20.04.2013 Views

Documentation of the Evaluation of CALPUFF and Other Long ...

Documentation of the Evaluation of CALPUFF and Other Long ...

Documentation of the Evaluation of CALPUFF and Other Long ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

• The best performing <strong>CALPUFF</strong> configuration from all <strong>of</strong> <strong>the</strong> sensitivity tests when looking<br />

at <strong>the</strong> performance across all <strong>of</strong> <strong>the</strong> fitted plume performance statistics was use <strong>of</strong> <strong>the</strong><br />

slug option with puff splitting in <strong>CALPUFF</strong>/MMIF.<br />

A key result <strong>of</strong> <strong>the</strong> GP80 600 km receptor arc evaluation was <strong>the</strong> need to invoke <strong>the</strong> near‐<br />

source slug option to adequately reproduce <strong>the</strong> <strong>CALPUFF</strong> performance from <strong>the</strong> 1998 EPA<br />

<strong>CALPUFF</strong> evaluation study. Given that <strong>the</strong> slug option is a very nonst<strong>and</strong>ard option for LRT<br />

dispersion modeling, this finding raises concern regarding <strong>the</strong> previous <strong>CALPUFF</strong> evaluation.<br />

Ano<strong>the</strong>r important finding <strong>of</strong> <strong>the</strong> GP80 <strong>CALPUFF</strong> sensitivity tests is <strong>the</strong> wide variation in<br />

modeling results that can be obtained using <strong>the</strong> various options in CALMET <strong>and</strong> <strong>CALPUFF</strong>. This<br />

is not a desirable attribute for regulatory modeling <strong>and</strong> emphasizes <strong>the</strong> need for a st<strong>and</strong>ardized<br />

set <strong>of</strong> options for regulatory <strong>CALPUFF</strong> modeling.<br />

1975 Savannah River Laboratory (SRL75) Field Experiment<br />

The 1975 Savannah River Laboratory (SRL75) field experiment released a tracer on December<br />

10, 1975 <strong>and</strong> measured it at receptors located approximately 100 km downwind from <strong>the</strong><br />

tracer release site. The fitted Gaussian plume model evaluation approach was used to evaluate<br />

numerous <strong>CALPUFF</strong> sensitivity tests. Several CALMET sensitivity tests were run to provide<br />

meteorological inputs to <strong>CALPUFF</strong> that varied whe<strong>the</strong>r MM5 data was used or not <strong>and</strong> how<br />

meteorological observations were used (surface <strong>and</strong> upper‐air, surface only or no<br />

observations). As in <strong>the</strong> GP80 sensitivity tests, three dispersion options were used in <strong>CALPUFF</strong><br />

(CAL, AER <strong>and</strong> PG). In addition, <strong>CALPUFF</strong>/MMIF sensitivity tests were performed using MM5<br />

output at 36, 12 <strong>and</strong> 4 km resolution.<br />

Because <strong>of</strong> <strong>the</strong> long time integrated sampling period used in <strong>the</strong> SRL75 experiment, <strong>the</strong> plume<br />

arrival, departure <strong>and</strong> residence statistics were not available <strong>and</strong> only <strong>the</strong> fitted Gaussian plume<br />

statistics along <strong>the</strong> 100 km receptor arc were used in <strong>the</strong> evaluation. The key findings <strong>of</strong> <strong>the</strong><br />

SRL75 <strong>CALPUFF</strong> evaluation are as follows:<br />

• The maximum plume centerline concentrations from <strong>the</strong> fitted Gaussian plume to <strong>the</strong><br />

observed tracer concentrations is approximately half <strong>the</strong> maximum observed tracer<br />

concentration at any monitor along <strong>the</strong> 100 km receptor arc. As a plume centerline<br />

concentration in a Gaussian plume represents <strong>the</strong> maximum concentration, this indicates<br />

that <strong>the</strong> fitted Gaussian plume is a very poor fit to <strong>the</strong> observations. Thus, <strong>the</strong> plume<br />

centerline <strong>and</strong> plume width statistics that depend on <strong>the</strong> fitted Gaussian plume are a poor<br />

indication <strong>of</strong> model performance for <strong>the</strong> SRL75 experiment. The observed fitted Gaussian<br />

plume statistics were taken from <strong>the</strong> 1998 EPA study (EPA, 1998a).<br />

• Given that <strong>the</strong>re are many more (~5 times) <strong>CALPUFF</strong> receptors along <strong>the</strong> 100 km receptor<br />

arc than monitoring sites where <strong>the</strong> tracer was observed, <strong>the</strong> predicted maximum<br />

concentration along <strong>the</strong> arc is expected to be greater than <strong>the</strong> observed maximum<br />

concentration. Such is <strong>the</strong> case with <strong>the</strong> <strong>CALPUFF</strong>/MMIF runs, but is not always <strong>the</strong> case<br />

for <strong>the</strong> CALMET/<strong>CALPUFF</strong> sensitivity tests using no MM5 data.<br />

• The <strong>CALPUFF</strong> plume centerline is <strong>of</strong>fset from <strong>the</strong> observed plume centerline by 8 to 20<br />

degrees. The largest angular <strong>of</strong>fset occurs (17‐20 degrees) when CALMET is run with no<br />

MM5 data. When MM5 data is used with <strong>the</strong> surface <strong>and</strong> upper‐air observations <strong>the</strong><br />

<strong>CALPUFF</strong> angular <strong>of</strong>fset is essentially unchanged (18‐19 degrees) <strong>and</strong> <strong>the</strong> removal <strong>of</strong> <strong>the</strong><br />

upper‐air observations also has little effect on <strong>the</strong> plume centerline angular <strong>of</strong>fset.<br />

However, when only MM5 data are used, in ei<strong>the</strong>r in CALMET (11‐12 degrees) or MMIF (9‐<br />

10 degrees), <strong>the</strong> <strong>CALPUFF</strong> plume centerline <strong>of</strong>fset is improved.<br />

11

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!