Black carbon in Arctic snow and ice - Atmospheric Sciences ...

Black carbon in Arctic snow and ice: 

Quantifying its effect on surface albedo 

Tom Grenfell & Steve Warren 

Dept of Atmospheric Sciences, University of Washington 

in collaboration with 

Tony Clarke (University of Hawaii) 

Other UW participants (past and future): 

Dean Hegg 

Sarah Doherty 

Mike Town 

Steve Hudson 

Hyun-Seung Kim (ESS) 

Jamie Morison, Andy Heiberg & Mike Steele(APL) 

Project website: www.atmos.washington.edu/sootinsnow

The question that motivates this 

research: 

Why is the Arctic 

warming so much faster 

than the global average?

Outline 

What affects snow albedo? 

Why are impurities so effective at reducing snow albedo? 

Why do impurities reduce snow albedo more than cloud albedo ? 

Measurement of black carbon (BC) in snow 

Measurements to date 

Planned measurements 

Uncertainties in measuring BC in snow 

Uncertainties in modeling effect of BC on snow albedo 

Errors in albedo measurement 

Why remote sensing is not useful to determine BC's effect 

on snow albedo 

Where and when does snow albedo matter for climate?

Beginnings: 

NCAR 1978 Steve Warren & Warren Wiscombe, RT Modeling


Grain size (age) 

Variation of grain size with depth 

Snow depth (& vegetation) 

Impurities: dust, black carbon (soot) – more effective for 

larger grains 

Sun angle 

density - if snow layer is thin 

Why does albedo vary with wavelength: 

Absorption spectrum of ice - strongly wavelength dependent 

Very transparent at blue and near UV wavelengths 

Opaque in the near infrared and beyond 

New values recently published (Warren, Brandt, Grenfell, 

Applied Optics, 45, 5320, 2006)

Surface snow at South Pole

Snow Albedo 

Effect of snow grain size (Wiscombe & Warren 1980) 

WW JAS I model

Interaction of sunlight with snow 

To get the same number of refraction events, the distance traveled 

through ice is greater in coarse-grained snow. Therefore, coarsegrained 

snow has lower albedo. 

Mostly refraction

Modelcalculated 

albedos were 

too high! 

Wiscombe & 

Warren 1980 

Grenfell & 

Maykut 1977, 

snow on Arctic 

sea ice: 

dry new snow 

wet new snow 

old melting 

snow 

Model albedos too high


Because absorption by ice is very weak at visible wavelengths: 

ice is nearly transparent. 

Almost any additional absorption is significant. 

Impurities effective

Snow Albedo 

0.5ppm 0.05ppm 

0.05 ppm 

5 ppm 

0.5 ppm 

5 ppm 

Warren & 

Wiscombe 

JAS, 1980, 

RT Model 

Results 

WW JAS II dirty snow

Grenfell & 

Maykut 

(1977) 

Obs 

Warren 

(1982) 

RT Model 

GM 77 Observations

Grenfell Perovich & Ogren (CRST, 1981). Obs of albedos & impurities for 

alpine snow. Dust ~ 20-60 ppm; maximum observed albedos between 0.8 

and 0.95. Clean snow ~0.98-0.99 from RT models. WWII values in range. 

GPO Observations

What impurities get into snow that affect the albedo on a 

regional scale? 

Soot; Windblown dust; volcanic ash; rafted sediments (sea 

ice); other exotic material of various colors on record. 

Many questions remain about how the material gets into the 

snow, but “clean” snow is not as common as we expected.

Where can we find clean snow?

South Pole Station, 1986 

South Pole

Warren & Clarke (JGR, 1990) 

Antarctic Plateau Soot

Grenfell, Warren, Mullen (1994) Hudson et al., J. Geophys. Res., 2006 

South Pole Dome C 

Observations consistent with RT models and soot concentrations 

Antarctic Plateau Spectral Albedo

Clarke & Noone A

AGASP 1 aircraft flights April 1983 and Snow Samples 1982-1983 

Range 1-120 ppb 

Most within 5-50 ppb. 

