, International Riverbank Filtration Conference, Rheinthemen 4, 33-39. 

Denecke, E. (1997). “Auswertung langzeitlicher Messreihen zur aeroben Abbauleistung der Uferpassage einer 

Wassergewinnungsanlage am Niederrhein. Z.” Wasser-Abwasser-Forschung, 25: 311-318. 


A review of recent research data.” Toxicology Letters, 131: 5-17. 

Karrenbrock, F., T.P. Knepper, F. Sacher, and K. Lindner (1999). “Entwicklung eines standardisierten 

Testfilters zur Bestimmung der mikrobiellen Abbaubarkeit von Einzelsubstanzen.” Vom Wasser, 92: 361-371. 

Knepper, T.P., F. Sacher, F.T. Lange, H.-J. Brauch, F. Karrenbrock, O. Rörden, and K. Lindner (1999). 

“Detection of polar organic substances relevant for drinking water.” Waste Management, 19: 77-99. 

Kühn, W. and U. Müller (2000). “Riverbank filtration — An overview.” Journal AWWA, 92: 60-69. 

Lange, F.T., M. Wenz, and H.-J. Brauch (1995). “The behaviour of aromatic sulfonates in drinking water 

production from River Rhine water and bank filtrate.” Analytical Methods and Instrumentation, 2: 277-284. 

Redin, C., F.T. Lange, H.-J. Brauch, and S.H. Eberle (1999). “Synthesis of sulfonated naphthaleneformaldehyde 

condensates and their trace-analytical determination in wastewater and river water.” Acta 

Hydrochim. Hydrobiol., 27: 136-142. 

Sacher, F., E. Lochow, and H.-J. Brauch (1998). “Synthetische organische Komplexbildner - Analytik und 

Vorkommen in Oberflächenwässern.” Vom Wasser, 90: 31-41. 

Sacher, F., and H.-J. Brauch (1999). “Bewertung organischer Einzelstoffe im Hinblick auf ihr Verhalten bei 

der Wasseraufbereitung.” Veröffentlichungen aus dem Technologiezentrum Wasser, 7: 111-127.

Sacher, F., H.-J. Brauch, and W. Kühn (2001a). “Fate studies of organic micropollutants in riverbank 

filtration.” Proceedings, International Riverbank Filtration Conference, Rheinthemen, 4: 139-148 

Sacher, F., F.T. Lange, H.-J. Brauch, and I. Blankenhorn (2001b). “Pharmaceuticals in groundwaters — 

Analytical methods and results of a monitoring program in Baden-Württemberg, Germany.” J. Chromatogr., 

A 938: 199-210. 

Sacher, F., and H.-J. Brauch (2002). “Experiences on the fate of organic micropollutants during riverbank 

filtration.” Understanding contaminant biogeochemistry and pathogen removal, C. Ray (ed.), Kluwer Academic 

Publishers, The Netherlands, p. 135-151. 

Schubert, J. (2000). “Entfernung von Schwebstoffen und Mikroorganismen sowie Verminderung der 

Mutagenität bei der Uferfiltration.” gwf-Wasser/Abwasser, 141: 218-225. 

Sontheimer, H. (1980). “Experiences with riverbank filtration along the Rhine River.” Journal AWWA, 72: 3,386-3,392. 

Sontheimer, H., and E. Völker (1987). Charakterisierung von Abwassereinleitungen aus der Sicht der Trinkwasserversorgung. 

Veröffentlichungen des Bereichs und Lehrstuhls für Wasserchemie am Engler-Bunte-Institut der 

Universität Karlsruhe 31. 

Sontheimer, H. (1991). Trinkwasser aus dem Rhein? Bericht über ein Verbundforschungsvorhaben zur Sicherheit 

der Trinkwassergewinnung aus Rheinuferfiltrat, Academia Verlag, Sankt Augustin. 

Ternes, T.A. (1998). “Occurrence of drugs in German sewage treatment plants and rivers.” Wat. Res., 

32: 3,245-3,260. 

HEINZ-JÜRGEN BRAUCH has over 20 years of experience in resolving water-quality 

problems, with special emphasis in drinking-water production. He has conducted many 

research projects on the development and optimization of analytical determination 

methods for organic micropollutants, including emerging contaminants and fate studies on 

ground surface and drinking water, as well as the design and implementation of monitoring 

strategies for controlling hazardous substances. In addition, he has completed several 

projects in cooperation with water utilities to improve and optimize the treatment 

techniques for removing organic substances. Brauch has published numerous articles and papers on modern 

analytical techniques for detecting organic micropollutants, fate and behavior of organics in drinking-water 

treatment, occurrence and distribution of organic substances between water, and sediment and soil, as well 

as on water-quality problems in surface water and groundwater. Brauch has been the Head of the Analytical 

Department of DVGW-Technologiezentrum Wasser (Technology Center for Water) in Karlsruhe, Germany 

since 1990, and is the Chairman or Member of many national and international working groups in this field. 

He received a Ph. D. in Chemical Engineering from the University of Karlsruhe. 

141

142

Session 7: Organics Removal 

Organics Removal by Riverbank Filtration 

at the Greater Cincinnati Water Works Site 

Jeffrey Vogt 



William Fromme 



Bruce Whitteberry, P.G. 






Objective 

With the upcoming promulgation of the Stage 2 Disinfectants and Disinfection By-Product Rule, 

water utilities are required to reduce DBPs in their finished water (USEPA, 2003). The 

concentration of NOM in source waters is directly related to the concentration of DBPs formed 

in finished waters (Owen et al., 1993). The challenge for water utilities is to reduce precursors and 

NOM prior to disinfection. The natural process of RBF may be an effective method to remove 

NOM. As part of a research study, TOC, UV 254, and THM formation potential were used to 

evaluate the organic removal capabilities of RBF. This project was a cooperative study between the 

USGS, Miami University in Oxford, Ohio, and the Greater Cincinnati Water Works. This was a very 

large project with many objectives. The objective of this paper is to evaluate the removal of NOM 

by RBF. 

Background 

The Greater Cincinnati Water Works owns and operates a 40-MGD drinking-water treatment 

plant located in southwestern Ohio. The source water for the plant comes from 10 production 

wells located in an alluvial aquifer system adjacent to the Great Miami River. The aquifer is a 

mixture of sand and gravel and has a hydraulic conductivity of 200 to 300 ft (62 to 91 m) per day. 

Five monitoring wells were drilled adjacent to the river and close to a production well (CMB1). 

Each monitoring well was drilled at varying depths in relation to the production well. Well 1A is 

the shallowest vertical well and is the nearest to the river. Well 1D is the deepest well and closest 

to CMB1. Wells 1B and 1C were placed at intermediate depths between Wells 1A and 1D. 


Jeffrey Vogt 

Chemist 


5651 Kellogg Ave • Cincinnati, Ohio 45228 USA 

Phone: (513) 624-5624 • Fax: (513) 624-5670 • Email: Jeff.Vogt@gcww.cincinnati-oh.gov 

143

144 

Well 1C is located at the top of the production well screen, and Well 1D is located at the bottom 

of the production well screen. To identify discrete sections of the aquifer, the monitoring wells 

were developed with 2-ft (0.6 m) screened interval. An inclined well (Well 1I) was drilled from 

the top of the riverbank at a 20- to 30-degree angle from the horizontal plane. This well allowed 

the study team to collect samples approximately 5- to 10-ft (1.5- to 3.1-m) below the streambed. 

Samples were collected and analyzed on a weekly basis and, later, at monthly intervals from 

September 1999 until May 2001. 

Materials and Methods 

TOC and UV 254 were used to evaluate NOM removal. TOC analyses were analyzed with a 

Tekmar/Dohrman Phoenix 8000 using the UV-Persulfate Method. UV absorbance at 254-mn 

wavelength was performed with a Hach DR4000. UV absorbance is primarily related to the humic 

fraction of NOM. UV at a wavelength of 254 nm is absorbed by double bonds and/or aromatic 

structures mostly produced from the breakdown of plant and animal matter (Owen et al., 1993). 

THM formation potential were used to evaluate the removal of DBP precursors. The concept 

behind the analysis is to maximize the THM formation reaction. Conditions (chlorine dose, pH, 

hold time, and high temperature) are set so that the reaction is “pushed” to form THMs. A dose 

of 15-mg/L free chlorine was applied to the samples using a sodium hypochlorite solution. The 

sample pH was adjusted to 9.5 with a borate buffer and then incubated at 35-degrees Celsius for 

7 days. After incubation, residual chlorine was measured. Samples were poured off and analyzed 

for THMs with a Varian 3400 Gas Chromatograph using USEPA Method 502.2. 


All three of the parameters demonstrated the ability of RBF to remove organic material. Table 1 

represents water-quality data obtained from the river, inclined well, monitoring wells, and production 

well. The removal values represent removals from the Great Miami River through the wells. The 

average river results show relatively high concentrations of NOM as compared to drinking-water 

standards. Organic material was removed through the streambed (Well 1I), and is continually 

reduced as it migrates downward into the aquifer. Peaks in organic concentrations in the Great 

Miami River correlated with peak concentrations in Well 1I, yet at a reduced concentration (Figure 1). 

TOC is reduced 37 percent from the Great Miami River to Well 1I. An additional 20 percent of 

TOC was removed to Well 1C. This produces a total reduction in TOC of 59 percent. UV and 

THM formation potential demonstrated even better reductions of, respectively, 63 and 72 percent. 

Table 1. Average Values and Percent Removals from the Great Miami River 

Over the Entire Study Period 

Great Well 1I Well 1A Well 1B Well 1C Production Well 1D 

Miami Well 

River CMB1 

DEPTH (ft) 5 to 10 31 46 60 60 88 

Avg. Avg. % Avg. % Avg. % Avg. % Avg. % Avg. % 

TOC 5.48 3.36 39 2.50 54 2.32 58 2.27 59 1.17 79 0.63 89 

UV 254 0.147 0.092 37 0.063 57 0.057 61 0.055 63 0.023 84 0.007 95 

THMFP 924 404 56 277 70 267 71 260 72 131 86 55.3 94 

THMFP = THM formation potential.

TOC (mg/L) 

9 

8 

7 

6 

5 

4 

3 

2 

1 

Site 1 Total Organic Carbon 

GMR FP1i FP1A FP1B FP1C CMB1 FP1D 

0 

9/1/99 12/2/99 3/3/00 6/3/00 9/3/00 

Date 

12/4/00 3/6/01 6/6/01 9/6/01 

Figure 1. TOC for the Great Miami River and well samples for the entire study period. 

Figure 2 presents UV 254 data for all study locations. The bulk of UV reduction is observed between 

the Great Miami River and Well 1I (37 percent). An additional 20-percent reduction is seen 

between Wells 1I and 1A. The reduction between Wells 1A and 1B is about 4 percent. Between 

Wells 1B and 1C, there is about 2-percent reduction. The downward trend of UV reduction 

continues through all of the locations, but the zone surrounding Wells 1A, 1B, and 1C 

demonstrates minor change. The similarities within that zone can be observed in the maximum, 

average, and minimum values for those wells. Additional reductions are seen at Well 1D. Results 

for Well 1D showed very low organic concentrations in all measured parameters when compared 

to others wells. It is likely that this water is a combination of regional groundwater and water with 

a much longer flow path to the production well. This downward trend or pattern for reduction can 

be observed in data plots of all three parameters. 

CMB1 has a 30-ft (9-m) intake screen. It draws in water from a much larger capture zone than the 

monitoring wells due to a much longer screen and significantly higher pumping rates. The water 

entering the production well is a mixture of water passing through Wells 1D and 1C, as well as 

other parts of the aquifer. The results for CMB1 fall between the results for both Wells 1C and 1D 

in three parameters. 

CMB1 showed the least amount of variation in the data, as compared to Wells 1A, 1B, and 1C in 

Figure 2. UV results of the samples collected at CMB1 showed a maximum value of 0.029 cm –1 , 

an average value of 0.023 cm –1 , and a minimum value of 0.016 cm –1 . Figure 2 shows the UV results 

for CMB1 plotting very close together. The calculated standard deviation for CMB1 UV data is 

0.003 cm –1 . Comparing that value to the standard deviation of Well 1C (0.009 cm –1 ), it is 

observed that CMB1 shows little variation in its results. This trend of consistent results between 

the data points can be seen in the THM formation potential and TOC results for CMB1. This is 

likely due to the large capture zone normalizing the concentration. 

145

146 

Absorbance (cm –1 ) 

0.30 

0.25 

0.20 

0.15 

0.10 

0.05 

0.00 

UV 254 Absorbance 

Maximum Average Minimum 

GMR FP1i FP1A FP1B FP1C CMB1 FP1D 

Location 

Figure 2. Maximum, average, and minimum UV values at sample locations. 

Looking again at Figure 1, all of the wells except for the inclined well demonstrated very 

consistent results. The average results for Wells 1A, 1B, and 1C are similar for all of the parameters 

(see Figure 2). These data demonstrate that RBF is very effective in reducing any organic shock 

load from the river. High concentration peaks in the river and inclined well are unnoticed in the 

remaining wells. Others (Kuehn et al., 2000) have also demonstrated this process when using RBF. 

THM formation potential can be used as a surrogate for THM concentrations in finished water 

(Owen et al., 1993). The ultimate goal of the Stage 2 Disinfectants and Disinfection By-Product 

Rule is to reduce THMs in finished waters produced by water treatment utilities across the 

country. UV and TOC measure bulk concentrations of organic matter, but THM formation 

potential is a measurement of actual THM concentrations. The organic matter, which directly 

reacts with the chlorine to form THMs, was observed to be the most reduced through the aquifer. 

Fifty-six percent of that matter was reduced at Well 1I, with an additional reduction of 21 percent 

through the 1A, 1B, and 1C zone. At CMB1, the matter was reduced to a total reduction of 

86 percent. These results suggest that RBF does an excellent job in reducing organic matter that 

forms THMs. 

The data from this project also supports the concept that the natural process of RBF produces 

consistent water quality through time. Through this study period, there was no apparent breakthrough 

of organic material. This would indicate that the processes, which remove NOM in the 

aquifer, do not reach a saturation point, as would be expected in an engineered system, such as 

GAC.

Conclusions 

RBF is very effective in reducing organic matter in source waters. The reduction of NOM was 

demonstrated using TOC and UV 254 absorbance. THM precursor removal was demonstrated 

using THM formation potential. 

• TOC reductions ranged from 39 to 79 percent between the Great Miami River and 

CMB1. 

• UV254 absorbance reductions ranged from 37 to 84 percent between the Great Miami 

River and CMB1. 

• THM formation potential demonstrated the greatest range of reductions between the 

Great Miami River and CMB1, with values of 56 to 86 percent. 

• RBF is very effective in reducing organic shock loading. The high concentrations seen in 

the river were not seen in the 1A, 1B, or 1C monitoring wells. During the study period, 

there was never a breakthrough of organic matter. The results for the three parameters 

remained consistent when compared to varying concentrations in the Great Miami River. 

• Production well (CMB1) data remained very consistent in all of the measured parameters. 

REFERENCES 

Owen, D.M., G.L. Amy, and Z.K. Chowdhury (1993). Characterization of Natural Organic Matter and Its 

Relationship to Treatability, American Water Works Research Foundation, Denver, Colorado. 

USEPA (2003). Stage 2 Disinfectants/Disinfection By Product Rule, United States Environmental Protection 

Agency, Washington, D.C. 

Kuehn, W., and U. Mueller (2000). “Riverbank filtration: An Overview.” Journal AWWA, 92(12): 60-69. 

JEFFREY VOGT is a Chemist with the Greater Cincinnati Water Works, where he has 

worked since 1992. For the last 5 years, he has been involved in riverbank-filtration 

issues/research and was largely involved in the 3-year Flowpath Study. His responsibilities 

during the study involved planning, sampling, analyses, data evaluation, and interpretation 

of the results. In 2002, he presented and published at the American Water 

Resources Association’s Groundwater/Surface Water Interactions Conference in 

Keystone, Colorado. Some of his current responsibilities include managing continuing 

riverbank-filtration issues, parasite-monitoring program for the Long Term 2 Enhanced Surface Water 

Treatment Rule at the Greater Cincinnati Water Works’ Richard Miller Treatment Plant, and treatment 

studies for the Greater Cincinnati Water Works’ three treatment plants. Vogt holds a Class 3 Water 

Treatment license in the State of Ohio and received a Science degree from the University of Cincinnati. 

147

148

Lunch Presentation 

Potential Uses of Riverbank Filtration 

for Regulatory Compliance 

Stig Regli 

United States Environmental Protection Agency 

Washington, D.C. 

RBF may have its most general application for systems seeking compliance with the LT2ESWTR. 

On August 11, 2003, the USEPA proposed the LT2ESWTR and included provisions by which 

RBF could be used as one of the compliance options for providing Cryptosporidium removal credits 

(USEPA, 2003). While the USEPA has previously recognized (through guidance implementation 

decisions) that RBF is a technology that can achieve pathogen removal, the LT2ESWTR is the 

first United States drinking-water regulation that specifically recognizes RBF as a compliance 

technology option. 

Under the proposed LT2ESWTR, filtered systems must monitor source water for Cryptosporidium 

to determine what source-water bin concentration category it belongs in and whether additional 

treatment is required. As part of this determination, the USEPA provides a “toolbox” of technologies 

by which systems can assess their total removal/inactivation credits for Cryptosporidium. 

The proposed LT2ESWTR recognizes RBF as a “toolbox” pretreatment technique that can 

provide a system 0.5- or 1.0-log additional pretreatment credit, if it meets specified design criteria 

and monitoring criteria. 

For RBF to be eligible for credit as a pretreatment technique, the following proposed criteria must 

be met: 

• Wells must be drilled in an unconsolidated, predominantly sandy aquifer, as determined 

by grain-size analysis of recovered core material — the recovered core must contain 

greater than 10-percent fine-grained material (grains less than 1.0-millimeter diameter) 

in at least 90 percent of its length. 

• Wells must be located at least 25 ft (in any direction) from the surface-water source to be 

eligible for 0.5-log credit; wells located at least 50 ft from surface water are eligible for 

1.0-log credit. 

• The wellhead must be continuously monitored for turbidity to ensure that no system 

failure is occurring. If the monthly average of daily maximum turbidity values exceeds 

1 ntu, the system must report this finding to the State. The system must also conduct an 

assessment to determine the cause of high turbidity levels in the well and consult with 

the State to determine whether the previously allowed credit is still appropriate. 

Systems using RBF as pretreatment to a filtration plant at the time that the system is required to 

monitor for Cryptosporidium must sample the well effluent for the purpose of determining the bin 


Stig Regli 

Environmental Engineer 


OGWDW (4607M) • 1200 Pennsylvania Avenue NW • Washington, D.C. 20460 USA 

Phone: (202) 564-5270 • Fax: (202) 564-3767 • Email: regli.stig@epa.gov 

149

150 

classification. Where bin classification is based on monitoring the well effluent, systems are not 

eligible to receive additional credit for RBF. The rationale for the above proposed criteria and 

opportunity for public comment are described in detail in the Federal Register (68FR47692) and 

are available on the web at http://www.regulations.gov/fredpdfs/03-18295.pdf. 

