Optimizing Multiple Scattering Calculations

in the CRTM: IR Results

Tom Greenwald, Jim Davies and Ralf BennartzUniversity of Wisconsin

The 12th JCSDA Technical Review Meeting & Science Workshop on Satellite Data Assimilation, May 21-23, 2014

MotivationMultiple scattering calculations dominate CRTM

FWD/ADJ calculations in cloud/precipitation-affected radiance data assimilation

Extend our work in optimizing the CRTM in the microwave spectrum to the infrared

“Optimize” means finding the minimum number of RT solver streams (i.e., angular resolution) needed to achieve a desired accuracy

2The 12th JCSDA Technical Review Meeting & Science Workshop on Satellite Data Assimilation, May 21-23, 2014

MethodsDeveloped a scattering indicator (SI) to estimate

degree of scattering; more scattering usually requires more angular resolution (i.e., streams)SI is function of optical depth and single-scattering albedo

Tested using high-resolution WRF model simulations of hurricane Katrina and marine stratocumulus

Applied to the HIRS (High Resolution Infrared Radiation Sounder) for longer wavelengths (6-15 µm)

Assumed target solution accuracy of 0.2 K

3The 12th JCSDA Technical Review Meeting & Science Workshop on Satellite Data Assimilation, May 21-23, 2014

Katrina results

4The 12th JCSDA Technical Review Meeting & Science Workshop on Satellite Data Assimilation, May 21-23, 2014

Band (µm) Opt New

15 0.032 0.032

14.7 0.032 0.039

14.5 0.048 0.083

14.2 0.131 0.479

14 0.164 0.522

13.7 0.201 0.559

13.3 0.228 0.569

11.1 0.179 0.525

9.7 0.197 0.502

12.5 0.239 0.545

7.3 0.449 0.559

6.5 0.250 0.585

Timings relative to current CRTM

5

Katrina results cont.

The 12th JCSDA Technical Review Meeting & Science Workshop on Satellite Data Assimilation, May 21-23, 2014

Marine Sc results

6The 12th JCSDA Technical Review Meeting & Science Workshop on Satellite Data Assimilation, May 21-23, 2014

Band (µm) Opt New

15 0.039 0.039

14.7 0.039 0.039

14.5 0.049 0.039

14.2 0.041 0.044

14 0.042 0.084

13.7 0.043 0.383

13.3 0.043 0.808

11.1 0.063 0.768

9.7 0.098 0.744

12.5 0.044 0.802

7.3 0.048 0.230

6.5 0.041 0.058

Timings relative to current CRTM

7

Marine Sc results cont.

The 12th JCSDA Technical Review Meeting & Science Workshop on Satellite Data Assimilation, May 21-23, 2014

SummaryA method for accelerating IR cloudy-sky calculations in

the CRTM has been developed Correctly predicts optimum number of streams over 75% of

the time for most sounding bands, leading to potential speedup of up to 30x

For window bands this drops to 25-50% of the time but still provides potential speedup of up to 2x as compared to current CRTM

Has tendency to predict too few streams more often than current CRTM

8The 12th JCSDA Technical Review Meeting & Science Workshop on Satellite Data Assimilation, May 21-23, 2014

Remaining tasksInvestigate ways to improve the methodIntegrate it into the CRTM (much easier than

before since code infrastructure already in place in V2.2)

Manuscript in preparation

9The 12th JCSDA Technical Review Meeting & Science Workshop on Satellite Data Assimilation, May 21-23, 2014