Western Sector Only 

1983 

Clarke & Noone Map

Warren & 

Wiscombe 

(1985); 

Warren & 

Clarke 

(1986) 

Soot 

contents 

from Clarke 

& Noone 

(1985) 

Albedo Change vs Mass Fraction

Now, 20 years after Clarke & Noone (1985), 

there is renewed interest in dirty snow. 

20 years after

Hansen & Nazarenko Proc. Nat. Acad. Sci. 2004

E 

i 

ΔTs 

/ Fi 

= 

ΔT 

/ F( CO ) 

Soot in snow has the greatest 

efficacy of all climate forcing 

components considered, 

because: 

(1) Peak of soot fallout (Spring) 

coincides with onset of 

snowmelt 

(2) Melting snow has lower 

albedo than cold snow 

(3) Earlier melt exposes dark 

underlying surface 

(4) Soot may concentrate at the 

surface during melting (?) 

s 

Efficacy of snow albedo 

2

New Project was launched in 2006: 

A comprehensive survey of soot in Arctic snow, 

to update Clarke & Noone’s survey from 1983-4.




Why do impurities reduce snow albedo more than cloud albedo? 



Planned measurements Uncertainties in measuring BC in snow 





Where and when does snow albedo matter for climate? 

Uncertainties in computing climatic effects of BC in snow

The Greenland Ice Sheet

Transportation

Sample jars inserted into the snow 

Surface samples Depth profile

Sample Jars dug out

NE Greenland Clean Tent

NE Greenland Microwave Oven

Decanting Melted Sample

Vacuum Filtering

Snow at 96 km Snow at 38 km 

(~ 400 ml of meltwater was passed through each filter.) 

Filters will be analyzed for light transmission at multiple 

wavelengths to separate contributions by soot and dust. Greenland filters

BC 

standards 

made by 

Tony 

Clarke 

“Monarch 

-71 soot” 

Standard Filters

NSF project 

began 2006 

Sites samples 

to date in red 

Planned sites 

in green 

Our sites

HOTRAX voyage, Summer 2005 (Grenfell) 

Arctic Transect

Canada & Alaska, Spring 2006 & 2007: Tom Grenfell, Von Walden, 

Dave Barber & Co., Matt Nolan, Matthew Sturm 

Arctic Canada

Cambridge Bay, Nunavut 2006

Resolute Bay and sampling traverse April 2006. Green points 

show the sample sites. 

Resolute

Konrad Steffen, Univ. 

Colorado. 

Automatic weather stations at 

2000-m level. 

Mike Town, Summit 

Lora Koenig, Swiss Camp 

Greenland

Matthew Sturm (CRREL) 

March-April 2007 

Sturm Traverse 07

The Russian Federation 

Большая Карта

Соруководитель 

Dr. Vladimir 

Radionov 

Arctic and 

Antarctic Research 

Institute 

St. Petersburg 

Victor 

Boyarsky 

VICAAR Polar Logistics 

Michael Lamakin & Valery Ippolitov

The quality of our shoes has been well tested for a long time. 

It is always strictly monitored – Peshekhod Shoe Center (in Siktivkar) 

Чувство Юмора

Prelim Effective Soot 

Concentration (ngC/g) 

1000 

100 

10 

Нарьян 

Мар 

Nar'yan 

Mar 

Воркута 

Vorkuta 

Хатанга 

Khatanga 

Диксон 

Dikson 

1 

0 50 100 150 

Site Index 

Наблюдение

Bronto 

Plan B 

Khatanga Transport

Vorkuta Transport

Results from NW 

Russia 2007 

Nar’yan Mar: 

Ave = 11 ppb 

Std Dev = 7 ppb 

Background ~ 5 ppb 

Vorkuta: 

Ave = 236 ppb 


‘Background’ ~ 90 ppb 

Khatanga: 

Ave = 36 ppb 



Dikson: 

Ave = 20 ppb 



Table of Values to date

In 20 years, the 

Arctic has become 

cleaner, by factors: 

1 

4 

2 

6 

Arctic is cleaner




Why do impurities reduce snow albedo more than cloud albedo? 