RBF also provides the opportunity for directly reducing organic DBP precursor levels or indirectly 

facilitating the application of advanced precursor removal technologies, such as nanofiltration. 

The USEPA is proposing the Stage 2 Disinfection By-Product Regulation to mitigate concerns for 

the potential risk of developmental and reproductive effects from DBPs. Since the compliance 

dates of the LT2ESWTR will coincide with those of the Stage 2 Disinfection By-Product 

Regulation, there may be opportunities for utilities to use RBF for helping to achieve compliance 

with both regulations. 

The simultaneous reduction of other regulated contaminants by irreversible adsorption, biodegradation, 

dilution with groundwater, or attenuation mechanisms is also possible with RBF. This is 

clearly a site-specific issue whose success cannot be assumed without adequate testing/monitoring. 

REFERENCE 


Rule; Propose Rule, Federal Register, 68(154): 47,691-47,696. 

STIG REGLI is an Environmental Engineer for the Office of Ground Water and Drinking 

Water of the United States Environmental Protection Agency. He has been with the 

United States Environmental Protection Agency since 1979 and is involved with 

developing national drinking-water regulations for public water systems. His major focus 

has been as a Regulation Manager (1985 to 1996) and, more recently, as a Senior Science 

Advisor pertinent to the control of pathogens and disinfection byproducts. He has also 

been on extended leave to work on drinking-water related projects in Somalia and 

Thailand. Prior to working at the United States Environmental Protection Agency, he taught environmental 

engineering courses as a Peace Corps volunteer at Kabul University in Kabul, Afghanistan. Regli received 

both a B.S. in Mechanical Engineering and an M.S. in Civil Engineering from Duke University.

Session 8: Emerging Contaminants Removal 

Transport and Attenuation of 

Pharmaceutical Residues During Bank Filtration 

Andy Mechlinski 

Institute of Food Chemistry 



Thomas Heberer, Ph.D. 




Bank filtration and artificial groundwater recharge are important, effective, and cheap techniques 

to treat surface water and remove microbes and inorganic and (some) organic contaminants; 

however, the purification capacity of these techniques varies and is limited in the removal of some, 

but not all, potential impurities. Thus, bank-filtration research began investigating pharmaceutically 

active compounds (PhACs) when a number of groundwater samples from bank filtration sites 

positively detected these compounds. To date, the mechanisms for removing impurities and 

chemical reactions of the water components have not sufficiently been understood. These subjects 

are currently addressed in a new research project called NASRI. In this interdisciplinary project, 

the fate and transport of some new emerging contaminants during bank filtration are investigated 

at Tegel and Wannsee Lakes and at a groundwater replenishment infiltration pond in Berlin, 

Germany. The locations of the field-sites are shown in Figure 1. 

The field sites are equipped with different types of monitoring wells that screen at various depths 

and are drilled between the infiltration bank and water-supply wells, as well as behind the watersupply 

wells. An example of the transects is shown in Figure 2. 

These transects are sampled monthly, which allows the fate and transport of PhACs during 

groundwater recharge to be monitored. The samples are analyzed by solid phase extraction, 

chemical derivatization, and gas chromatography-mass spectrometry (GC/MS) applying selected 

ion monitoring. Two novel analytical methods detect PhACs even in complex environmental 

samples (Reddersen, 2003). Additionally, some other PhACs (such as antibiotics and estrogenic 

steroids) are analyzed by high-pressure liquid chromatography mass spectrometry (HPLC-MS/MS). 

In Berlin surface waters, PhACs are found up to the microgram-per-liter level as highly persistent 

residues. Six PhACs (Figure 3), including the analgesic drugs diclofenac and propyphenazone, the 

antiepileptic drugs carbamazepine and primidone, and the drug metabolites clofibric acid and 

1-acetyl-1-methyl-2-dimethyl-oxamoyl-2-phenylhydrazide (AMDOPH), were found to leach from 

contaminated watercourses into groundwater aquifers. They also occur at low concentrations in 

receiving water-supply wells. Bank filtration was, however, also found to decrease concentrations 

(e.g., of diclofenac) or even remove some PhACs (bezafibrate, indomethacine, antibiotics, and 

estrogens). Other PhACs (such as carbamazepine and, especially, primidone) were identified as 

being excellent tracers of sewage contamination in surface water and groundwater. 


Andy Mechlinski 

Graduate Student 


Technical University of Berlin • Sekr. TIB 4/3-1 • Gustav-Meyer-Allee 25 • 13355 Berlin, Germany 

Phone: +49 (30) 314-72267 • Fax: +49 (30) 314-72823 • Email: andymechlinski@web.de 

151

152 


The authors would like to thank Veolia Water and the Berlin Water Company for financing the 

NASRI project. 

REFERENCE 

TS Tegeler See 

GWA Tegel 

3339 

3338 

Sickerbecken 

TS Lieper Bucht 

Unterhavel 

TEG365 TEG248 

TEG364 

TEG247 

3337 3335 

4 

3336 3334 

TS Wannsee 

Galerie Wannsee 

3333 

3332 

3311 

3310 

Br.20 

Tegeler See 

3301 

3302 

3309 

3308 

3307 

3306 12 

3303 13 3305 

14 

Bernauerst. 

3304 

Galene West 

WW Kladow 

Testfeld 

Marienfelde 

WW Spandau 

Spandau 

Havel 

WW Stolpe 

Kleinmachnow 

KW Stahnsdorg 

Tegeler See 

PEA-Tegel 

WW Tegel 

WW Jungfernheide 

Spree 

Langsamsandfilter 

Uferfiltration 

8 Peilrohre 

Tegeler Fließ 

Charlottenburg 

KW Ruhleben 

Unterschleuse 

WW Tiefwerder 

WW und PEA 

Beelitzhof 

Einleitung 

KW Ruhleben 

Teitkowkanal 

Messstellen 

KW Schönerlinde 

(KW Marienfelde) 

KW Waßmannsdorf 

KW Münchehofe 

WW Friedrichshagen 

Reddersen, K., and T. Heberer (2003). “Multi-methods for the trace-level determination of pharmaceutical 

residues in sewage, surface and ground water samples applying GC-MS.” J. Sep. Sci. (in press). 

Nordgraben 

Panke 

Speicherteich 

Mühlendamm 

Oberschleuse 

Landwehrkanal 

Neukölln 

Spree 

Panke 

Tettowkanal 

MHG 

KW Falkenberg 

WW Wuhlheide 

Wuhle 

WW Johannisthal 

TS Müggelsee E. 

Dahme 

Erpe 

Wasserwerk waterworks 

Klärwerk sewage treatment plant 

Phospat-Eliminationsanlage 

surface water treatment plant 

TS Transsekte Testfield 

Wehr weir 

Schleuse lock 

Nottekanal 

Neue Mühle 

Fredersdorfer 

Fließ 

Wernsdorf 

© Free University of Berlin 

Federal Environmental Agency (UBA) 

TS Müggelsee C. 

Flakenfließ und 

Löcknitz 

Figure 1. Locations of the Tegel transect, Wannsee transect, and groundwater recharge area in Berlin, Germany. 

SW 

? ? 

existing observation wells 

Pegel Tegler See 

3310 3309 

3311 

? 

3301 3308 

TEG371OP 

TEG371UP 

3302 3307 

approximately 120 m 

TEG372 

3303 

3306 

new observation wells (08/2002) 

glacial till 

Br 13 

NE 

3304 3305 

Figure 2. Profile of the Tegel transect with 14 observation wells and Water Works Well Br. 13. 

Woltersdorf 

Große Tränke 

depth below 

ground [m] 

0 m 

5 m 

10 m 

15 m 

20 m 

25 m 

30 m 

35 m 

40 m 

45 m 

veritas 

iustitia 

libertas 

Freie Universität Berlin

H 3C O O N 

N 

H3C N CH3 Cl 

AMDOPH Bezafibrate 

O 

H 5C 6 

H 5C 2 

Cl 

CH 3 

H 

N 

O 

CH 3 

Clofibric Acid 

NH 

CH 3 

C C COOH 

CH 3 

N 

O 

C 

N 

HOOC 

Primidone Propyphenazone Indometacine 

Figure 3. Structures of some detected PhACs. 

NH-CH 2CH 2 

O 

ANDY MECHLINSKI is a Ph.D. student at the Institute of Food Chemistry of the 

Technical University of Berlin in Germany. His research interests include the analysis of 

pharmaceutically active compounds by gas chromatography-mass spectrometry and the 

investigation of the environmental fate and transport of pharmaceuticals during 

groundwater recharge at bank-filtration sites in Berlin. In 2001, he took the first part of 

the final examination covering food chemistry at the Technical University of Berlin. After 

this, he prepared a diploma thesis concerning the transport and attenuation of pharmaceuticals 

during bank filtration at two field sites in Berlin. From 2002 until 2005, he will be involved in the 

interdisciplinary Natural and Artificial Systems for Recharge and Infiltration project. Mechlinski recieved 

his undergraduate degree at the Institute for Food Chemistry of the Technical University of Berlin, where he 

investigated the behavior of several pharmaceuticals and other contaminants during riverbank filtration at 

two field sites in Berlin. 

H 3C 

CH 3 

C 

CH 3 

O 

CH 3 

CH 2 

H 

N 

Cl 

Diclofenac 

N 

O 

Carbamazepine 

Cl 

C 

2H 2 

CH 3 

N 

C CH 2 

O 

COOH 

Cl 

153

154


Attenuation of Pharmaceuticals 

During Riverbank Filtration 

Traugott Scheytt, Ph.D. 

Institute of Applied Geosciences 

Technical University Berlin 


Petra Mersmann 




Marcus Leidig 




Thomas Heberer, Ph.D. 




Objective 

Occurrences of PhACs from human medical care are reported in groundwater not only from the 

Berlin (Germany) area (Heberer et al., 1998; Scheytt et al., 1998), but also from several places 

worldwide (Heberer, 2002). These findings initiated more specific investigations concerning the fate 

and transport of those pharmaceuticals under water-saturated and unsaturated conditions (Scheytt 

et al., in preparation). Field investigations at the bank infiltration site at Lake Tegel (Berlin, 

Germany) show the presence of several classes of pharmaceuticals, such as antirheumatics 

(e.g., diclofenac), analgesics (e.g., propyphenazone), and blood lipid regulators (clofibric acid) in 

both surface water and groundwater (Verstraeten et al., 2003). At Lake Tegel, clofibric acid was found 

at concentrations up to 290 nanograms per liter and propyphenazone up to 250 nanograms per liter, 

whereas concentrations of diclofenac were around detection limit. 

These results will be compared to preliminary results from an ongoing study at the Santa Ana 

River in Orange County, California. Using its forebay facilities, the Orange County Water District 


Traugott Scheytt, Ph.D. (after December 2003) 

Department of Civil and Environmental Engineering 

609 Davis Hall • University of California • Berkeley, California 94720 USA 

Phone: (510) 642-0151 • Fax: (510) 642-7483 • Email: scheytt@ce.berkeley.edu 

Traugott Scheytt, Ph.D. (until December 2003) 

Associate Professor for Hydrogeology 


Technical University Berlin • Ackerstr. 71-76 • 13355 Berlin, Germany 

Phone: +49 30 314-72417 • Fax: +49 30 314-25674 • Email: traugott.scheytt@tu-berlin.de 

155

156 

recharges surface water from the Santa Ana River into the Orange County Groundwater Basin. 

The fact that a high amount of the Santa Ana River’s flow is of wastewater origin has prompted 

concern that PhACs that survive secondary and tertiary treatment and groundwater transport may 

end up in the aquifer. 

To understand the transport and attenuation processes of those substances, laboratory column 

experiments were conducted (Mersmann et al., 2002). Based on these results, it is possible to 

identify and predict the transport and fate of PhACs during bank infiltration and to identify the 

pharmaceuticals, which may enter the aquifer via bank infiltration. 


Pleistocene sediments from a well drilling campaign carried out by the Berlin Water Works were 

used to prepare the soil column for the column leaching experiment. The sediment was sampled 

at a depth of approximately 60-m below ground level and consisted of medium-grained sand. The 

sediment was manually packed into a stainless steel column measuring 35 × 14 centimeters (inner 

diameter). A gauze net and 0.5-diameter globes were placed at both the top and bottom of the 

column to prevent soil particles from leaching. The column was pre-wetted and, afterwards, 

equilibrated with groundwater originating from the same location and depth like the sediment 

itself. Groundwater was led through the column with a flow rate of about 0.3 m per day and a 

bottom-to-top flow direction to ensure saturated conditions in the column. 

Equilibration of the column with pure groundwater took about 5 days (4.83 pore volumes) before 

pharmaceutical compounds were applied. The groundwater used for the column experiment 

represents a typical groundwater from the Berlin area not contaminated by any PhAC residues. 

Lithium chloride (LiCl) (used as tracer) and the pharmaceuticals were applied to the same 

groundwater, which was then passed through the column for approximately 10 days. Beside the 

tracer and pharmaceutical compounds, all other parameters (e.g., flow rate) were kept the same 

during all three phases of the study. The experiments took place at a room temperature of 

approximately 20-degrees Celsius and all parts of the column experiment, including the tank, were 

protected against exposure to light. The eluted liquid was collected in fractions of approximately 

25 milliliters and analyzed for contents of anions, cations, pharmaceutical chemicals, and the 

lithium chloride used as a tracer. 

The concentration of the lithium chloride tracer was 10 mg/L, and the pharmaceuticals had a 

concentration of 10 µg/L in the spiked water. Physico-chemical parameters (redox potential, pH, 

temperature, oxygen saturation, specific conductance) were measured every 10 minutes using 

respective electrodes coupled to a data logger. Lithium chloride was chosen as a tracer because the 

background concentration of lithium was definitely below detection limit in all sediment and 

groundwater samples used for the experiments. Lithium also shows a transport behavior comparable 

to a nonreactive tracer and can be analyzed in a rapid and cost-effective manner. 

For the analysis of pharmaceutical compounds, water samples were adjusted to a pH of 2 and then 

extracted by solid-phase extraction using a non-endcapped reversed phase adsorbent 

(RP-C18 Bakerbond Polar Plus). Then the analytes and surrogate standard were derivatized, 

making them amendable to gas chromatographic separation (Heberer et al., 1998). 

Two microliters of the sample extracts (100 microliters for each sample) were analyzed by capillary 

GC-MS with selected ion monitoring. Depending on the sample volume (100 to 1,000 milliliters), 

the limits of determination were between 1 and 10 nanograms per liter, and the limits of quantitation 

were between 5 and 25 nanograms per liter. The analytical recoveries range between 80 and 120 percent. 

For further analytical details, refer to Heberer et al. (1998).

Results and Conclusion 

The Santa Ana River study will provide information about the concentration of some high-volume 

pharmaceuticals from human medical care in the Santa Ana River, especially acetaminophen, 

carbamazepine, diclofenac, gemfibrozil, ibuprofen, metoprolol, naproxen, primidone, and propranolol. 

Laboratory experiments show that clofibric acid exhibits no degradation and almost no 

retardation (Rf = 1.1). After spiking groundwater with lithium chloride and diclofenac, the 

movement of diclofenac within the column is much slower than the movement of the lithium 

tracer (Figure 1), leading to a retardation factor of Rf = 2.6 for diclofenac. Propyphenazone 

(Rf = 2.0) is also retarded, whereas significant degradation was not observed for both pharmaceuticals, 

diclofenac and propyphenazone, under prevailing conditions in the soil column. Ibuprofen 

is degraded under aerobic conditions, whereas only little degradation was observed under 

anaerobic conditions. The retardation factor for ibuprofen was extrapolated to be 4.0. 

Carbamazepine shows no degradation in the soil column experiments, but significant retardation 

(Rf = 2.8) under prevailing conditions. 

C/C 0 

1.2 

1.0 

0.8 

0.6 

0.4 

0.2 

0.0 

0.0 5.0 10.0 15.0 

Pore Volume 

20.0 25.0 30.0 

Lithium Diclofenac 

Figure 1. Concentration of lithium and diclofenac at the outflow of the soil 

column; results from a single-substance laboratory soil column experiment. 

It was concluded that at least clofibric acid, propyphenazone, and carbamazepine are recalcitrant 

under groundwater conditions and will be transported within the aquifer at the Berlin bank-filtration 

site. Additionally, due to anaerobic conditions in the deeper part of the aquifer, diclofenac and 

ibuprofen may also occur in groundwater and are transported if these compounds are not completely 

degraded in surface water or in the aerobic part of the aquifer. 

Compared to the bank-infiltration site at Lake Tegel in Berlin, the situation is quite different at 

the Santa Ana River, especially in respect to the climate and hydrogeological setting. Among the 

already mentioned pharmaceuticals of interest at the Santa Ana River, acetaminophen is 

prescribed in high amounts, but it is expected that this compound will be attenuated during bank 

infiltration due to its high biodegradability. Additionally, due to mostly aerobic conditions in the 

aquifer at the Santa Ana River, diclofenac and ibuprofen will be degraded under these aerobic 

conditions. Among the pharmaceuticals that might be expected in groundwater are carbamazepine, 

primidone, gemfibrozil, metoprolol, naproxen, and propranolol. 

157

158 

Acknowledgments 

Deutsche Forschungsgemeinschaft has funded parts of this work, and the National Water Research 

Institute is funding the research on attenuation of pharmaceuticals during recharge at the Santa 

Ana River. The authors appreciate the cooperation of the Berliner Wasserbetriebe in gaining 

access to waterworks wells and the help of the Orange County Water District, which has greatly 

contributed to the study. 

REFERENCES 


A review of recent research data.” Toxicology Letters, 131: 5-17. 

Heberer, T., K. Schmidt-Bäumler, and H.-J. Stan (1998). “Occurrence and distribution of organic contaminants 

in the aquatic system in Berlin. Part I: Drug residues and other polar contaminants in Berlin surface 

and groundwater.” Acta Hydrochimica et Hydrobiologica, 26: 272-278. 

Mersmann, P., T. Scheytt, and T. Heberer (2002). “Column experiments on the transport behavior of 

pharmaceutically active compounds in the saturated zone.” Acta Hydrochimica et Hydrobiologica, 

30(5-6): 1-10. 

Scheytt, T., T. Mersmann, M. Leidig, A. Pekdeger, and T. Heberer (in preparation). “Transport of 

pharmaceutically active compounds (PhACs) clofibric acid, diclofenac, and propyphenazone under water 

saturated conditions.” Ground Water, Westerville, OH. 

Scheytt, T., S. Grams, and H. Fell (1998). “Occurrence and behavior of drugs in groundwater.” Gambling with 

groundwater – physical, chemical, and biological aspects of aquifer-stream relations, J.V. Brahana, Y. Eckstein, 

L.K. Ongley, R. Schneider, and J.E. Moore, eds., IAH/AIH Proc., St. Paul., MN. 

Verstraeten, I.M., T. Heberer, and T. Scheytt (2003). “Occurrence, characteristics, transport, and fate of pesticides, 

pharmaceutically active compounds, and industrial products and personal care products at bank-filtration 

sites.” Riverbank Filtration: Improving Source-Water Quality, C. Ray, G. Melin, and R.B. Linsky, eds., 


TRAUGOTT SCHEYTT has been an Associate Professor for Hydrogeology in the 

Department of Civil Engineering and Applied Geosciences at the Technical University 

Berlin since 1996. His teaching duties include courses on groundwater chemistry, 

groundwater flow, and transport of organic contaminants. Current research areas include 

the transport and attenuation of pharmaceutically active compounds, natural attenuation 

of contaminants emanating from landfills, and prediction of transport and degradation of 

organic contaminants in groundwater. Presently, he is on sabbatical at University of 

California, Berkeley, and will stay in California until January 2004 to conduct research on the transport and 

attenuation of pharmaceutically active compounds during riverbank filtration at the Santa Ana River in 

Orange County, California. Scheytt received an M.S. and Ph.D. in Geology from Christian-Albrechts- 

University Kiel in Germany.


The Fate of Bulk Organics and Emerging Contaminants 

During Soil-Aquifer Treatment 

Dr. Jörg E. Drewes 

Colorado School of Mines 

Golden, Colorado 

Dipl.-Ing. Tanja Rauch 

Colorado School of Mines 

Golden, Colorado 

Introduction 

Wastewater reuse employing soil-aquifer treatment is becoming an increasingly important strategy 

for many utilities in the United States and abroad to augment local drinking-water sources where 

supplies are limited. One major water-quality issue associated with soil-aquifer treatment leading 

to indirect potable or nonpotable reuse of wastewater effluents is the fate and transport of organic 

constituents. Effluent-derived bulk organic matter can impair the quality of recovered water by 

interfering with post-treatment processes, such as coagulation, adsorption, or membrane 

treatment. Bulk organic matter is also a known precursor for DBPs. The presence of bioavailable 

organic carbon after soil-aquifer treatment could also increase the microbial regrowth potential in 

distribution systems. In addition, BOM might be comprised of organic micropollutants, which 

survive during soil-aquifer treatment and which are associated with potential adverse human 

health effects. The objective of this study was to investigate the fate and transport of bulk and 

trace organics present in reclaimed water during soil-aquifer treatment leading to indirect potable 

reuse. The American Water Works Association Research Foundation and USEPA funded this study. 

Methodology 

The fate of organics during soil-aquifer treatment was investigated using controlled biodegradation 

studies in adapted soil-columns and full-scale infiltration facilities in Arizona and California. The 

study design followed a watershed-guided approach considering hydraulically corresponding 

samples of drinking-water sources, soil-aquifer treatment-applied wastewater effluents, and 

subsequent post-soil-aquifer treatment samples representing a series of different travel times in the 

subsurface. Extensive characterization of organic carbon in the different samples was performed 

using state-of-the-art analytical techniques (such as size-exclusion chromatography with online 

DOC and UV absorbance detection, carbon-13 nuclear magnetic resonance spectroscopy, Fourier 

transform infrared spectroscopy, and elemental analysis) and biomass/bioactivity measurements 

(dehydrogenase activity; total viable biomass through phospholipid extraction; substrate induced 

respiration) to distinguish between primary removal mechanisms (biodegradation versus adsorption). 

The mechanisms contributing to bulk organic matter removal during initial recharge were 


Dr. Jörg E. Drewes 

Assistant Professor 

Environmental Science & Engineering Division 

Colorado School of Mines • Golden, Colorado 80401-1887 USA 

Phone: (303) 273-3401 • Fax: (303) 273-3413 • Email: jdrewes@mines.edu 

159

160 

identified by isolating three bulk fractions from treated wastewater effluents: 

• Hydrophilic carbon (HPI). 

• Hydrophobic acids (HPO-A). 

• Colloidal organic matter. 

HPI and HPO-A were isolated from a domestic secondary effluent by XAD-8 fractionation. 

Colloidal organic matter was operationally defined as organic matter in the size range of 

approximately 6,000 Dalton to 1 micrometer and was isolated using dialysis diffusion after a preconcentration 

using large volume rotary evaporation. Trace organics selected for this study were 

PhACs and personal care products, as well as selected endocrine disrupting compounds 

(17β-estradiol, estriol, and testosterone). 

Pharmaceutical compounds were selected for this study based on their production, per-capita 

consumption, and occurrence in domestic effluents and surface waters in the United States and 

Middle Europe. Caffeine has been a well-known PhAC of wastewater origin for more than 

20 years. In the category of analgesics/anti-inflammatory drugs, diclofenac, ibuprofen, ketoprofen, 

naproxen, fenoprofen, propyphenazone, meclofenamic acid, and tolfenamic acid were identified. 

Carbamazepine and primidone were identified as antiepileptic drugs. Pentoxifylline represented a 

common blood viscosity-affecting agent. Gemfibrozil, clofibric acid, and fenofibrate were selected 

to represent blood lipid regulators. Organic iodine was used as a surrogate parameter for the 

detection of X-ray contrast agents (triiodinated benzene derivates such as iopromide and 

diatrizoate) and their halogenated metabolites. 

Results 

Data derived from several soil-aquifer treatment field sites consistently indicate a substantial 

removal of DOC during travel through the subsurface. At one site employing tertiary treatment 

prior to recharge, average DOC concentrations of 5.6 mg/L in the tertiary effluent decreased from 

5.6 mg/L to approximately 1.45 mg/L in groundwater monitoring wells after traveling 6 to 12 months 

in the subsurface. At the same time, the specific UV absorbance increased, indicating a preferred 

removal of aliphatic compound groups within bulk organic matter. Further DOC removal 

occurred during long-term travel in the subsurface of 1 to 2 years from 1.45 mg/L to approximately 

1.1 mg/L. At another site employing only secondary treatment without nitrification prior to 

recharge, the DOC concentration was efficiently reduced to approximately 2 mg/L after 

percolating through the vadose zone. The specific UV absorbance increased only slightly from 

1.7 liters per milligram and meter (L/mg m) in the secondary treated effluent to 2.0 L/mg m in 

groundwater samples. These findings pointed to a fast and substantial removal of biodegradable 

organic carbon during the initial phase of groundwater recharge. The results of spectroscopic 

investigations generally demonstrated that naturally derived and effluent-derived organic matter 

after SAT overlap extensively in molecular weight distribution, amount and distribution of 

hydrophobic and hydrophilic carbon fractions, and chemical characteristics based on elemental 

analysis and structural analysis; however, the residual portion of DOC contained both 

effluent-derived organic matter and NOM. 

Results of this study indicate that during vadose zone treatment, HPI is removed by a combination 

of physical and biological mechanisms. HPO-A is least efficiently removed of all fractions during 

soil infiltration. Both adsorption and biotransformation seem to contribute to this removal. Low 

biomass activity responses in soil-columns fed with HPO-A indicated that these substances are 

refractory in nature. The fate of effluent-derived colloidal organic matter during groundwater

echarge is not well understood. Our studies give reason to believe that colloidal organic matter 

can be efficiently removed during initial soil recharge, and that this removal is based on the 

filtration mechanism and biodegradation. Colloidal organic matter stimulated high biomass 

activities, but was not fully mineralized. Remobilization and the breakthrough of colloidal organic 

matter was observed in column experiments fed with colloidal organic matter under higher 

infiltration rates. The removal of colloidal organic matter might, therefore, depend upon hydraulic 

regimes in the aquifer during recharge operation. The majority of biological removal in the HPI 

and HPO-A fractions occurred in the first 30-centimeters of soil infiltration. 

Findings of this study demonstrated that the dominating removal mechanism for steroids during 

soil-aquifer treatment is adsorption, although biodegradation is also taking place and is important 

for a sustainable operation avoiding compound accumulation in the system. The study showed 

that steroid removal was not dependent upon the type of organic background matrix present (HPI, 

HPO-A, colloidal organic matter) or redox and flox conditions (aerobic versus anoxic; saturated 

versus unsaturated). 17β-estradiol, estriol, and testosterone were efficiently removed from initial 

concentrations as high as 500 nanograms per liter after only 5.2 hours of contact with subsurface 

media in the presence of soil microbial activity. 

In addition, the study revealed that the stimulant caffeine, analgesic/anti-inflammatory drugs, and 

blood lipid regulators were efficiently removed to concentrations near or below the detection limit 

of the analytical method after retention times of less than 6 months during groundwater recharge. 

The antiepileptics carbamazepine and primidone were not removed during groundwater recharge 

under either anoxic saturated or aerobic unsaturated flow conditions during travel times of up to 

8 years. Organic iodine showed a partial removal only under anoxic saturated conditions (as 

compared to aerobic conditions) and persisted in recharged groundwater. 

Conclusions 

Findings of this study demonstrated that soil-aquifer treatment represents an efficient treatment 

step to reduce bulk organic carbon and to remove trace pollutants of concern present in reclaimed 

water. 

JÖRG DREWES is an Assistant Professor of Environmental Science and Engineering at 

the Colorado School of Mines. He has been actively involved in research in the area of 

water reclamation, water reuse, and groundwater recharge for more than 11 years. His 

research interests include water and wastewater treatment engineering; potable and nonpotable 

water reuse (soil-aquifer treatment and microfiltration/reverse osmosis); state-ofthe-art 

characterization of natural and effluent organic matter; contaminant transfer 

among environmental media; and the fate of endocrine disrupting compounds and 

pharmaceuticals in natural and engineered systems. Drewes has published more than 50 journal papers, book 

contributions, and conference proceedings. Among his honors, he was awarded the Willy-Hager Award in 

1997, Quentin Mees Research Award in 1999, and Dr. Nevis Cook Excellence in Teaching Award in 2003. 

Drewes received both an M.S. and Ph.D. in Environmental Engineering from the Technical University of 

Berlin, Germany. 

161

162


Ethylenediaminetetraacetic Acid Occurrence and 

Removal Through Bank Filtration in the Platte River, 

Nebraska 

Jason R. Vogel, Ph.D. 



Larry B. Barber, Ph.D. 


Boulder, Colorado 

Tyler B. Coplen, Ph.D. 


Reston, Virginia 

Ingrid M. Verstraeten, Ph.D. 



Thomas F. Speth, Ph.D., P.E. 



Jerry Obrist, P.E. 

City of Lincoln Water System 


The USGS, USEPA, and City of Lincoln Water System (Nebraska) have conducted a study to 

determine the occurrence and removal of EDTA, NTA, and nonylphenol monoethoxycarboxylate 

to nonylphenol pentaethoxycarboxylate (NP1EC-NP5EC) in the hydrologic system at the City of 

Lincoln well field. The objective of the study is to evaluate the occurrence and removal of EDTA, 

NTA, and total nonylphenolpolyethoxycarboxylate (NPEC) by bank filtration at the City of 

Lincoln well field. This presentation will discuss removal during two sampling periods — May and 

August 2002 — based upon surface-water fractions in the collector well calculated using stable 

isotope ratios of hydrogen and oxygen. 

The rationale for selecting compounds evaluated in this study was based on the hierarchical 

analytical approach and includes a range of compounds covering a spectrum of uses and effects. 

For example, EDTA is a low-toxicity, high-production volume chemical used in multiple 

domestic, commercial, and industrial applications to form stable, water-soluble complexes with 

trace metals. Because of its uses and chemical characteristics, EDTA occurs at relatively high 

concentrations and can persist in the aquatic environment (Barber et al., 1996; Barber et al., 2000; 

Leenheer et al., 2001; Barber et al., 2003). 


Jason R. Vogel, Ph.D. 

Hydrologist 


Room 406, Federal Building • 100 Centennial Mall North • Lincoln, Nebraska 68508 USA 

Phone: (402) 437-5129 • Fax: (402) 437-5139 • Email: jrvogel@usgs.gov 

163

164 

Site Description 

There are 40 active production wells at the City of Lincoln well field. Two of these wells are 

horizontal collector wells screened in alluvial sand and gravel approximately 26-m below the 

Platte River; they provide approximately 50 percent of the municipal water during most times of 

the year. The remaining 38 wells are vertical production wells that are developed in alluvial 

sediments mainly consisting of sand and gravel. At this location, the quality of the river water has 

a large effect on the quality of the bank-filtered water obtained from the collector wells. Similarly, 

the water quality of the groundwater in the vertical production wells directly corresponds to the 

distances of the wells from the river. Because of the direct link between the collector wells and 

river, the collector well is usually turned off during the month of May and first part of June to avoid 

the flush of herbicides associated with spring planting in this agricultural area. After collection, 

well water is treated by ozonation, filtration, and chlorination before distribution. 

Sampling 

Representative surface-water samples were collected quarterly from the Platte River at the wellfield 

site using equal width-increment, flow-weighted sampling. Groundwater samples were also 

collected quarterly from one of the collector wells and from two of the vertical groundwater wells. 

One groundwater well was located relatively close to the river (within 100 m), with another 

located away from the river (1,000 m). All samples were filtered through 0.7-m glass fiber filters 

and collected in pre-cleaned amber glass bottles. Samples for EDTA, NTA, and NPEC analyses 

were preserved with 2-percent by volume formalin. This presentation will discuss the results of 

samples from May and August 2002. 

Analysis 

EDTA, NTA, and NP1EC-NP5EC were measured using a modification (Barber et al., 2000) of 

the method of Schaffner and Giger (1984). Samples (100 milliliters) were evaporated to dryness, 

acidified with 5-milliliter 50-percent (volume per volume) formic acid/distilled water, and evaporated 

to dryness. Acetyl chloride/propanol (10-percent volume per volume) was added, the sample 

heated at 90-degrees Celsius for 1 hour to form the propyl-esters, and the propyl-esters were 

extracted into chloroform. The chloroform extracts were evaporated to dryness and re-dissolved 

in toluene for analysis by GC/MS, as described below. 

The propyl-ester extracts were analyzed by electron-impact GC/MS in both the full-scan and 

select ion monitoring modes. The general GC conditions were: 

• Hewlett Packard (HP) 6890 GC. 

• Column-HP Ultra II (5-percent phenylmethyl silicone), 25 m × 0.2 millimeters, 

33-micrometer film thickness. 

• Carrier gas, ultra-high purity helium, with a linear-flow velocity of 27 centimeters per second. 

• Injection port temperature, 300-degrees Celsius. 

• Initial oven temperature, 50-degrees Celsius. 

• Split vent open, 0.75 minutes. 

• Ramp rate, 6-degrees Celsius per minute to 300-degrees Celsius. 

• Hold time, 15 minutes at 300-degrees Celsius.

The MS conditions were as follows: 

• HP 5973 Mass Selective Detector. 

• Tune with perflurotributylamine. 

• Ionization energy, 70 electron volt (eV). 

• Source pressure, 1 × 10 –5 torr. 

• Source temperature, 250-degrees Celsius. 

• Interface temperature, 280-degrees Celsius. 

• Full scan, 40 to 550 atomic mass units at one scan per second. 

Concentrations were calculated based on select ion monitoring data using diagnostic ions for each 

compound. Each compound was identified based on the matching of retention times (±0.02 minutes) 

and ion ratios (±20 percent) determined from the analysis of authentic standards. An eight-point 

calibration curve (typically ranging from 0.01 to 50 nanograms per microliter) and internal 

standard procedures were used for calculating concentrations. Surrogate standards were added to 

the samples prior to extraction and derivatization to evaluate compound recovery and whole 

method performance. 

Results 

In general, based upon results from May and August 2002, measured EDTA concentrations for 

surface-water samples were larger than for groundwater samples. In addition, EDTA concentrations 

decreased and total NPEC concentrations increased in the groundwater as the distance of 

the well from the river increased. NTA was only detected in one surface-water sample at very low 

levels and not at all in groundwater samples. 

Using surface-water fractions in the collector well determined from deuterium and oxygen-18 ratios, 

the transport of EDTA was nearly conservative during these two sampling periods. Total NPEC 

concentrations were lower than predicted. Further analysis will be forthcoming in the presentation. 

REFERENCES 

Barber, L.B., G.K. Brown, and S.D. Zaugg (2000). “Potential endocrine disrupting organic chemicals in 

treated municipal wastewater and river water.” Analysis of Environmental Endocrine Disruptors, L.H. Keith, 

T.L. Jones-Lepp, and L.L. Needham, eds., American Chemical Society Symposium Series 747, American 

Chemical Society, Washington, DC, p. 97-123. 

Barber, L.B., E.T. Furlong, S.H. Keefe, G.K. Brown, and J.D. Cahill (2003). “Natural and Contaminant 

Organic Compounds in Boulder Creek, Colorado under High-Flow and Low-Flow Conditions, 2000.” 

Comprehensive water quality of the Boulder Creek Watershed, Colorado, under high-flow and low-flow conditions, 

S.F. Murphy, L.B. Barber, and P.L. Verplanck, eds., U.S. Geological Survey Water Resources Investigations 

Report 03-4045. 

Barber, L.B., J.A. Leenheer, W.E. Pereira, T.I. Noyes, G.A. Brown, C.F. Tabor, and J.H. Writer (1996). 

“Organic contamination of the Mississippi River from municipal and industrial wastewater.” U.S. Geological 

Survey, Circular 1133, p. 114-135. 

Leenheer, J.A., C.E. Rostad, L.B. Barber, R.A. Schroeder, R. Anders, and M.L. Davisson (2001). “Nature and 

chlorine reactivity of organic constituents from reclaimed water in groundwater, Los Angeles County, 

California.” Environmental Science and Technology, 35: 3,869-3,876. 

Schaffner, C., and W. Giger (1984). “Determination of nitrilotriacetic acid in water by high-resolution gas 

chromatography.” Journal of Chromatography, 312: 413-421. 

165

166 

Hydrologist JASON VOGEL has been with the United States Geological Survey in 

Lincoln, Nebraska, for a little over a year. During that time, he has been the Project Chief 

for a bank-filtration study in cooperation with the United States Environmental Protection 

Agency and the City of Lincoln Water System, and is also Lead Scientist of the agricultural 

chemical transport team in the Nebraska District. Before joining the United States 

Geological Survey, Vogel was a research engineer in the Biosystems Engineering Department 

at Oklahoma State University for 5 years. He has published articles on a wide variety of 

topics, including geostatistics and stochastic design, vadose zone transport, and microbial transport in 

groundwater, and co-authored the chapter on Geostatistics in Statistical Methods in Hydrology, Second Edition, 

with C.T. Haan. Vogel received a B.S. in Biological Systems Engineering from the University of Nebraska, 

an M.S. in Agricultural Engineering from Texas A&M University, and Ph.D. in Biosystems Engineering from 

Oklahoma State University.

Session 9: Public Policy and Regulatory 

Riverbank Filtration as a Regional Supply Option 

for the United States 

Leo Gentile, P.G., CPG 

Jordan Jones & Goulding 


David Haas, P.E. 

Jordan Jones & Goulding 


D. Joseph Hagerty, Ph.D., P.E. 

University of Louisville, Kentucky 


Peggy H. Duffy, P.E. 