Planned measurements 






Where and when does snow albedo matter for climate? 

Uncertainties in computing climatic effects of BC in snow


1. Incomplete capture of BC by the filter (Clarke & Noone got 85-88%) 

2. Inadequate distinction of BC from other absorbers (dust) 

3. Absorption coefficient will vary depending on type of soot 

But we don’t need the mass; we need only its effect on albedo; that’s why 

we use an optical method. 

The mass we report would be the true mass if the absorber in the sampled 

snow were identical to the soot used to make the weighed standards. 

4. Snow may be unrepresentative 

- contaminated during sampling (sooty clothes after cooking) 

- too close to cities or research station

Greenland Summit, July 2006 (Mike Town) 

Greenland summit variations

Use of wavelength dependence of Absorption to separate Soot and Dust Absorption 

Kevin Noone & Tony Clarke 

Abisko, Sweden, 1984 

Soot filtered from snow 

from above tree-line 

Dust from stream-flow 

above tree-line 

Soot 

Volcanic 

Ash dust 

reference 

Calibration 

soot 

Dust 

Separate soot dust ash spectral analysis

Uncertainties in translating from filter measurement 

to effect of BC on snow albedo 

1. The effective absorptive properties of soot particles depend on the 

refractive index of the surrounding medium. A soot particle on the 

surface of the nuclepore filter material may not have the same mass 

absorption cross-section as a soot particle on the surface of a snow 

grain. 

2. Soot particles are on the surface of the filter but may be embedded 

within the snow particles, so k abs (soot on filter) differs from 

k abs (soot in snow). 

Uncertainty in computing effect on albedo as snow ages: 

1. Growth of snow grains 

2. Vertical redistribution of BC during melting 

Filter index and ageing

Vertical redistribution of BC during melting 

Do particles collect at the surface as the snow melts? 

Soot collecting at surface?

Conway et al., Albedo of dirty snow during conditions of melt, 

Water Resources Res., 1996 

Migration of soot with melting refs

Sources of error in albedo measurement 

1. Imperfect cosine-collector (deviation from cosine response varies 

with wavelength, and correction varies with diffuse-fraction of 

incident radiation) 

2. Imperfect leveling of instrument 

Albedo errors A






3. Sloping surface: 

Albedo errors B






3. Sloping surface: 

4. Sampling of bumps (sastrugi) 

(These 4 errors are much smaller under overcast sky.) 

Albedo errors C

Sastrugi

Sources of error in albedo measurement (continued) 

5. Fluctuations of incident irradiance under cloud (need to monitor 

this) 

6. Shadowing correction (can be made



1. It's hard to distinguish snow from clouds-over-snow. Thin near-surface 

layers of atmospheric ice crystals ("diamond-dust") are common in the 

Arctic. 

2. The bidirectional reflectance (BRDF) is affected by small-scale surface 

roughness: ripples, sastrugi, suncups, pressure-ridges. 

(The effects of sastrugi on BRDF are different at different wavelengths, 

because they depend on the ratio of sastrugi width to flux-penetration 

depth.) 

3. When a thin ground-fog (or diamond-dust layer) covers the rough snow, 

the forward peak is enhanced and the nadir view is darker, even though 

the albedo is probably unchanged or slightly higher. This darkening at 

nadir could be mistaken for soot-contamination. 

4. Grain shape affects BRDF (even if does not affect albedo). 

Remote sensing not useful 

A

Surface Frost on West Antarctica 

(Photo by Nadine Nereson)


on snow albedo (continued) 

5. Increase of grain size with depth reduces visible albedo more than near- 

IR albedo, masquerading as soot. 

6. Albedo reduction by BC in snow can be mimicked: 

- by thin snow. Sooty snow has the same spectral signature as thin snow. 

Arctic snow is thin in many places. 

- by sub-grid-scale leads in the Arctic Ocean. 