Hagerty Engineering 

Jeffersonville, Indiana 

Introduction 

RBF has been used on the lower Rhine River for over 130 years and, today, comprises 15 to 20 percent 

of the water supply in Germany as a whole (Schubert, 2000). In the United States, the majority 

of larger water systems (serving 10,000 or more) use surface water (Wang et al., 2002). Smaller 

systems supplying 10,000 or less often rely on groundwater for supply because of reduced treatment 

requirements. RBF is employed in communities in the United States such as Cedar Rapids, Iowa; 

Lincoln, Nebraska; Louisville, Kentucky; and Sonoma County, California. Some communities 

such as St. Helens, Oregon, and Sioux Falls, South Dakota, essentially use RBF by drawing water 

from collector wells constructed adjacent to rivers, although this is usually recognized as GWUDI 

(Wang et al., 2002). Still, RBF is an underutilized supply whose benefits are likely available to 

more communities to help provide better and more uniform raw-water quality. RBF potential is 

greatest for communities in the Midwestern United States that are located where glacial valleyfill 

aquifers occur that are interconnected to adjacent surface-water bodies. 

Conditions 

The need for water is the basic condition for RBF or any other supply source. The magnitude of 

the demand and physical conditions dictates what supply source is most feasible. If a community 

is located along a major river or stream, then surface-water intake is a relatively simple solution; 

however, when water treatment issues and costs are considered, then groundwater — which may 

require no more treatment than disinfection — is the next logical choice. RBF is an alternative 


Leo Gentile, P.G., CPG 

Senior Project Hydrogeologist 


6801 Governors Lake Parkway • Norcross, Georgia 30071 USA 

Phone: (678) 333-0148 • Fax: (770) 455-7391 • Email: LGENTILE@JJG.com 

167

168 

supply option when these conditions are not available that provides typically larger quantities 

than wells, and more uniform and higher quality raw water compared to surface-water intake. 

Aquifer Thickness and Extent: To supply the millions of gallons a day needed for a typical 

municipal water supply, the aquifer needs to be sufficiently thick and extensive to sustain such a 

yield without depleting the aquifer. If the aquifer is thin (less than about 100-ft thick), the cone 

of depression and amount of drawdown from wells penetrating the aquifer will be narrow and 

steep, and the well will likely not be efficient in producing the required yields. Excessive 

drawdown indicates an inefficient well and/or an over-drafted aquifer. Constructing collector wells 

with multiple horizontal lateral collector galleries will help increase well efficiency and create a 

broader cone of depression with less drawdown. Still, if the aquifer is relatively thin and limited 

in aerial extent, the sustained (or safe yield) may not be sufficient to meet demand. With RBF, 

conventionally constructed wells, collector wells, or specially designed tunnels overcome the 

physical shortcomings of the aquifer. The leakance from an adjacent river or lake recharges the 

aquifer, augmenting the safe yield (Figure 1). Thus, a relatively thin and aerially limited aquifer 

can yield the water quantities needed for municipal supplies. RBF also yields water of more 

consistent and better overall quality. 

River or 

Recharge Pond 

Water Table 

A B 

NOT TO SCALE 

River or 

Recharge Pond 

NOT TO SCALE 

Pumping Well 


Modified from Gallaher and Prize, 1966 

EXPLANATION 

Figure 1. Groundwater flow recharge to a river (A) and an RBF well (B) in a sand and gravel aquifer (from 

Lloyd and Lyke, 1995). 

Aquifer Characteristics: Glacial and alluvial sand and gravel deposits are some of the most 

productive aquifers in the world (Todd, 1980). The hydraulic conductivities and amount of water 

in storage are often high. The recharge potential from adjacent streams, rivers, or lakes is also 

considered highly favorable as natural or induced vertical gradients, coupled with good hydraulic 

conductivity, allows for leakance from the surface-water body to the aquifer. Pleistocene-age sand 

and gravel deposits have been used for well fields and RBF sites. By comparison, high hydraulic 

conductivity and yields can also occur in bedrock aquifers with fracture or karst porosity; however, 

storage capacity may be limited. Furthermore, groundwater flow through fractures or karst porosity 

is essentially conduit flow. Recharge from adjacent rivers or lakes would be minimally filtered, 

reducing the benefit from RBF. 

Other Considerations: There must be effective interconnection between the aquifer and surface-water 

body. Occasionally, an aquifer is confined or semi-confined by a layer of finer sediments that 

reduces the potential for leakance from the surface to the aquifer (Figure 2). RBF sites are more 

efficient where there is good interconnection (over 90 percent of withdrawn water can be bank 

filtrate [Eckert et al., 2000]). Clogging of the interconnection through RBF use, or the natural 

accumulation of fines, also reduces the efficacy of RBF. The removal of streambed fines by water 

flow from naturally high-gradient streams and rivers or during floods will help restore infiltration.

FEET 

1,000 

900 

800 

700 

600 

VERTICAL SCALE GREATLY EXAGGERATED 

DATUM IS SEA LEVEL 

Modified from Gann and others, 1973 

0 1 2 MILES 

Figure 2. Glacial valley-fill aquifer overlain by fine grained sediments in the Great Miami River Valley, Ohio 

(from Miller and Appel, 1997). 

German researchers experimented with excavating a “window” by dredging accumulated 

sediment. While the measure increased RBF efficiency, it was considered temporary, lasting only 

weeks before silting over (Schubert, 2000). Rivers whose flow has been altered for flood control 

and navigation improvement, such as the Ohio River, are particularly prone to clogging. Steep 

gradients and floodwaters that may have scoured the riverbed in the past have been eliminated. 

Wells or specially designed tunnels used to extract groundwater and bank filtrate must also be 

properly located to maximize system efficiency. Rivers and streams that occur with the thumbprint 

of the most recent continental glacial period in the United States (e.g., Wisconsin ice age) have 

a similar terraced profile. The unaltered modern river channel carries normal stage flows within 

the primary terrace. This terrace corresponds to a 100-year flood plain. Where significantly altered 

for flood control or navigation, the river may now have inundated the primary terrace extending 

from bank to bank up to the secondary terrace. The secondary terrace extends outward to the 

500-year flood plain. This channel was cut into the bedrock by glacially fed rivers. The rivers were 

large braided streams laden with sand and gravel. As the glacial outwash diminished, the scour 

channels filled with sand and gravel to become aquifers. The tertiary terrace is the remnant of 

peak stage flooding and its resultant scouring into the bedrock. Relatively thin and finer grained 

sediments were deposited as the flood stage fell (see Figure 2). These sediments may have formed 

local, relatively less-productive aquifers that are not connected to the modern river channel. The 

secondary terrace is the most productive place to locate RBF wells (e.g., out of main navigation, 

interconnected to the river, relatively thicker aquifer, greater potential for effective filtration). 

Potential RBF Regions 

Grand River 

B 

0 1 2 KILOMETERS 

The greatest potential for RBF — and, in fact, where RBF has been implemented — in the United 

States is in the Midwestern states. Figure 3 shows the location of mid- to large-sized communities 

in the United States that are located along rivers and on alluvial aquifers. Due to site conditions, 

RBF is a viable water supply option for many, but not all, communities. 

Midwest stream-valley aquifers in Midwestern states west of the Mississippi River are important 

sources of water for communities and industries (Figure 4). These aquifers are unconsolidated sand 

and gravel deposits that are thicker, greater in extent, and more productive in the major river 

valleys of the region. The stream-valley aquifers are unconfined and in direct connection to adjacent 

streams and rivers. Average yields from wells range from 100 to 1,000 gallons per minute, with some 

wells yielding over 3,000 gallons per minute (Olcott, 1992). RBF opportunities may be available 

Buried Valley 

B 

EXPLANATION 

Clay, silt, and fine-grained sand 

Coarse-grained sand and gravel 

Sand, gravel, and boulders 

Till 

Bedrock 

169

170 

EXPLANATION 

Dune Sand 

NOT TO SCALE 

Recharge 

Discharge 

South 

N 

Figure 3. Potential RBF areas in the United States (from USGS, 1965). 

Evaporation 

Arkansas 

River 

Evapotranspiration 


Stream to 

Aquifer Leakage 

Aquifer 

Bedrock 

East 

using stream-valley aquifers that are located along major rivers such as the Arkansas, Des Moines, 

Grand (Michigan), Mississippi, Platte, Missouri, and Wisconsin rivers, plus numerous mediumsized 

streams that course the region (Miller and Appel, 1997). 

West 

Precipitation 

Canal Diversion System 

Irrigation 

Well 

Crop Consumptive Use 

Deep Percolation 

of Precipitation and 

Applied Irrigation Water 

Pumpage 

Modified from Barker and Others, 1983 

North 

Figure 4. Midwestern stream valley site favorable for RBF (Arkansas River Valley) (from Miller and Appel, 

1997). 

Legend 

Metropolitan Areas with 

Bank Infiltration Potential 

Population 250,000–1,000,000 

Population 1,000,001–2,500,000 

Population greater than 2,500,000 

Unconsolidated Aquifers 

Alluvial aquifers suitable for bank infiltration 

Sand and gravel 

Consolidated Rock Aquifers 

Sandstone: includes some unconsolidated sand 

Carbonate rock: limestone and dolomite; 

and in Texas and Oklahoma, some gypsum 

Sandstone and carbonate rocks 

Volcanic rocks, chiefly basalt 

Crystalline rocks, igneous and metamorphic 

Withdrawals from wells

The surficial aquifers in Midwestern states that border the Ohio River and are east of the 

Mississippi River are similar to those described previously, but are primarily of glacial origin. These 

aquifers are very productive (1,000 gallons per minute) and supply almost 50 percent of the fresh 

groundwater produced in the region. The course-grained aquifers are divided into two categories: 

• Deposits at or near the land surface occurring in stream and river valleys. 

• Deposits buried by a layer of fine-grained material that occur in former river valleys cut 

into bedrock and filled with coarse-grained glacial outwash. 

The sands and gravels range in thickness from less than 100 to over 600 ft in some buried bedrock 

valleys. Large yields are possible from wells completed in the glacial outwash aquifers that are 

hydraulically connected to streams, rivers, and lakes, and the wells are sufficiently close to the surfacewater 

body (Lloyd and Lyke, 1995). Communities located near rivers such as the Illinois, Kaskaski, 

Wabash, White, Kankakee, Maumee, Great Miami, and Scioto rivers could benefit from RBF. 

Valley-fill glacial aquifers also occur in the Northeastern United States. While extensive, the 

aquifer thickness and productivity is less than those occurring in Midwestern states (Olcott, 

1995). In general, the valley-fill aquifers are less common along major streams and rivers, so the 

potential for RBF is considered lower than in the Midwestern United States. Alluvial aquifers 

occur along streams and rivers in the Northwestern states, but sparse population and relatively low 

water demands can be met through either surface-water intakes or groundwater wells. In the arid 

Southwest, few major rivers or streams are perennial, thereby limiting RBF potential. Water 

supplies in Gulf Coast states are met through groundwater withdrawals from productive regional 

aquifers and surface water. In the Piedmont region, water needs are met through surface water, as 

rainfall is usually adequate to meet demands and aquifers are fractured crystalline rock, which are 

not considered viable RBF areas. 

REFERENCES 

Eckert, P., C. Blomer, J. Gothardt, S. Kamphausen, D. Liebich, and J. Schubert (2000). “Correlation between 

the Well Field Catchment and Transient Flow Conditions.” Proceedings, International Riverbank Filtration 

Conference, November 2 - 4, Dusseldorf, Germany. 

Lloyd, O.B., and W.L. Lyke (1995). “Ground Water Atlas of the United States, Segment 10 - Indiana, 

Illinois, Kentucky, Ohio.” U.S. Geological Survey Hydrologic Investigations Atlas 730 -K. 

Miller, J.A., and C.L. Appel (1997). “Ground Water Atlas of the United States, Segment 3- Kansas, 

Missouri, Nebraska.” U.S. Geological Survey Hydrologic Investigations Atlas 730 -D. 

Olcott, P.G. (1995). “Ground Water Atlas of the United States, Segment 12 - Connecticut, Maine, 

Massachusetts, New Hampshire, New York, Rhode Island, Vermont.” U.S. Geological Survey Hydrologic 

Investigations Atlas 730 -J. 

Olcott, P.G. (1992). “Ground Water Atlas of the United States, Segment 12 - Iowa, Michigan, Minnesota, 

Wisconsin.” U.S. Geological Survey Hydrologic Investigations Atlas 730 -M. 

Schubert, J. (2000). “How Does it Work: Field Studies on Riverbank Filtration.” Proceedings, International 

Riverbank Filtration Conference, November 2 - 4, Dusseldorf, Germany. 

Todd, D.K. (1980). Groundwater Hydrology, John Wiley & Sons, New York. 

United States Geological Survey (1965). Productive Aquifers in the Conterminous United States, United States 

Geological Survey. 


Process, American Water Works Association Research Foundation and American Water Works Association. 

171

172 

LEO GENTILE is Senior Hydrogeologist and Group Manager with Jordan, Jones & 

Goulding in Atlanta, Georgia. He has a broad range of experience in applied geology and 

hydrogeology in the United States. His background includes over 20-years of experience 

in technical support and management on a broad range of projects, including waterresources 

development, remedial investigations, and feasibility studies at Comprehensive 

Environmental Resource Compensation and Liability Act sites; Resource Conservation 

and Recovery Act facility investigation and closure; investigation and remediation of 

agricultural chemical sites; and assessments and reclamation for mining and petroleum-related sites. Gentile 

received a B.S. in Geology/Mineralogy from Ohio State University, an M.S. in Petroleum Geology from 

Oklahoma State University, and an MBA in Enterprise Risk Management/Finance from Georgia State 

University. He is a Registered Professional Geologist in seven states and Puerto Rico, and is a Certified 

Professional Geologist by the American Institute of Professional Geologists.


Application of the Long Term 2 Enhanced Surface 

Water Treatment Rule Microbial Toolbox at Existing 

Water Plants 

Richard A. Brown 

Environmental Engineering and Technology, Inc. 

Newport News, Virginia 

All surface-water utilities in the United States will be required to comply with specific Cryptosporidium 

removal/inactivation targets in the LT2ESWTR, depending upon their bin classification as a 

result of raw-water Cryptosporidium occurrence levels. The “microbial toolbox” is intended to 

provide utilities with a range of treatment options for meeting LT2ESWTR compliance 

requirements. 

This presentation will include a general discussion of LT2ESWTR requirements, including two of 

its main components: 

• Cryptosporidium sampling. 

• Treatment credits from the microbial toolbox. 

In particular, this will include a discussion of how these requirements relate to systems with RBF 

or other groundwater wells that are GWUDI. LT2ESWTR requirements for RBF are different for 

systems in place at the time the Rule is published versus well systems installed later. New systems 

are eligible for treatment credits if they meet USEPA-defined requirements for media gradation, 

separation distance from the associated surface-water source, and turbidity monitoring. Existing 

RBF and GWUDI wells are not eligible for any direct credits, but will produce indirect credits if 

Cryptosporidium bin assignment samples are collected from well effluent. According to the 

LT2ESWTR, these Cryptosporidium samples must be collected from well effluent if well water is 

sent to a subsequent treatment process before being delivered to consumers. Public water-supply 

GWUDI or RBF wells that provide water directly to customers without subsequent physical 

treatment must collect Cryptosporidium bin assignment samples from the surface-water source. 

Additional benefits of RBF (removal or degradation of other contaminants and microorganisms, 

dampening of water quality and temperature spikes, etc.), as well as potential issues of concern 

(clogging of pores, scour associated with flooding/high stage events), will also be outlined. 

Other microbial toolbox alternatives will be outlined, though in less detail, and cost comparisons 

for different treatment strategies will also be discussed. Strategies for source-water sampling and 

the preparation of contingency plans will also be addressed. 


Richard A. Brown 

Engineer 

Environmental Engineering and Technology 

712 Gum Rock Court • Newport News, Virginia 23606 USA 

Phone: (757) 873-1534 • Fax: (757) 873-2392 • Email: rbrown@EETinc.com 

173

174 

RICHARD BROWN, an Environmental Engineer at Environmental Engineering and 

Technology, Inc. in Newport News, Virginia, has almost 20 years of experience working 

on surface-water and groundwater quality, regulatory compliance, and water-treatment 

projects. Recently, this has included extensive involvement with the American Water 

Works Association, Association of Metropolitan Water Agencies, and various United 

States drinking-water utilities during negotiations with the United States Environmental 

Protection Agency regarding the Stage 2 Disinfection By-Products Rule and Long Term 2 

Enhanced Surface Water Treatment Plant. Prior, Brown worked for 11 years at the Los Angeles County 

Sanitation Districts, performing and directing numerous projects involving the investigation and 

interpretation of groundwater chemistry at six municipal solid waste landfills. The American Water Works 

Association Research Foundation study that is the subject of Brown’s presentation during the Second 

International Riverbank Filtration Conference will be published in the American Water Works Association’s 

e-Journal in September 2003. Brown received a B.S. in Civil Engineering from Purdue University and an 

M.S. in Environmental Engineering from the University of North Carolina at Chapel Hill.


Draft Protocol for the Demonstration of 

Effective Riverbank Filtration 




Verna J. Arnette, P.E. 



Bruce L. Whitteberry 



Introduction 

RBF is a natural in situ filtration process by which microbial pathogens in surface water are 

removed via the porous media of the streambed and aquifer under induced infiltration conditions 

created by a pumping well. With the promulgation of the LT2ESWTR, the USEPA will allow 

0.5- and 1.0-log treatment credit for Cryptosporidium removal using RBF (USEPA, 2003). 

Based upon a review of data from several RBF systems in the United States, it is not uncommon 

for RBF to achieve a removal performance better than the proposed 1.0-log reduction credit. The 

0.5- and 1.0-log credits are based on general conservative requirements obtainable by most RBF 

systems; however, there are still several concerns with the effectiveness of RBF during worst-case 

hydrologic and water-quality conditions. For example, high flow events may significantly scour a 

streambed, thus reducing its filtration capabilities. To date, there has been a reluctance to provide 

additional credit because an acceptable procedure has not been developed that will allow for the 

demonstration of consistent removal during these periods of high risk. 

The Greater Cincinnati Water Works has developed the following draft protocol for demonstrating 

RBF pathogen removal. The draft protocol will: 

• Allow utilities a method arguing for the re-designation of a groundwater from GWUDI 

back to “groundwater.” 

• Allow for demonstration of 2.0-log Cryptosporidium removal in lieu of engineered filtration 

under the Interim Enhanced Surface Water Treatment Rule. 

• Allow for greater than 1.0-log treatment credit under LT2ESWTR. 

• Provide consistency for evaluating RBF sites. 

The draft protocol is based upon a review of several regulatory guidance publications dealing with 

GWUDI evaluations and “alternative filtration technology”(USEPA, 1991; USEPA, 1992; 



Supervisor of Treatment, Water Quality & Treatment Division 



Phone: (513) 624-5657 • Fax: (513) 624-5670 • Email: William.Gollnitz@gcww.cincinnati-oh.gov 

175

176 

Wilson et al., 1996; USEPA, 2001). In addition, the Greater Cincinnati Water Works has 

reviewed data from an extensive flowpath study conducted in cooperation with the USGS at the 

Greater Cincinnati Water Works’ Charles M. Bolton Well Field (Sheets et al., 2002; Gollnitz et 

al., 2003), as well as published data from other RBF sites such as Louisville, Kentucky (Wang et 

al., 2002); Casper, Wyoming (Gollnitz et al., 1997); and others (Cote et al., 2002). The protocol 

is in response, in part, to requests for comments on additional RBF credit in the LT2ESWTR. It 

is hoped that the protocol will provide initial guidance to the USEPA, primacy agencies, and 

water utilities. The draft format of the protocol will facilitate comment and modification by 

regulatory and utility personnel. 

The RBF process is very complex and is still not fully understood; however, our knowledge to date 

does pinpoint several controlling factors that are primary with respect to the removal of microbial 

pathogens and surrogates. These factors are identified in Cote et al. (2002) and include: 

• Assessment of river and groundwater quality. 

• Flow velocity through the streambed and aquifer. 

• Dilution with regional groundwater. 

The RBF protocol has, therefore, been designed around: 

• Characterizing the aquifer and its capability as a RBF system. 

• Identifying conditions for testing during periods of high flow velocity. 

• Testing during periods of minimal dilution with regional groundwater. 

The latter two items involve identifying periods of maximum induced filtration during high stage 

events (as related to storm or reservoir release events) and maximum groundwater pumping. Theoretically, 

these two factors would correspond with periods of high velocity and minimum dilution. 

Pre-Demonstration Evaluation 

The purpose of the pre-demonstration evaluation is to characterize the aquifer and collection 

devices with respect to hydrology and water quality. This includes a determination of the worst-case 

conditions for minimal natural filtration of pathogenic protozoa. This evaluation will then identify 

the type of demonstration project needed for determining treatment credit. The pre-demonstration 

evaluation will organize and evaluate existing data and will identify what additional data is needed 

to complete this first phase. The pre-demonstration project will specifically look at the following 

important aspects of the natural filtration process: 

1. Surface-water and groundwater quality. 

• Filtration capability. 

• Contaminant and surrogate concentrations. 

• Continuous monitoring of turbidity. 

2. Potential infiltration rate variability. 

3. Time of travel between surface-water and groundwater collection devices. 

The first aspect of Item 1 is to characterize source-water quality with respect to its filtration 

capability. It is recommended that the utility perform analyses to determine the ionic strength of 

both surface water and groundwater. The ionic strength may provide an indication of the filtration 

capability of the sediments. Cote et al. (2002) and others have suggested that a high ionic strength 

improves the removal of particles; however, its role is not yet fully understood. It should be noted

that multiple samples should be evaluated. Ionic strength may vary between low (primarily 

groundwater inflow) and high (primarily runoff) river flow conditions. 

The second aspect of Item 1 is to determine the level of contaminant (Giardia, Cryptosporidium) 

and surrogate (algae, diatoms, spores, particle counts) concentrations and their seasonal 

variability. If this data does not already exist, it is recommended that the utility perform monthly 

Giardia/ Cryptosporidium monitoring using Method 1623 (the same as required under 

LT2ESWTR), monthly Microscopic Particulate Analysis using the USEPA Consensus Method 

(1992), and weekly endospore analysis (Rice, 1996). Another optional surrogate parameter is 

particle counts. Particle counts should be in the two size ranges for Giardia (7 to 10 micrometers) 

and Cryptosporidium (3 to 5 micrometers). Data collection for all parameters selected should be 

representative of one full hydrologic cycle (e.g., 1 year). The concentrations of pathogens, 

primarily Cryptosporidium, will determine the level of treatment needed. Because Giardia and 

Cryptosporidium concentrations are typically low in surface waters, combined with detection 

limitations during sample analysis, they cannot be used to determine log-reduction credit. Log 

reductions should be based upon levels of algae, diatoms (from Microscopic Particulate Analysis), 

endospores, and particle counts; however, prior to performing the demonstration project, 

measurable concentrations need to be established. Particle counts may be used; however, they 

should be considered secondary due to the fact that they detect sediment or natural microbial 

growth (e.g., iron bacteria) in groundwater from wells. Typically, log reductions using particle 

counts are lower as compared to algae and spores due to the measurement of other inert and 

microbial particles. 

The third aspect of Item 1 is the continuous monitoring of turbidity. Turbidity monitoring is a 

requirement for engineered filtration systems, and has been included as a requirement on each 

collection device for RBF credit under LT2ESWTR for the period that the device is in use. 

Groundwater from RBF sites should be consistently low (0.01 to 1.0 ntu). Significant excursions 

above the normal level, especially after high infiltration periods, should be investigated. As with 

particle counts, turbidity also measures inert geologic material and other non-pathogenic microbials. 

As part of the pre-demonstration phase, it is important to estimate the range of potential unit 

infiltration rates out in the active area of the streambed, articularly to estimate the highest value 

under conditions of high-river stage and high groundwater pumping during periods when river 

water is warm. Typically, infiltration rates at RBF sites are lower than slow sand filters, but should 

be investigated to see if they are higher. Infiltration rates can be calculated using Darcy’s Law, with 

input variables being river stage, groundwater elevations below the river stage, streambed permeability, 

streambed thickness, and surface-water temperature (affecting viscosity) (Gollnitz et al., 

2002). River-stage data may be obtained from a local United States geological gaging station. If 

this data is not available, the maximum stage elevation may be estimated from the depth of the 

river channel. The groundwater elevation should be measured below the active infiltration area 

of the streambed using piezometers in the streambed or monitoring wells on each side of the river. 

Streambed permeability can be estimated using pump test analysis (preferred, but costly), seepage 

meters, temperature probes, and piezometers (each method has inherent problems and should be 

considered carefully before selection). Streambed thickness can be estimated by visually 

inspecting sediments from shallow excavations of the streambed during low flow. Temperature can 

be measured directly in surface water. Surface-water temperature typically reflects the average 

daily ambient air temperature. If this data is not available, it may be necessary to collect it during 

the pre-demonstration period. Another important aspect is to look at the frequency of the 

occurrence of high infiltration events. Frequency analysis requires long-term river stage data, 

which may be limited for most RBF sites. 

177

178 

Water-quality monitoring during the demonstration project should consider the travel time 

between the river and collection device. Travel time should be calculated for the shortest flowpath 

(e.g., fastest potential time of travel) during continuous pumping of the collection device. Time of 

travel can be estimated using various techniques ranging from tracer studies (conductivity, 

temperature, and chloride), analytical flow calculations, and groundwater flow modeling. Each 

method has inherited limitations that must be considered. It is recommended that the utility 

determine a sampling window based upon a realistic range of travel times using two or more methods. 

Another important consideration is the travel time of particles. Larger particles such as algae may 

have longer travel times due to retardation in the aquifer. Sampling windows should be long 

enough to take these into consideration. For example, Table 1 lists various travel-time estimates 

for Production Well 1 at the Charles M. Bolton Well Field, including limitations. The fastest 

theoretical time is 1 day; however, a more accurate estimate using conductivity indicates that the 

travel time of groundwater ranges from 10 to 14 days (Sheets et al., 2002). Interestingly, algae 

appear to arrive at the production well much later (approximately 38 days). In this case, a high 

infiltration-event monitoring sampling window may be from 1 to 60 days after the event peak. It 

is recommended that 10 to 15 samples be collected during this time period, with increased 

sampling frequency around the anticipated period for possible breakthrough of microbial surrogates. 

Table 1. Time of Travel Estimates between the Great Miami River and Production Well 1 

at the Charles M. Bolton Well Field 

TOT 

Methodology (Days) Comments 

Calculated Fixed Radius 1 Short TOT; considers only horizontal flowpath; 

no gradient 

Groundwater Flow Model 8 Limitations with model design and scale 

Conductivity 10 to 14 Real time measurement; range based upon multiple 

measurements; considered most accurate method 

Event Temperature Lag 26 Real-time measurement; problems due to 

thermodynamics 

Algae Event Peak 11 to 38 Limited data; considers flow characteristics of algae 

TOT = Time of travel. CMB = Charles M. Bolton Well Field. 

Demonstration Evaluation 

The demonstration evaluation is broken down into three options: 1-year monitoring, multi-event 

monitoring, or seasonal monitoring. One of these options is selected based upon the results of the 

pre-demonstration evaluation. Log reduction for all three options should be based upon the 

average of Microscopic Particulate Analysis and endospore data. 

The 1-year monitoring option is for systems with little or limited infiltration rate and water-quality data. 

It is recommended that utilities monitor river stage, groundwater elevations, and surface-water 

temperature on a daily basis. Daily calculations can be made with Darcy’s law using conservative 

values of permeability (e.g., highest measured value) and thickness (lowest measured value). 

These calculations can be easily made using a computer spreadsheet (Gollnitz et al., 2002). With 

respect to water quality, it is recommended that utilities collect weekly samples of turbidity and 

endospores from surface water, along with monthly Giardia, Cryptosporidium, and Microscopic 

Particulate Analysis. Each collection device being evaluated should be continuously monitored for

turbidity. Endospores, Giardia, Cryptosporidium, and Microscopic Particulate Analysis data should 

be sampled at each collection device on a period and frequency determined from the time-of-travel 

data. During the 1-year period, an attempt should be made to identify conditions for high 

infiltration and to monitor one or more events, if possible. 

The multi-event monitoring option is provided for systems with infiltration rates that are 

significantly high, as compared to engineered filtration systems, during storm events, upstream 

reservoir releases, and during periods of heavy groundwater pumping. Water-quality monitoring 

should include multiple turbidity, Giardia, Cryptosporidium, Microscopic Particulate Analysis, and 

endospore data collected from the river during and shortly after the river stage peak. As with all 

options, groundwater turbidity should be continuously monitored. Groundwater samples for the 

other parameters should be collected within a defined sampling window using the time-of-travel 

data. Endospores and particle counts should be collected more frequently due to their lower cost. 

A major advantage to this option is that monitoring resources can be concentrated during these 

periods, which in turn may reduce overall costs. Infiltration rate parameters should be monitored 

to determine the magnitude of the event. 

Option three is seasonal monitoring. Pre-demonstration evaluations may identify if a system is at 

risk during a 2- to 4-month period of the year. For example, a controlled river in the arid West 

may have a high stage only during the summer months, when water is released from reservoirs for 

irrigation. Water-quality monitoring should be concentrated during this high-risk period, taking 

time of travel into consideration, as well as estimating the magnitude of infiltration. 

REFERENCES 

Cote, M.M., M.B. Emelko, and N.R.Thomson (2002). “Factors Influencing Prediction of Cryptosporidium 

Removal in Riverbank Filtration Systems: Focus on Filtration.” Proceedings, American Water Works Association 

WQTC Conference, Denver, CO. 

Gollnitz, W.D., J.L. Clancy, and S.C. Garner (1997). “Reduction of Microscopic Particulates by Aquifers.” 

Journal AWWA, 89(11). 

Gollnitz, W.D., J.L. Clancy, B.L. Whitteberry, and J.A. Vogt (2003). “Riverbank Filtration as a Microbial 

Treatment Process.” Journal AWWA (in press). 

Gollnitz, W.D., B.L. Whitteberry, and J.A. Vogt (2002). “Induced Infiltration Rate Variability and Water 

Quality.” Proceedings, SW-GW Interactions Conference, American Water Resource Association, Keystone, CO. 

Rice, E.W., K.R. Fox, R.J. Miltner, D.A. Lytle, and C.H. Johnson (1996). “Evaluating Plant Performance 

with Endospores.” Journal AWWA, Denver, CO. 

Sheets, R.A., R.A. Darner, and B.L. Whitteberry (2002). “Lag Times of Bank Filtration at a Well Field, 

Cincinnati, Ohio, USA.” Journal of Hydrology, 233: 162-174. 

USEPA (1991). Guidance Manual for Compliance with the Filtration and Disinfection Requirements for Public 

Water Systems using Surface Water Sources, American Water Works Association, Denver, CO. 

USEPA (1992). Consensus Method for Determining Groundwaters under the Direct Influence of Surface Water 

using Microscopic Particulate Analysis (MPA), Port Orchard, WA. 

USEPA (2001). Draft, Considerations for Evaluating the Effectiveness of Alternative Filtration at Central Wyoming 

Regional Water System (CWRWS), Region 8, Denver, CO (unpublished letter to CWRWS). 


Rule, 40 CFR 141 and 142, Washington, D.C. 


Process, Tailored Collaboration with Louisville Water Co., American Water Works Association Research 

Foundation, Denver, CO. 

Wilson, M.P., W.D. Gollnitz, S.N. Boutros, and W.T. Boria (1996). Determining Ground Water under the Direct 

Influence of Surface Water, American Water Works Association Research Foundation Project 605, Denver, CO. 

179

180 

BILL GOLLNITZ is a Supervisor of Treatment for the Greater Cincinnati Water Works, 

where he is responsible for water quality and treatment for both the Charles M. Bolton and 

Mason Ground Water Facilities. Gollnitz has been in the water supply industry for over 

30 years, having previously managed surface-water and groundwater utilities in New York 

and Rhode Island. He has also had extensive experience in water-supply protection and 

surface-water/groundwater interactions. In addition, he was a Project Manager and co-author 

on the American Water Works Association Research Foundation’s project “Determining 

Ground Water under the Direct Influence of Surface Water.” Gollnitz has completed groundwater under the 

direct influence of surface water evaluations in Connecticut, New York, Ohio, Rhode Island, and Wyoming, 

and he has been published in the Journal American Water Works Association and elsewhere. In 1992, he was 

honored with an American Water Works Association Best Paper Award. Gollnitz received a B.S. in Biology 

from Mount Union College in Alliance, Ohio, and a M.S. in Environmental Science-Water Resources from 

the State University of New York, College of Environmental Science and Forestry in Syracuse, New York.

Session 9: Public Policy and Registration 

Source Water Protection and Riverbank Filtration 

in the Dyje River Basin 

Prof.-Dr. Petr Hlavínek 

Brno University of Technology 

Brno, Czech Republic 

Prof.-Dr. Jaroslav Hlavác˘ 

Vodárenská Akciová Spolec˘nost a.s. 

Brno, Czech Republic 

Water is not a commercial product like any other but, rather, a heritage that must be protected, 

defended, and treated as such. Achieving regional water-quality goals often involves substantial 

capital investments and changes in public attitudes concerning resource management. Economic 

impacts include: 

• The cost of facilities designed to reduce the discharge of contaminants into natural waters 

or to improve the quality of waste-receiving waters. 

• Limitations on economic activities and economic development in a particular region or 

river basin. 

Those responsible for the formulation and adoption of water-quality plans and management 

policies must have a means of estimating and evaluating the temporal and spatial economics, 

environmental, and ecologic impacts of these plans and policies. This need has stimulated the 

development and application of a wide range of mathematical modeling techniques for predicting 

the impact of alternative pollution control plans. 

The Dyje River is one of the biggest boundary streams in the Czech Republic. It is located in 

southern Moravia, crosses the state boundary between Austria and the Czech Republic several 

times, and creates the state boundary (with a total length of approximately 50 kilometers). There 

are 86 urban areas (agglomerations) larger than 2,000 population equivalent, with a total of more 

than 1.2-million inhabitants in the area of interest. The area of the Dyje River Basin is 

approximately 12,000 square kilometers, and the monitored river network is approximately 

1,000-kilometers long (Figure 1). There is number of RBF plants along the Dyje River basin that 

are, in the long run, affected by river-water quality. 

The Dyje Project deals with the areas that do not yet meet European Union legislation and 

correspond with the Instrument for Structural Policies for Pre-Accession (ISPA) definition of 

agglomerations. 


Prof.-Dr. Petr Hlavínek 

Professor of Civil Engineering, Faculty of Civil Engineering 

Brno University of Technology 

Zizkova 17 • 602 00 Brno, Czech Republic 

Phone: 420-541147733 • Fax: 420-543245032 • Email: hlavinek.p@fce.vutbr.cz 

181

182 

Figure 1. Location of the Dyje River Basin in the Cezch Republic. A= Austria. SK = Slovakia. 

PL = Polland. D = Germany. 

The goal of the project is to ensure that: 

• The quality and volume of water collected from polluters and treated in wastewater 

treatment plants complies by the year 2005 and is within the area managed by Union 

water supply and sewer systems with the values set down by Directive No. 91/271/EEC. 

• All wastewater discharged to the sewer is collected. 

• Within any drained area serviced by a wastewater treatment plant, all wastewater is 

carried to that wastewater treatment plant and treated in compliance with the values set 

down by Directive No. 91/271/EEC. 

• National parks and landscape-protected areas within the region are protected in terms of 

surface-water and groundwater quality. 

A group of related projects covers the following priority measures, as classified below: 

Category A.1: Reconstruct and upgrade existing wastewater treatment plants in municipalities 

with populations over 2,000 (or equivalent). 

Category A.2: Refurbish existing large wastewater treatment plants (for populations over 10,000 

[or equivalent]), including equipment for the removal of nitrogen compounds and phosphorus. 

Category A.3: Reconstruct existing sewer systems connected to existing wastewater treatment 

plants to provide sufficient capacity and a high level of treatment. 

Category A.4: Complete sewer systems and connect them to existing wastewater treatment plants 

to provide sufficient capacity and a high level of treatment. 

Category A.5: Construct wastewater treatment plants in municipalities with populations over 

2,000 (or equivalent). 

Category A.6: Complete sewer systems and wastewater treatment plants in municipalities with 

populations over 2,000 (or equivalent).

For the first phase of the Dyje Project, 10 agglomerations were chosen based on area investigations 

and comparisons with European Union standards. A technical solution was designed for those 

locations. Wastewater treatment and collection needs were identified and their standards 

compared with European Union standards. See Figure 2 and Table 1. 

1. Dyje – Moravská Dyje 

2. Jihlava – lower part 

3. Jihlava – upper part 

4. Oslava 

5. Bobrava 

6. Svratka – lower part 

7. Svratka – upper part 

8. Litava 

Figure 2. Sub-catchments 1 through 8 of the Dyje River. 

Table 1. Basic Data of the Dyje River Catchment Area 

Increase in Connected Inhabitants PE 203,606 

Increase in Removed Pollution PE 154,905 

Total Capacity of New and Upgraded WWTP PE 350,133 

Increase in Removed Pollution – SS ton/year 3,110 

Increase in Removed Pollution – BOD 5 ton/year 3,392 

Increase in Removed Pollution – COD ton/year 6,785 

Increase in Removed Pollution – N-NH4 ton/year 622 

Increase in Removed Pollution – Total ton/year 141 

Increase in Treated Wastewater m 3 /year 9,103,587 

Wastewater Treated on WWTPs m 3 /year 20,975,687 

Total Length of New and Upgraded WWTP kilometers 601 

WWTP = Wastewater treatment plant. SS= Suspended solid. BOD 5 = Biochemical oxygen demand. 

COD = Chemical oxygen demand. N-NH4 = Ammonia nitrogen. PE = Population equivalent. 