- by BC in the atmosphere above the snow ("Arctic haze"). 

Remote sensing not useful B

Where and when does variation of snow albedo matter for climate? 

Whenever snow is exposed to significant solar energy 

(snow albedo is less important in winter and in boreal forest) 

Arctic snow 

- Tundra in spring 

- Sea ice in spring (covered with snow) 

- Greenland Ice Sheet in spring (cold snow) 

- Greenland Ice Sheet in summer (melting snow) 

Non-Arctic snow 

- Great Plains of North America 

- Steppes of Asia: Kazakhstan, Mongolia, Xinjiang, Tibet 

Glacier ice and sea ice are also important: 

- Ablation zone of Greenland Ice Sheet in summer 

- Arctic ocean in summer 

Where and when does soot matter?

Summary of our Strategy 

Climatic effect of BC on snow albedo is expected to be at most a 

few percent, and snow albedo depends on several other 

variables. Therefore we do not directly measure the albedo 

reduction. Instead: 

1. Collect snow samples, melt and filter them. Analyze the filters 

for BC and dust. 

2. Use the BC amount together with snow grain size in a radiativetransfer 

model to compute albedo reduction - comparison with 

selected albedo observations. 

3. Direct measurement of albedo-reduction (to verify the modeling) 

may be possible where all other variables are constant: e.g. a 

transect outward from a pollution source.

Summary 1. Reduction of albedo 

To reduce albedo of old melting snow 

by 1% requires mass fractions: 

~10 ppb black carbon (soot) 

~500 ppb soil dust 

~1000 ppb volcanic ash

Summary 2. Preliminary results. 

Median BC values in Greenland are in the same range as found by 

Clarke & Noone (1985). 

Median background values from Canada, Alaska, and the Arctic 

Ocean are lower by factors of 2-6 than in early 1980’s. 

Background values from NW Russia are ~5-20 ppb 

Quantification refinements due shortly: 

new spectrophotometer to analyze filters 

new standards

Summary 3. Uncertainties in computing climatic effects 

of BC in snow: 

BC content of snow: inadequate temporal and spatial sampling 

Seasonal variation 

Geographical variation 

Vertical redistribution 

Effect on snow albedo: variations of 

Snow grain size (varies seasonally and geographically) 

Snow depth 

Importance of snow albedo for climate: what is hiding the snow? 

Vegetation cover 

Cloud cover 

Cloud thickness

Planned Activities 

1. Survey (update of Clarke & Noone 1985) 

(expand to eastern Arctic, several years at all possible locations) 

Additional data needed from: Arctic tundra, ice sheets & sea ice in Alaska, 

Canada, Greenland, Russia, Arctic Ocean, Scandinavia 

March-May (time of maximum snow depth, sample deposition at several 

vertical levels – intra-annual variations) 

Summer: Greenland Ice Sheet (melting snow), Arctic Ocean (melting sea ice). 

Who: Our group + Volunteers 

2. Process studies 

Scavenging ratio for BC in snowfall 

Vertical redistribution of BC with melting 

Characterization of soot by source type 

3. Methods comparison with Gerland, Berntsen 

4. Collaborate on source attribution with Dean Hegg (PC/factor analysis) + 

Planned Activities

Arctic Russia 

March-May 2007/ 08 

Tom Grenfell, Steve Hudson, Steve Warren 

Arctic Russia 2007-08

Jim Hansen 

for inspiring us to take on this project 

Ellen Baum and the Clean Air Task Force 

for getting us going with foundation funding 

(Torrance, Redman, Hatch, Kendall, Oak) 

NSF Arctic Program 

Thanks to: 

Our volunteer force for greatly extending coverage of our results 

Polar Continental Shelf Project of Canada 

for logistics support at Resolute, Nunavut 

Danish Polar Center 

for logistics support in northeast Greenland 

Russian Hydrometeorological Servive 

for support and permissions for northern Russia

Black carbon in Arctic snow and ice - Atmospheric Sciences ...

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?