183

184 

As a result of improvements in pollution source-management at the Dyje River Basin, mathematical 

modeling techniques were used to assess stream-water quality. The study aimed to evaluate 

the effect of improvements on stream-water quality at main streams in the Dyje River basin, with 

special respect to the Dyje River transboundary profile (downstream of the City of Br˘eclav). The 

results were used as bases in the decision-making process. By doing this, financial sources can be 

distributed more efficiently in the basin and the application of remediation measures would better 

improve water quality while considering the whole spectrum of requirements connected to the 

process of approaching the European Union. 

There are a number of RBF plants along the Dyje River Basin. After decades of safe operation, 

some came under heavy pressure due to increasing river-water pollution. As an example, the RBF 

plant at Vojkovice is discussed. See Figures 3 and 4. 

Concentration mg/L 

2.5 

2.0 

1.5 

1.0 

0.5 

0 

1979 

1981 

1983 

Average 

1985 

1987 

1989 

Date 

Maximum 

Figure 3. Maximum and average concentration of N-NH4 at the RBF plant at Vojkovice from 1979 to 2000. 

In 1967, 170,000 tons of biochemical oxygen demand (BOD 5) per year were discharged into rivers 

in the Czech Republic. From 1957 to 1970, 800 new wastewater treatment plants were built, and the 

load decreased to 142,000 tons of BOD 5 in 1975. Between 1970 and 1980, the construction of 

wastewater treatment plants was stifled; therefore, the load increased to 198,000 tons of BOD 5 in 

1980. In 1990, the load decreased to 148,000 tons of BOD 5. Due to the massive construction of 

wastewater treatment plants after 1990, the load decreased to 67,000 tons of BOD 5 in 1995 and 

to 65,000 tons of BOD 5 in 2000. 

Gradual improvements in quality are expected after the construction and completion of the 

project, “Protection of Water in the Dyje River Basin.” It is expected that a number of RBF plants 

will be immediately affected (Ivancice, Moravské Bránice), while others will be gradually affected 

(Lomnicka, Omice, Vojkovice). 

1991 

1993 

1995 

1997 

1999

Concentration mg.l 1 

60 

50 

40 

30 

20 

10 

0 

1979 

1981 

1983 

1985 

Average 

1987 

1989 

Date 

Maximum 

Figure 4. Maximum and average concentration of N-NO 3 in the RBF plant at Vojkovice from 1979 to 2000. 

The simulation models are a relatively crude approximation of interactions among various 

constituents that occur in water bodies. The best water-quality simulation model is the simplest 

one that will adequately predict water-quality impacts within a particular water body associated 

with a particular water-quality management policy. Yet, in spite of current limitations, simulation 

models are the only reasonable means available for predicting surface-water quality. The state-ofthe-art 

in water-quality modeling and the understanding of physical, chemical, and biological 

processes that affect water quality are improving rapidly. The model provides preliminary 

estimates; a more precise tool should be used after collecting all necessary data (i.e., especially 

more accurate pollution data, corresponding real discharge data at the time of sampling, more 

accurate flow hydrodynamics calculation, etc.). The activities mentioned are quite time-consuming 

and expensive, yet necessary when improvements to the receiving waters must be more precisely 

assessed and quantified. 

Regional projects for water protection are complex, complicated, multidisciplinary, and long-term. 

They are politically important, with complicated relations among authorities both at the local and 

international levels. Close cooperation of all involved parties and companies is absolutely 

necessary. The Dyje Project was established as a pilot project for a regional solution in the Czech 

Republic. The project, “Water Protection of the River Dyje Basin,” has been accepted for 

financing from the ISPA fund. 

1991 

1993 

1995 

1997 

1999 

185

186 

PETR HLAVÍNEK has 20 years experience in wastewater treatment and water quality. He 

has 10 years of experience as a lead designer in the design company, HYDROPROJECT. 

At present, he is Professor of Civil Engineering at Brno University of Technology, Faculty 

of Civil Engineering. He also acts as Vice-President of the Czech Wastewater Treatment 

Experts Association, as well as Deputy Head of the Institute of Urban Water Management 

and as Member of the governing board of the Faculty of Civil Engineering at Brno 

University of Technology. Hlavínek has completed more than 200 research reports, expert 

opinions, and reports dealing with water quality and wastewater treatment. He has published more than 120 

articles dealing with wastewater treatment and water quality, and is author and co-author of six books, 

including Industrial Wastewater Treatment, Upgrading of Wastewater Treatment Plants, Hydraulics of WWTP, 

Wastewater Treatment-Examples of Calculation, Drainage of Urban Areas — Conceptual Approach and Water 

Structures, and Water Structures. Hlavínek received an M.S. at Brno University of Technology, a Dipl. in 

Sanitary Engineering at IHE Delft Netherlands, and a Ph.D. at Brno University of Technology.

Session 11: Case Studies “Lessons Learned” 

Greater Cincinnati Water Works Flowpath Study 

Field Design: Methodology and Evaluation 







Jeffrey Vogt 



Objective 

The Greater Cincinnati Water Works, USGS, and Miami University of Oxford, Ohio, entered 

into a joint research study to evaluate the effectiveness of RBF on an unconsolidated sand and 

gravel aquifer at the Greater Cincinnati Water Works’ Charles M. Bolton Well Field. One of the 

goals of this study was to better understand the aquifer’s effectiveness at reducing biological 

drinking-water contaminants. The purpose of this presentation is to describe and evaluate the 

study’s field design to determine its effectiveness for achieving study goals. This presentation will 

also discuss water-quality anomalies in the observed data and how they relate to field design. 

Methodology 

The Charles M. Bolton Well Field is located within the Great Miami Buried Valley Aquifer, which is 

composed of sand and gravel. The well field consists of 10 production wells located 50 to 600 ft 

from the Great Miami River. The well field is situated along the southern edge of the Great Miami 

Buried Valley Aquifer and is bounded by the bedrock valley wall to the south and the Great Miami 

River to the north. Due to this configuration, the well field is dependent upon the river as a major 

source of induced recharge or infiltration. 

Two study sites were chosen within the well field. The sites are adjacent to Production Well 1 

(referred to as Site 1) and to Production Well 8 (referred to as Site 8). At each site, four vertical 

monitoring wells with 2-ft long well screens were installed along a theoretical flowpath between 

the river and production well. Each well was installed progressively deeper, moving away from the 

river and toward the well. The two deepest wells were installed to correspond to the top and 

bottom of the production well screen. In addition, an inclined well was installed at each site. The 

inclined wells were installed at 20- to 30-degree angles from the horizontal and drilled beneath 



Hydrogeologist 



Phone: (513) 624-5611 • Fax: (513) 624-5670 • Email: Bruce.Whitteberry@gcww.cincinnati-oh.gov 

187

188 

the river. This allowed water-sample collection from approximately 5- to 10-ft beneath the Great 

Miami River. 

Each of the vertical flowpath wells was constructed using 4-inch diameter polyvinyl chloride wells 

with 2 ft of 10-slot (0.010-inch) screen. The lengths of the screens were designed to provide the 

shortest interval necessary to accommodate a sample pump and downhole data. 

Once the wells were installed and developed (to clean the well of construction-induced silt), a 

data sonde was installed in each well. The data sonde was used to record temperature, conductivity, 

pH, dissolved oxygen, and water levels. Each well was fitted with a bladder pump to collect 

water samples. By using dedicated equipment for sampling, the possibility of introducing contamination 

into the well was minimized, sampling was more consistent and efficient, and eliminating 

equipment blank samples from the quality control program reduced analytical costs. The inclined 

wells were constructed similarly to vertical wells, but with a 6-inch diameter polyvinyl chloride 

well casing and 5 ft of 10-slot (0.010-inch) well screen. A submersible sampling pump and data 

sonde were installed in each inclined well. 

To monitor the river, USGS installed a stream gage station at Site 1. The gage continuously 

measured and recorded river stage. It was also fitted with an automated sampling pump and data 

sonde to collect readings of the same parameters measured in the wells. 

Prior to each sampling event, the monitoring wells were purged using low-flow (minimal 

drawdown) purging methodology. Each well was pumped at a rate of less than 1 liter per minute 

(1 to 3 liters per minute for the inclined wells), while minimizing drawdown in the wells (usually 

less than 0.02 ft). Temperature, pH, specific conductance, and dissolved oxygen were measured 

during purging, and the well was sampled when the parameters stabilized according to USGS 

National Water-Quality Assessment Program Sampling Protocol (Koterba et al., 1995). With 

low-flow techniques, sampling-induced turbidity problems can sometimes be minimized. Because 

the goal of this study was to sample discrete portions of the aquifer in close proximity, this 

technique was also desirable to minimize the mixing from portions of the aquifer above or below 

the screened area. 

Anomalous Data 

Because the purpose of this study was to evaluate the effectiveness of the aquifer in acting as a 

filtration system, significant anomalies in the data were of concern because of their potential to 

represent a preferred flowpath through which water is not effectively filtered. One anomaly 

observed was the concentrations of TOC in Well FP8B. Because this well is of intermediate depth, 

the TOC concentration was expected to be between the concentrations of shallower (FP8A) and 

deeper (FP8C) wells. This is the trend seen at Site 1; however, during the first 2 months of the 

study, FP8B had TOC concentrations several times higher than the river (Figure 1). Throughout 

the first 2 months of monitoring, TOC dropped steadily and stabilized, with concentrations 

between those of Wells FP8A and FP8C. 

Another anomaly identified in the study was high particle counts in Wells FP8B and FP8D. Well 

FP8B particle counts were lower than the river, but higher than FP8A (a shallower well closer to 

the river). Particle counts in Well FP8D, the deepest monitoring well, were higher or very near 

the levels of particles in the river (Figure 2).

Concentration (mg/L) 

35 

30 

25 

20 

15 

10 

5 

0 

Aug-99 Nov-99 Feb-00 May-00 Sep-00 Dec-00 Mar-01 Jul-01 

Figure 1. Total organic carbon at Site 8. 

Counts/100 ml 

10000000 

1000000 

100000 

10000 

1000 

100 

Discussion 

Production Well 8 FP8A FP8B FP8C River 

Date 

Production Well 8 FP8A FP8B FP8D River 

10 

Aug-99 Dec-99 Mar-00 Jun-00 Oct-00 Jan-01 Apr-01 Jul-01 

Date 

Figure 2. Particle counts (3 to 5 micrometers in size) at Site 8. 

In general, the field design was appropriate for this project. Several monitoring wells were intentionally 

placed close together to evaluate the aquifer in detail. Although appropriate for this study, 

the design was costly. Similar designs with several monitoring wells and intensive water-quality 

analyses may be prohibitively expensive for many utilities. Based on the results of this study, 

natural filtration at similar sites could be evaluated using fewer wells. With fewer wells, the wells 

can be designed with longer screens to accommodate higher pumping rates for analyses, such as 

189

190 

Microscopic Particulate Analysis. In addition, longer screen lengths may be less susceptible to 

local aquifer heterogeneities because water will be contributed from a larger portion of the aquifer. 

For other study designs, using data sondes that measure only temperature and specific conductance, 

as well as eliminating the inclined wells (depending on the goals of the study), would reduce costs. 

The construction of a simple stilling well fitted with a transducer, data sonde, and sampling pump 

would be adequate for short-term river monitoring. 

In general, the monitored parameters showed the most significant drop in concentration from the 

river through the riverbed to the inclined wells. Further reductions of concentrations continued 

(although not as dramatically) as the water traveled from below the streambed to the production 

well. Another significant drop in concentrations occurred from the “C” wells (nearest the 

production wells) to the production wells. This is believed to be caused by the dilution effect of 

regional groundwater as a result of the large capture zone of the production well. 

The high TOC concentrations observed in Well FP8B during the first 2 months of the study 

(described earlier) may be due to the growth of indigenous aerobic bacteria after oxygen was 

introduced during well installation and development. Once oxygen was consumed, the aerobic 

bacteria likely died off and TOC dropped to natural levels. This indicates that this particular well 

intercepted a zone of the aquifer with different biological properties than other parts of the aquifer. 

The presence of strong odors of hydrogen sulfide, indicative of sulfur-reducing bacteria, and the 

excessive corrosion of metal pump fittings are also indicative of biological activity in this well. Had 

these high concentrations been due to a preferential flow path, TOC concentrations would not 

have exceeded the levels of the river and would not have been expected to decline. 

If the high particle counts in Wells FP8B and FP8D were a result of poor filtration, other constituents 

also show poor reduction from the river to the well. Aside from the initial high concentrations in 

FP8B, TOC concentrations were between those of Wells FP8A and FP8C. Additional parameters 

such as UV 254, nitrate, and chloride (not shown) also reflected effective filtration through the 

aquifer. This demonstrates that the high particle counts in Well FP8B originated in the aquifer and 

were not a result of a preferential flowpath. In the case of FP8D, the particle levels are above the 

counts in the river, indicating a secondary source of particles. As in FP8B, other parameters in FP8D 

indicated the high particle counts were not a result of preferential flow. 

It is widely understood that even a relatively homogeneous aquifer contains localized heterogeneities 

in particle size, in situ biology, and water chemistry. These heterogeneities, however, do 

not necessarily compromise filtration through the aquifer. Some anomalies can be recognized and 

evaluated by incorporating sufficient monitoring points, lengthening well screens to intercept 

larger sections of the aquifer, and monitoring for sufficient parameters so that anomalies can be 

recognized and evaluated. 

Acknowledgement 

Funding for this project was provided by a grant from the Ohio Water Development Authority, a 

grant from USGS, and funds from the Greater Cincinnati Water Works. The authors thank each 

of these agencies for their contributions and support throughout this project.

REFERENCE 

Koterba, M.T, F.D. Wilde, and W.W. Lapham (1995). Ground-water data-collection protocols and procedures for 

the National Water-Quality Assessment Program: Collection and documentation of water-quality and samples and 

related data, U.S. Geological Survey Open File Report 95-399. 

BRUCE WHITTEBERRY has been a Professional Hydrogeologist for Greater Cincinnati 

Water Works for the past 5 years. In addition to RBF research, Whitteberry oversees the 

wellhead protection programs for two well fields and addresses various groundwater 

quantity and quality concerns for the Greater Cincinnati Water Works. He is one of the 

Greater Cincinnati Water Works’ representatives for the Hamilton to New Baltimore 

Ground Water Consortium, and he also oversees the Consortium’s regional groundwater 

monitoring program, as well as serves in an advisory capacity on groundwater issues. Prior 

to joining the Greater Cincinnati Water Works, he spent several years in the groundwater industry, both in 

consulting and in a regulatory capacity. Whitteberry received a B.S. in Geology from Olivet Nazarene 

University and an M.S. in Hydrogeology from Wright State University. 

191

192


The Hungarian Experience with Riverbank Filtration 

Ferenc Laszlo, Ph.D. 

Institute for Water Pollution Control 

Water Resources Research Centre 

Budapest, Hungary 

The water-supply systems of Hungary provide drinking water for 9.5-million inhabitants 

(95 percent of the population) in 2,100 settlements of the country. About 88 percent of the total 

volume supplied originates from groundwater sources. About 30 percent of this (approximately 

800,000 m 3 /d) is abstracted from RBF water resources. 

There is an additional 3-million m 3 /d free RBF capacity along the Hungarian section of the 

Danube River. 

The largest RBF system in Hungary provides drinking water for 2-million inhabitants in Budapest. 

The present capacity is about 900,000 m 3 /d. Approximately 850 RBF wells are located on two 

islands (Szentendre Island and Csepel Island) of the Danube River. Some parts of this RBF system 

have operated for more than 100 years (Laszlo and Homonnay, 1986). The hydraulic conductivity 

of the gravel terraces, the composition of the filtration layer, and the water quality of the river 

enables the production of high-quality drinking water. 

In recent years, considerable efforts were focused on investigating and developing methods of 

practical application for the reliable utilization and protection of bank-filtered water resources. 

The activities were mainly directed toward evaluating pollution impacts and exploring the 

processes and causes of water-quality changes in RBF systems. 

Studies in an RBF pilot area on Szentendre Island included: 

• An extended field investigation to provide adequate data for a comprehensive 

water-quality evaluation and for modeling activities. 

• Adsorption experiments in the laboratory. 

• Application and development of numerical models to assess and predict water-quality 

changes in the RBF system that are induced by various potential changes in the 

conditions of the river and filtration media. 

Numerical modeling was applied to: 

• Calibrate hydraulic and transport parameters. 

• Calculate the transport of various pollutants from the Danube River to the wells. 

• Assess quality changes in the water of the production wells. 

• Predict the consequences of accidental pollution in the Danube River. 


Ferenc Laszlo, Ph.D. 

Director, Institute for Water Pollution Control 

Water Resources Research Centre 

H-1095 Budapest • Kvassay ut 1., Hungary 

Phone: +36 1 215 9045 • Fax: +36 1 216 8140 • Email: laszloferenc@vituki.hu 

193

194 

During the field investigation program, hydrogeological, hydraulic, hydrological, chemical, 

physico-chemical, biological, and microbiological data were collected and evaluated. The results of 

the water-quality studies revealed characteristic quality changes in the filter media: the chemical 

oxygen demand concentration gradually decreased from the river towards abstraction sites, 

indicating the decomposition of organic pollutants. Hydro-biological studies indicated that the 

Danube River had considerable fecal coliform pollution, but that bacteria removal during RBF was 

very efficient. Coliphages used as indicators of virus pollution were also measured. The results of 

these tests verified that coliphages were not able to pass through filter media. 

The retardation of specific organic micropollutants is very different depending on their biodegradability, 

sorption affinity, etc. The mobile, persistent organic micropollutants (e.g., atrazine, 

simazine, trichloroethene, and benzene) have low removal efficiency (less than 30 percent); on 

the other hand, some pesticides (e g., terbutrin, 2,4-D, and carbaryl) and petroleum hydrocarbons 

are removed substantially. The retardation or remobilization of some heavy metals is influenced 

by redox conditions. 

Adsorption experiments in the laboratory provided information on the partition coefficients of 

various micropollutants between spiked Danube water and the material of filtration medium taken 

from RBF sites (Laszlo and Literathy, 2002). Two riverbed materials were used: one represented the 

sandy alluvium having low organic material; the other was taken from a sedimentation zone of the 

riverbed where river water infiltrates through sediment having high organic material content and fine 

particle size. The partition coefficient varied widely for the different organic micropollutants and 

heavy metals, depending mainly on the character of the pollutants. The type of bed material was also 

important: the partition coefficients were higher in silty riverbed material than in the sandy matrix 

for all pollutants. 

Numerical transport modeling — based on measured hydraulic, hydrogeological, and physicochemical 

parameters — concluded that relatively short-duration accidental pollution has limited 

risk for RBF schemes along the Danube River due to a wide range of travel times from the river to 

abstraction wells. Higher risk would be long-lasting, continuous pollution of the Danube. 

Another concern is the impacts of river training and gravel dredging on the quality of riverbankfiltered 

water (Laszlo et al., 1990). Training structures (groynes, parallel dykes) and dredging 

operations affect the hydraulic conditions in the river that are conducive to silting in areas with 

reduced flow velocities. Adverse hydrochemical changes occur in the silted filter layer, especially 

the dissolution of iron and manganese, and higher concentrations of ammonium are observable. 

Bed degradation owing to dredging causes changes in the inflow ratio to the abstraction wells, 

increasing the proportion of polluted groundwater from the background areas in the wells. 

Conclusions 

RBF has been an effective method for drinking-water supply in Hungary for a long time. 

The sustainability of RBF abstraction schemes depends on several factors, including the quality of 

river water, characteristics of the filtration media, retention time within the aquifer, and quality 

of the groundwater adjacent to the extraction site.

REFERENCES 

Laszlo , F., and Z. Homonnay (1986). “Study of effects determining the quality of bank-filtered well water.” 

Conjunctive Water Use, IAHS Publ. No. 156, 181-188. 

Laszlo, F., Z. Homonnay, and M. Zimonyi (1990). “Impacts of river training on the quality of bank-filtered 

waters.” Wat. Sci. Tech., 22 (5): 167-172. 

Laszlo, F., and P. Literathy (2002). “Laboratory and field studies of pollutant removal.” Riverbank filtration: 

Understanding contaminant biogeochemistry and pathogen removal, C. Ray, ed., Kluwer Academic Publishers, 

Dordrecht. 

FERENC LASZLO is Director of the Institute for Water Pollution Control of the Water 

Resources Research Centre in Hungary. A chemical engineer, Laszlo has 30 years of 

experience in research related to water pollution and aquatic chemistry. His research 

interests include the protection of riverbank-filtration systems, investigation of 

water-quality changes in different water resources, study of micropollutants, simulation of 

the fate and transport of contaminants in the aquatic environment, development and 

operation of water-quality monitoring system, and accident emergency warning systems. 

In addition, he has provided short-term consulting and expert services to the World Health Organization and 

United Nations Development Program. Laszlo received an M.S. in Chemical Engineering and a University 

Doctorate Degree in Aquatic Chemistry from Veszprem University in Hungary. 

195

196


Nitrate Pollution of a Water Resource – 15 N and 18 O 

Study of Infiltrated Surface Water 

Frantisek Buzek, Ph.D. 

Czech Geological Survey 

Prague, Czech Republic 

Renata Kadlecova 



Miroslav Knezek, Ph.D. 


This study was undertaken to determine the effects of agricultural and human activities at the 

village of Karany near Prague, the Czech Republic, on the quality of water resources. Three groups of 

wells use bank infiltration water from the Jizera River to produce about 1,000 liters per second of 

quality drinking water for Prague. Some wells have exhibited a steady increase in nitrate content 

in recent years (Figure 1), although river-water quality remains good. A question arises of the 

origin of nitrate contamination and its transportation to wells, and to what extent bank-filtered 

water contributes to produced water. Also of interest are other contributions and how long it takes 

for these contributions to reach the wells. 

Figure 1. Nitrate content (in milligrams per liter, 6-months average value) of contaminated capture wells. 


Frantisek Buzek, Ph.D. 

Head of Stable Isotope Lab 


Geologicka 6 • 152 00 Prague 5, Czech Republic 

Phone: 420 251085345 • Fax: 420 251818748 • Email: buzek@cgu.cz 

197

198 

Site Description 

The study area is formed by Cretaceous sandstones and marlstones (the Jizera formation) of Turonian 

age. The Jizera River represents a base for local drainage. A shallow aquifer is situated in Late 

Pleistocene and Holocene alluvial deposits. They consist of sands and sandy gravel 7- to 9.5-m thick. 

The Quaternary aquifer is 2.5- to 5-m thick. It is hydraulically connected with the Jizera River 

stream. The Quaternary sediments have a high degree of transmissivity and high permeability. The 

Cretaceous deposits have an average transmissivity and free groundwater surface. 

On the right bank of the Jizera River, where the Turonian aquifer has a sandy development, its 

transmissivity is only about 1 order inferior to that of the alluvial plain deposits. On the left bank 

of the Jizera River, the permeability is lower due to a higher proportion of marly component in the 

Turonian sediments. Contamined water Wells 227 to 299 (Skorkov capture wells) are located 

along the Jizera River at a distance of 200 to 300 m from the right bank (Wells 227 to 270) and 

left bank (Wells 271 to 299) (Figure 2). Sojovice capture wells are located on the left bank only; 

one part is recharged by artificial infiltration (Wells 120 to 187), another part is traditional bank 

infiltration at a distance 300 m from the river (Wells 188 to 226). The river terraces are used by 

local farmers for the cultivation of corn, potatoes, and maize. Local villages have poor sewer 

systems, and there are several landfills located in the area. 

Methods 

From 1999 to 2000, we took water samples at various time intervals from the river, precipitation 

in the recharge area, precipitation at the site, and wells for the oxygen isotope composition of 

water (δ 18 O - H 2O), and river and wells for the isotope composition of nitrogen in nitrate 

(δ 15 N-NO 3). 1 

δ 18 O data were used to: 

• Describe the river system. 

• Specify groundwater recharge. 

• Calculate the contribution of infiltrated river water to well production. 

• Model infiltration water flow and transportation time to the well. 

• Specify additional water contributing to well water. 

δ 15 N-NO 3 data were used to: 

• Specify the origin of nitrate and possible mixtures. 

• Trace seasonal trends in nitrate content. 

• Follow reactions in the groundwater system (denitrification). 

River System 

The mean residence time T equals 7.2 months, as calculated from variations of δ 18 O values of the 

river and precipitation in the recharge area. An average contribution of direct precipitation to 

runoff is about 13 percent, with a maximum of 36 percent during the highest flood event. 

1 Isotopic composition is measured in delta units, δ(in ‰) = Rx /R s -1) × 1000, where R denotes the ratio of the heavyto-light 

isotope (e.g., 18 O/ 16 O or 15 N/ 14 N), and R x and R s are the ratios in the sample and standard, respectfully. As 

international reference standard is used, Standard Mean Ocean Water (SMOW) stands for water and atmospheric N 2 

for nitrogen.

Figure 2. A sketch of the study area. 

Key: 1 = Forested area. 

2 = Jizera-Elbe watershed. 

3 = Capture well. 

4 = Groundwater flow. 

5 = Deep well. 

6 = Landfill. 

Nitrate content and δ 15 N-NO 3 of the Jizera River follow seasonal variations. δ 15 N-NO 3 varies from 

8 ‰ in the winter to about 1.5 ‰ (base flow only). The highest nitrate content (about 25 mg/L) 

was measured during snowmelt. Base flow nitrate content is very low (about 8 to 10 mg/L). 

Additional nitrate to base flow originates mostly from drained precipitation events (short time or 

fast component of runoff), flushing fertilizers bringing mostly organic nitrogen in the spring 

(manure applied in the winter period), and inorganic nitrogen during the growing season 

(fertilizers applied on leaves). 

199

200 

The nitrate content and δ 15 N-NO 3 of the Jizera River do not follow the same seasonal variation. 

δ 15 N-NO 3 varies from 8 ‰ in winter to about 1.5 ‰ (base flow only). The highest nitrate content 

(about 25 mg/L) was measured during snowmelt. Base flow nitrate content is very low (about 8 to 

10 mg/L). Figure 3 shows the nitrate content and isotopic composition of the Jizera River. 

Figure 3. Nitrate content and isotopic composition of the Jizera River. 

As the response of river runoff to precipitation is quite fast (from 3 to 5 days after a storm), a 

simplified two-component model of runoff generation is applicable. The discharged water consists 

of groundwater and short residence-time components. Short time components are formed by 

drained precipitation events flushing fertilizers — predominantly the organic nitrogen applied in 

winter — and by inorganic nitrogen applied during the growing season. Increasing δ 15 N-NO 3 

levels during the autumn corresponds to residual nitrate levels from the unsaturated soil zone 

below root level. A similar seasonal variability was observed during the next year as well. 

Infiltration System 

Water Balance 

The actual and modeled δ 18 O data were compared during a flood event and steady-state 

conditions. At flood conditions, the contribution of infiltrated river water increased up to 

100 percent, with a transit time of only about 5 days. At steady-state conditions, infiltrated river 

water forms about 60 percent of infiltrated water, with a transit time of 28 days. The additional 

40 percent of infiltrated water cannot be explained by local groundwater only. It must originate 

from a shallow aquifer. Fast transit times of this recharging water were confirmed by a variability 

of δ 18 O in the wells, which is higher than in the river. Additional recharging water can be 

modeled by local precipitation input, with a delay in the order of weeks. 

Nitrate Contamination of Skorkov Wells 

From comparing the high nitrate content of contaminated wells and their seasonal δ 15 N-NO 3 

variability, it is obvious that: 

• Nitrate originates mostly from water other than infiltrated water. 

• Sources of nitrate are very similar to river sources (i.e., similar sequences of nitrogen 

inputs during the year. 

Wells have a higher concentration of nitrates than river water.

By analyzing the δ 15 N-NO 3 of single wells along the contaminated part of bank infiltration, we 

identified the possible sources of nitrate. Besides the seasonal application of fertilizers and manure, 

village sewer systems and landfills in the recharge area also contribute to water pollution in the 

wells (Figure 4). 

Figure 4. Nitrate content and isotopic composition of contaminated water (Skorkov single wells). 

The importance of local precipitation for recharging wells is obvious from the long-time variation 

of nitrate content of contaminated wells (see Figure 1). Minimal values on an otherwise 

continuously increasing plot correspond to years with very low precipitation input (25-percent less 

than average). With low precipitation input, wells are recharged preferentially by deeper 

groundwater with a lower nitrate content. Within a high precipitation period, wells are recharged 

by fast infiltrated local precipitation washing out nitrate in the unsaturated zone. 

Nitrate Contamination of Sojovice Wells 

Sojovice capture wells have two parts: one part is recharged by artificial infiltration (Wells 120 to 

187), another part is traditional bank filtration at a distance of 300 m from the river (Wells 188 

to 126). Contrary to the Skorkov system, the Sojovice wells are not affected directly by 

precipitation (Figure 5). Besides, both the bank infiltration and artificial infiltration parts are 

partially recharged by denitrified groundwater. From a time sequence of nitrate content and its 

δ 15 N-NO 3 on single wells, we could separate three components of the recharge: 

• River water. 

• Shallow groundwater with a nitrate source typical for infiltration in the left bank. 

• Deep groundwater with a reduction zone and denitrification. 

Flow paths of groundwater components are different — perpendicular to river flow (shallow 

component) and along the river (deep component). 

Conclusions 

Nitrate contamination of the bank infiltration systems originates in local sources of nitrate, 

including inappropriate uses of manure/fertilizers and leaking village sewers. A considerable 

201

202 

Figure 5. Nitrate isotopic composition of contaminated water (Sojovice single wells). 

increase of nitrate in the 1980s originated both because of the change in agriculture practices (due 

to an increased proportion of maize cultivation) and the change in hydrology dynamics of the 

system (due to the extensive use of water resources, new infiltration pathways were developed that 

recharge more shallower groundwater than before). 

FRANK BUZEK has worked for the Czech Geological Survey for about 25 years, and is 

Head of the Stable Isotope Group. His primary research interest includes the application 

of isotope geochemistry to environmental problems, hydrology, and organic geochemistry. 

As an all-round geochemist, he was responsible for stable isotope data on carbon and 

nitrogen cycling in forest soils (European Union international projects CANIF, NIPHYS, 

and FORCAST) and tracing atmospheric emissions and groundwater pollution 

(International Atomic Energy Agency project on isotope technique in groundwater 

pollution). Most of the recent case studies solved by Buzek and his team have dealt with nitrate in water 

resources. Buzek received an M.S. in Physical Chemistry from the Institute of Chemical Technology and a 

Ph.D. in Applied Surface Chemistry from the Institute of Chemical Processes Fundamentals at the Czech 

Academy of Sciences in Prague.


Microbial Growth in Artificially Recharged Groundwater: 

Experiences from a 4-Year Project 

Ilkka T. Miettinen, Ph.D. 

National Public Health Institute 

Kuopio, Finland 

Markku J. Lehtola, Ph.D. 



Prof. Terttu Vartiainen 



Prof. Pertti J. Martikainen 

Kuopio University 


Introduction 

There are regions in Finland where large groundwater aquifers are not available and where the 

artificial groundwater recharge technique is an important alternative for drinking-water production. 

In Finland, 9 percent of drinking water is produced using the artificial recharge of groundwater or 

bank-filtration techniques. The production of artificially recharged groundwater has increased 

during the last decade. The high content of organic carbon (humus) present in surface water is a 

serious problem when this water is used to produce drinking water. 

Organic carbon can react with chlorine during disinfection, forming halogenated byproducts. The 

high availability of organic carbon may also cause microbial problems in distribution networks. 

The fraction of organic carbon that microbes can use for growth (i.e., AOC) is considered to be 

the most important nutrient affecting microbial growth in drinking waters (van der Kooij, 1992). 

In boreal areas, phosphorus — in addition to organic carbon — has been shown to regulate 

microbial growth in drinking waters (Miettinen et al., 1996). 

Objectives 

A 4-year project examined how the artificial recharge of lake water affects the following: 

• Molecular weight distribution of organic matter. 

• AOC content. 

• Microbially available phosphorus (MAP) content. 

• Microbial growth potential in water. 


Ilkka Miettinen, Ph.D. 

Researcher, Laboratory of Environmental Microbiology 


Department of Environmental Health • P.O. Box 95 • FIN-70701 Kuopio Finland 

Phone: (358) 17-201371 • Fax: (358) 17-201155 • Email: Ilkka.Miettinen@ktl.fi 

203

204 

Methodology 

Experimental Sites 

Changes in water quality during artificial groundwater recharge were studied in five water works. 

All water works (Water Works A, B, C, D, and E) are located in esker areas, where the soil consists 

mainly of sand and gravel. Lake water is filtrated into the ground by basin infiltration in Water 

Works A, C, D, and E. Water Works A, B, and E use sprinkling infiltration through forest soil, 

while Water Works B and C use the same lake as their raw-water source. 

Water Analyses 

The quality of raw-water samples and water samples from different monitoring wells at different 

filtration distances (raw water, the first monitoring point after infiltration, and the last monitoring 

point after filtration) were analyzed from 1998 to 2001. Filtration distance varied from 10 to 280 m in 

the first monitoring point and from 390 to 1,200 m in the last monitoring point (artificially recharged 

groundwater). The quality of organic carbon was analyzed in terms of its molecular weight distribution 

using the high-pressure size-exclusion chromatography (HPSEC) method (Vartiainen et al., 1987). 

AOC was measured with a modified bioassay (Miettinen et al., 1999) originally presented by van der 

Kooij et al. (1982), and MAP was analyzed by bioassay (Lehtola et al., 1999). Microbial growth 

potential (heterotrophic growth potential) was tested using a method by Miettinen et al. (1997). 


Organic matter in lake waters consisted mainly of high and intermediate molecular-size fractions. These 

fractions decreased rapidly in the filtration process. In contrast, the lowest molecular weight fractions 

of organic matter decreased only slightly during filtration. The strongest reduction in the content of 

AOC occurred in the beginning of filtration. Later on, there was only a slight decrease in AOC 

content. On average, there was a 53-percent reduction in the concentration of AOC during the 

artificial recharge of groundwater (Table 1). Changes in MAP concentrations during filtration varied 

among water works. MAP content decreased strongly (89- to 99.9-percent removal) during filtration 

in Water Works A, B, and D. Water Works E was the only exception where MAP content increased 

during the filtration process, indicating that phosphorus was dissolving from soil into water (see Table 1). 

Table 1. Concentrations (Range and Mean Value in Parentheses) of AOC and MAP 

in Raw Water and Artificially Recharged Groundwater. Number of Observations: 4 to 11. 

Water AOC: µg Acetate eq. C/L MAP: µg MAP-P/L 

Works RW ARW RW ARW 

A 54 to 366 (134) 17 to 132 (46) 0.98 to 9.90 (3.84) 0.22 to 0.34 (0.27) 

B 15 to 120 (56) 15 to 73 (46) 1.79 to 5.31 (2.57) 0.15 to 0.37 (0.28) 

C 39 to 144 (78) 27 to 90 (49) Not Determined Not Determined 

D 27 to 288 (151)

Microbial growth was weak in all lake-water samples, despite the fact that these waters had high 

AOC and MAP contents; however, there was strong microbial growth in artificially recharged 

groundwater. Lower microbial growth in surface water could be associated with the grazing activity 

of protozoa (Hahn and Hofle, 2001). The microbial growth potential in groundwater was strongest 

in the first sampling point and weakened during filtration (Figure 1). The AOC content did not 

correlate with microbial growth potential; however, the maximum microbial counts correlated 

with the MAP content (r = 0.70, p = 0.000, n = 26) when raw-water samples and samples with 

MAP over 5 µg/L (no phosphorus limitation) were excluded from the data. 

Figure 1. Increase in microbial cell numbers (R2A plate counts) with time when water samples were 

incubated in the laboratory. The figure shows mean numbers in samples taken from raw water or 

artificially recharged groundwater of Water Works A, B, D, and E. Symbols: RW = Raw water. 

1.point = The first monitoring point after infiltration. 2. Point = The last monitoring point after 

filtration). Number of observations: 18 to 21. 

Conclusions 

• The quality of organic matter changed during artificial recharge: high and intermediate 

molecular weight organic compounds were removed from infiltrated lake water. 

• The decrease in AOC content was strongest in the beginning of infiltration. An increase 

in filtration distance had a minor effect on AOC removal. 

• MAP removal depended on soil characteristics. In most of the studied water works, 

removal was good. 

• Microbial growth was higher in artificially recharged groundwater than in raw water 

(surface water). The microbial growth potential decreased during filtration. 

• Because the removal efficiency for phosphorus was higher than that for carbon, phosphorus 

became the main nutrient regulating microbial growth in artificially recharged 

groundwater. 

205

206 

REFERENCES 

Hahn, M.W., and M.G. Höfle (2001). “Grazing of protozoa and its effect on populations of aquatic bacteria.” 

FEMS Microbiol. Ecol., 35: 113-121. 

Lehtola, M., I.T. Miettinen, T. Vartiainen, and P.J. Martikainen (1999). “A new sensitive bioassay for 

determination of microbially available phosphorus in water.” Appl. Environ. Microbiol., 65(5): 2,032-2,034. 

Miettinen, I.T., T. Vartiainen, and P.J. Martikainen (1996). “Contamination of drinking water.” Nature, 

381: 654-655. 

Miettinen, I.T., T. Vartiainen, and P.J. Martikainen (1997). “Microbial growth and assimilable organic 

carbon in Finnish drinking waters.” Wat. Sci. Techn., 35(11/12): 301-306 

Miettinen, I.T., T. Vartiainen, and P.J. Martikainen (1999). “Determination of assimilable organic carbon 

(AOC) in humus-rich waters.” Wat. Res., 3(10): 277-2,282. 

Vartiainen, T., A. Liimatainen, and P. Kauranen (1987). “The use of size exclusion columns in determination 

of the quality and quantity of humus in raw waters and drinking waters.” Sci. Total Environ., 62: 75-84. 

van der Kooij, D., W.A.M. Hijnen, and A. Visser (1982). “Determinating the concentration of easily 

assimilable organic carbon.” Journal AWWA, 74: 540-545. 

van der Kooij, D. (1992). “Assimilable organic carbon as an Indicator of Bacterial Regrowth.” Journal AWWA, 

84(2): 57-66. 

ILKKA MIETTINEN has been a researcher for the National Public Health Institute in 

Finland since 1987. Currently, he serves as Researcher of the Laboratory of Environmental 

Microbiology at the Institute, as well as Researcher for Kuopio University. For the past 

3 years, his research has focused on the role of phosphorous in microbial growth potential 

and biofilm formation in distribution networks. A recent project of his is called, 

“Surveillance and control of microbiological stability in drinking-water distribution 

networks.” Past research has focused on chemical and microbiological quality in humusrich 

lake water during bank filtration, as well as the effects of strong oxidants on the quantity and quality of 

organic matter and microbial regrowth in treated waters. Miettinen received a B.S. in Biochemistry, M.S. in 

Biotechnology, and Ph.D. in Environmental Sciences at the University of Kuopio, where he currently is 

Docent of Environmental Sciences.


Evaluation of the Existing Performance of Infiltration 

Galleries in Alluvial Deposits of the Parapeti River 

Dip.-Eng. Alvaro Camacho 

Bolivian Association of Sanitary Engineers 

La Paz, Bolivia 

Introduction 

The Parapeti River is part of the drainage system of the Parapeti-Izozog watershed, which covers 

about 52,000 square kilometers and is the sole water resource for supplying drinking water to the 

City of Camiri, Bolivia. The strata of the basin consist of Quaternary deposits and the permeability 

is good; however, precipitation is low (only 700 to 800 millimeters) and, as this zone is located in 

the watershed between the Amazon and La Plata river systems, existing groundwater quantity is 

very low. The aquifer depth is between 100 to 200 m in the vicinity of Camiri (yield capacity is 

less than 1.0 liters per second). The City is located in the region of Santa Cruz, Bolivia, at an 

altitude of 810-m above sea level, in the southeast, and has a population of 25,000 inhabitants. 

The average temperature is 25-degrees Celsius, with a maximum of 38-degrees Celsius in the 

summer and a minimum of 11-degrees Celsius in the winter. 

The City’s water demand is supplied by five infiltration galleries buried below the riverbed 

(Figure 1), on the bedrock, at a depth of about 4 to 5 m. The galleries are interconnected through 

Gallery 

1 

Pumping 

Station 

Collector 

Well 

Parapeti River 

Gallery 

2 Gallery 

5 

Figure 1. Location of the study area. 


Dip.-Eng. Alvaro Camacho 

Consultant Engineer 

Bolivian Association of Sanitary Engineers 

Casilla 9348 • La Paz, Bolivia 

Phone and Fax: (591-2) 241- 6283 • Email: alcamachog@unete.com 

Study 

Area 

Camiri 

Location of the Infiltration Galleries 

Collector 

Well 

Collector 

Well 

Gallery 

3 Gallery 

4 

Collector 

Well 

Collector 

Well 

207

208 

a network and linked to a collector well sited on the riverbank. From there, the collected water is 

pumped to a disinfection tank prior to delivery to customers. 

The infiltration galleries are channels with a 1.0-m × 1.0-m section or drains of 12-inch covered 

by four layers (Table 1) of a filter medium (Figure 2) 

4 to 5 m 

Permeable 

Alluvial 

Deposits 

Figure 2. Infiltration gallery system. 

Table 1. Artificial Filterbed Layers of Infiltration Galleries 

Bed Layers Type/Size Height 

First Layer (Top) Coarse Sand 1.75 m 

Second Layer Gravel, 1 to 2 inches 1.0 m 

Third Layer Gravel, 2 to 3 inches 1.0 m 

Fourth Layer (Bottom) Small Stone, 3 to 6 inches 1.0 m 

Natural Sand River 

Section 1 m × 1 m 

Water Surface 

Gravel 1 to 2 inches 



Bed Rock 

River Bed 

In this area of the Parapeti River, where the galleries are in operation, the riverbed consists of 

permeable alluvial material, sand, and gravel of about 4- to 5-m deep, which creates favorable 

conditions for groundwater flow through the unconfined aquifer (Huisman, 1978). Because the 

movement of groundwater develops small velocities in granular material, the flow is laminar (Harr, 

1990) and this is the driving force for the sedimentation process in the sand and gravel layers of 

the river. The Parapeti is a mountain river whose water flow is subject to dramatic variations, both 

in quality and quantity, during the dry season (May to October) and rainy season (November to 

April). With a relatively steep slope (0.5 to 1 percent), intensive sediment transport rates occur 

in the rainy season (the maximum flow exceeds 1,000 m 3 /s, as compared to 5 m 3 /s in the dry 

season). Regarding water quality in the river, turbidity levels in the rainy season have been reported 

at about 15,000 ntu and suspended solid concentrations at 30,000 mg/L (Binnie & Partners 

Consultants, 1982). 

In spite of Parapeti’s water quality, infiltration galleries have worked continuously for more than 

15 years, producing water of good quality and using disinfection as the only treatment process, 

with a discharge of 48 to 50 liters per second. The only annual O&M consists of removing the 

finest particles of sand that have reached the bottom of the galleries.

Objectives 

The aim of the study was to evaluate the existing performance of infiltration galleries to improve 

on the water quality of surface waters. The study was oriented to answer the following question: 

“To what extent do infiltration galleries work to remove contaminants from surface-water supplies?” 


A sampling program was implemented during the dry and rainy seasons. Two complete sets of water 

samples were collected and analyzed between 2002 and 2003 (Schubert, 2002). The first set of samples 

was taken in the dry season, from early October to November 2002, when suspended-solid 

concentrations in the river are low. The second set of samples was collected during the rainy 

season, from February and April, when the river carries higher suspended-solid concentrations. In 

both periods, water samples were taken over a period of 41 to 45 days. Grab samples were collected 

three times per day in the Parapeti River and in two of the five infiltration galleries (Numbers 1 

and 4). Raw-water samples collected in the river were taken from a depth of 0.30-m below the 

surface and 100-m upstream of the galleries. At the same time, water samples were taken from the 

effluents of the two infiltration galleries (Numbers 1 and 4) at the inlet point of the disinfection tank. 

Eight water-quality parameters were chosen: temperature, color, turbidity, suspended solids, total 

dissolved solids, pH, conductivity, and fecal coliform counts. Grab samples for temperature and pH 

were collected in glass bottles and analyzed on site. Random samples for color, turbidity, suspended 

solids, total dissolved solids, and conductivity were collected in glass bottles for analysis in a 

laboratory. Microbiological samples were collected in sterile 50-mL Pyrex glass for analysis in a 

laboratory within 24 hours. The test was conducted using a portable OXFAM-DEL AGUA water 

test kit (membrane filtration technique). A duplicate was prepared from each sample. All the 

experimental tests were based on the “Standard Methods for the Examination of Water and 

Wastewater” (1975). 

In addition, exploration holes were dug into different layers of the coarse material to take samples from 

different filter media that cover the galleries. Each layer of the filter media, from Galleries 3, 4, and 5, 

was washed with distilled water to perform a column-settling test. This experiment was conducted to 

determine the frequency distribution of settling velocities and particle diameters. Also, each sample 

was subject to physical analysis, such as mass density, porosity, and granular grain-size distribution. 


Figures 3 to 5 illustrate seasonal variations in water quality in the Parapetí River and infiltration 

galleries. For the purposes of this paper, only two indicators of water quality were chosen: suspended 

solids and fecal coliform. 

During the rainy season, raw water in the Parapeti River had suspended-solid levels ranging from 

45 to 4,850 mg/L, with a mean value of 618 mg/L. These figures differ substantially in comparison 

to the dry season, where the average value was 72 mg/L and the maximum was 410 mg/L. In the 

rainy season, because the river’s flow is high (from 5 to 1,000 m 3 /s, with a mean value of 30 m 3 /s), 

the elevated suspended-solid concentration is caused by erosion that occurs on the steep slopes of 

watershed highlands. As shown in Figures 3 to 8, the infiltration galleries that were evaluated 

during the same period had very low levels. The values were ranged from 0 to 7 mg/L, with an average 

of 2 mg/L far below the common standards. From this evidence, it has been demonstrated that the 

performance of the infiltration galleries reached a removal ratio of more than 90 percent. Turbidity 

of 5 ntu or less in 98 percent of the daily samples was reported in the effluents (see Figure 5). 

209

210 

6045 

5045 

4045 

3045 

2045 

1045 

45 

8 

7 

6 

5 

4 

3 

2 

1 

0 

River 

Gal. 1 

Gal. 4 

Suspended Solids (mg/L) 

Feb-03 Feb-03 Feb-03 Mar-03 Mar-03 Mar-03 Mar-03 Mar-03 Mar-03 Apr-03 Apr-03 

Figure 3. Raw water and infiltration gallery water quality. 

70150 

60150 

50150 

40150 

30150 

20150 

10150 

150 

200 

150 

100 

50 

0 

River 

Gal. 1 

Gal. 4 

Fecal Coliforms (FC/100 mL) 

Feb-03 Feb-03 Feb-03 Mar-03 Mar-03 Mar-03 Mar-03 Mar-03 Apr-03 

Figure 4. Raw water and infiltration gallery water quality. 

% Cummulative Distribution 

120% 

100% 

80% 

60% 

40% 

20% 

Faecal Coliforms in Galleries. Frequency Distribution. 

Rainy Season (2003) 

0% 

0 20 40 60 80 100 120 140 160 180 

NTU 

Figure 5. Quality of water produced in the infiltration galleries. 

Gallery 1 

Galeria 4

6000 

4000 

2000 

Suspended Solids (mg/L) 

0 

Minimum Mean Maximum 

Gallery 1 0 1 5 


River 45 618 4850 

Figure 6. A comparison of raw water and water produced by the galleries. 

80000 

60000 

40000 

20000 

FE Faecal RMS (FC/100 mL) 

0 

Minimum Mean Maximum 


Gallery 4 1 32 169 

River 160 8958 60685 

Figure 7. A comparison of raw water and water produced by the galleries. 

Microbiological water-quality parameters in the Parapeti River show poor quality with a wide range 

of 160 to 60,000 fecal coliform counts/100 milliliters, with an average value of 9,000 fecal coliform 

counts/100 milliliters. The infiltration gallery data illustrate that in one gallery (Gallery 1), the 

mean value was zero fecal coliform counts/100 milliliters, with a maximum of 20 fecal coliform 

counts/100 milliliters achieving a removal ratio of more than 90 percent. Three fecal coliform 

counts /100 milliliters or less in 98 percent of the samples was found. In the other gallery (Gallery 4), 

the mean value was a concentration of 32 fecal coliform counts/100 milliliters, with a maximum 

of 169 fecal coliform/100 milliliters representing a removal ratio of more than 90 percent. In this 

gallery, 120 fecal coliform counts/100 milliliters or less in 98 percent of the daily samples was found. 

These differences may be attributed to variations in gallery O&M. 

211

212 

Cummulative Distribution 

100% 

90% 

80% 

70% 

60% 

50% 

40% 

30% 

20% 

10% 

0% 

Conclusions 

Suspended Solids in Galleries. Frequency Distribution. 

Rainy Season (2003) 

0 1 2 3 4 5 6 7 8 

NTU 

Figure 8. Quality of water produced by the galleries. 

Gallery 4 

Gallery 1 

Infiltration galleries built in the permeable alluvial deposits of the rivers can be suitable systems 

for improving water quality from surface waters. Removal ratios of more than 90 percent for 

turbidity, suspended solids, and fecal coliform have been recorded (during this research project). 

Infiltration galleries produce high-quality and stable volumes of water independent of seasonal 

variations in the quality and quantity of raw-surface stream, minimizing environmental effects. 

It seems that there is a natural and self-cleaning filtration process at work here. The permeable 

deposits of the Parapeti River form a highly efficient system for the pretreatment of water that 

contains high concentrations of solids. 

Moreover, the filtration qualities of the deposits are constantly maintained by the action of the 

river itself: during periods of flooding, the top layer of filtration material — the coarse sand — is 

stirred-up and cleaned by the sheer force of water flow. As flooding subsides, the clean, loose sand 

re-settles on the riverbed, thus preventing the natural process of clogging and hardening, which 

would otherwise impair its efficacy as a filter. It is, therefore, possible that the natural composition 

of the river deposits, assisted by the cycle of the river itself, are acting as an effective and 

self-cleaning filter of the solids contained in the river. 

With proper studies and depending on local conditions and the characteristics of raw-water 

quality, infiltration galleries are a suitable alternative for supplying drinking water to small and 

medium communities, with minimum capital costs and lower O&M costs. The system in Camiri 

has been running since the early 1980s, and no critical O&M problems have been observed 

(particularly clogging). It is essential that galleries must have easy access to facilitate the periodic 

cleaning of sediments from conduits (the minimum channel section should be 1.0-m × 1.0-m or 

more for manual cleaning). The infiltration gallery system in Camiri has shown that the amount 

of maintenance required on the galleries is very small indeed. For this reason, as well as the 

benefits of improved or maintained water quality, infiltration galleries have a significant advantage 

over other conventional systems. 

Further research is still needed to form a complete understanding of the whole process and the 

mechanisms that are involved in removing contaminants at higher concentrations.

REFERENCES 

American Public Health Association (1975). Standard Methods for the Examination of Water and Wastewater, 

Fourteenth Edition, American Public Health Association, New York. 

Binnie & Partners Consultants (1982). Final Report on the design of the surface water treatment plant in Camiri, 

Binnie & Partners Consultants, Santa Cruz, Bolivia. 

Harr, M.E. (1990). Groundwater and Seepage, General Publishing Company, Toronto, Ontario. 

Huisman, L. (1978). Ground Water Recovery, International Institute for Hydraulic and Environmental 

Engineering, Technical University of Delft, The Netherlands. 

Schubert, J. (2002). “German Experience with Riverbank Filtration Systems Ray.” Riverbank Filtration 

Improving Source-Water Quality, C. Ray, G. Melin, and R.B. Linsky, editors, Kluwer Academic Publishers, 

Dordrecht. 

Since 2002, ALVARO CAMACHO has been a consultant engineer for the water and 

sanitation sectors, including the Bolivian Association of Sanitary Engineers. Prior, he was 

the Director General of Water Supply and Sanitation for the Ministry of Housing Services 

of the Republic of Bolivia, a Project Leader in the Water and Sanitation Program for rural 

areas in Bolivia for the World Bank, and Guest Lecturer of Sanitary Engineering at the 

Universidad Mayor de San Andrés in Bolivia. He is also the author of several reports on 

topics such as water-treatment plants designed within the multi-stage filtration concept, 

developing the water and sanitation sector in Bolivia, technical standards for the design of unconventional 

sewer systems in Bolivia, and guidelines for the design of water-treatment plants in small and medium 

communities. Camacho received a degree in Civil Engineering from the Universidad Mayor de San Andrés 

in Bolivia and a Dipl.-Engineer in Sanitary Engineering from the International Institute for Hydraulics and 

Environmental Engineering, in conjunction with the Technical University of Delft, The Netherlands. He is 

a M.S. researcher at the Universidad del Valle in Cali, Columbia, where he is conducting a field study on 

infiltration galleries for water supply. 

213

214


Sensitivity and Implications of Microscopic Particulate 

Analysis – A Collector Well Owner’s Perspective 

Barry C. Beyeler 

City of Boardman 

Boardman, Oregon 

The City of Boardman, Oregon, owns two horizontal collector wells situated along the banks of 

the Columbia River, which provides the City’s water supply. The horizontal collector well system 

was chosen in 1975 for its unique ability to produce high volumes of high-quality water. The 

horizontal collector wells fill a role of vital importance in the City by addressing the balance of 

plentiful water supplies for industrial development and the need for high-quality drinking water 

for the citizens of the community. 

The City has been confident of the water quality produced by the collector and the filtration 

effectiveness it possesses; however, under provisions of the Surface Water Treatment Rule, the 

potential water-quality effects of the hydraulic connectivity to the Columbia River came under 

question. The City chose to perform Microscopic Particulate Analysis on the collector and Columbia 

River in the attempt to indicate and quantify the filtration effectiveness of the horizontal collector 

well system. 

Performance of Microscopic Particulate Analysis in the field presented many challenges that 

needed to be addressed to ensure that a quality representative sample was taken. There are many 

ways this sample can be compromised, which could produce non-representative or inaccurate 

results. Understanding the sensitivity of the Microscopic Particulate Analysis process from start to 

finish, combined with an assessment of what the indicated results would trigger under the Safe 

Drinking Water Act concerning additional treatment and subsequent associated costs, required 

significant care and diligence. 

As a result of the sampling and analysis that the City performed, the State of Oregon Department 

of Human Resources Drinking Water Program has determined that the City is operating a 

“groundwater” system. This is based on the quality of the water and assurances provided by 

accurate and representative data obtained through Microscopic Particulate Analysis. As the City 

brings its second collector online, the coordination of the Microscopic Particulate Analysis 

sampling efforts are being discussed with the Department of Human Resources Drinking Water 

Program to ensure that this collector performance can be quantified appropriately. The Microscopic 

Particulate Analysis process will be accomplished over the next year to provide the information 

to make appropriate decisions. The City will follow the same process of sampling the river and 

collector to indicate filtration effectiveness. 

In retrospect, there have been numerous other benefits to the process of Microscopic Particulate 

Analysis sampling. The increased knowledge of how the system works has been invaluable in 


Barry C. Beyeler 

Community Development Director 

City of Boardman, Oregon 

202 N Main Street • P.O. Box 229 • Boardman, Oregon 97818 USA 

Phone: (541) 481-9252 • Fax: (541) 481-3244 • Email: bbeyeler@cityofboardman.com 

215

216 

forums concerning Endangered Species Act listings of salmon and steelhead populations in the 

Columbia and Snake River basins. As proposals for river operations are brought forward, the City 

of Boardman can more accurately respond to what the potential impacts may be. The increased 

knowledge and communication between the City and State and Federal agencies has also been 

beneficial in allowing the City to get through permitting processes associated with this water right. 

Although there is no direct method to measure the impact Microscopic Particulate Analysis has 

had in this regard, there is no doubt that it has played a role in improving communication linkages 

and the level of understanding between the agencies and public involved. 

Since April 2003, BARRY BEYELER has been Community Development Director for the 

City of Boardman, Oregon, where he has been employed for over 22 years. Among his 

responsibilities, he coordinates and reviews all land-use activities within the City and its 

urban growth boundary. He also oversees all planning and land-use issues relating to 

utilities operations (water, wastewater, storm drainage, and traffic infrastructure), natural 

resource impacts and environmental policy, wellhead protection, and public education. 

Among his professional affiliations, Beyeler is a member of the Oregon Water Resources 

Department Ground Water Advisory Committee, where he is currently serving a third 4-year term and is 

former Committee Chair. Beyeler received the Northeastern Oregon Sub-Section American Water Works 

Association Activities Award in both 1992 and 1995.

RBF_Cover (for eps) - National Water Research Institute

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?