1
Assimilation of AIRS radiances and SFOV retrieval profiles in the Rapid Refresh model system Haidao Lin 1,2 , Steve Weygandt 1 , Ming Hu 1,3 , Stan Benjamin 1 , Curtis Alexander 1,3 , Patrick Hofmann 1,3 , Tim Schmit 4 , Jinlong Li 5 , Jun Li 5 1 NOAA/ESRL/GSD Assimilation and Modeling Branch 2 Cooperative Institute for Research in the Atmosphere, Colorado State University 3 Cooperative Institute for Research in Environmental Sciences, University of Colorado at Boulder 4 NOAA/NESDIS/STAR Advanced Satellite Products Branch 5 Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin-Madison Rapid Refresh RAP and HRRR • Hourly updated assimilation / model system using GSI analysis and WRF ARW model • RAP replaced RUC at NCEP as NOAA operational model May 2012 • RAP 13km domain covers all of North America, large oceanic region (few conventional obs.) • HRRR 3km domain covers CONUS, initialized from RAP BACKGROUND Evaluate the impact of AIRS data on the Rapid Refresh (RAP) and High Resolution Rapid Refresh (HRRR) mesoscale prediction systems, examine ways to maximize forecast improvement Atmospheric Infrared Sounder (AIRS) data provide high- resolution temperature and water vapor information Single Field of View (SFOV) profiles (temperature, moisture for clear sky conditions) obtained from application of CIMSS hyperspectral IR sounder retrieval (CHISR) algorithm 1-hr fcst 1-hr fcst 1-hr fcst 11 12 13 Time (UTC) Analysis Fields 3DVAR Obs 3DVAR Obs Back- ground Fields Rawinsonde (12h) 150 NOAA profilers 35 VAD winds ~130 PBL profilers / RASS ~25 Aircraft (V,T) 3500–10,000 TAMDAR 200 – 3000 METAR surface 2000 -2500 Mesonet (T,Td) ~8000 Mesonet (V) ~4000 Buoy / ship 200-400 GOES cloud winds 4000-8000 METAR cloud/vis/wx ~1800 GOES cloud-top P,T 10 km res. satellite radiance AMSUA/MHS/HIRS Radar reflectivity 1 km res. Data types – counts/hr Experiment design GOAL: Evaluate impact of AIRS data within full mix of observations. ASSUME no AIRS observation reduction due to data latency / cutoff • 9-day retrospective period (May 8-16, 2010) with 3-hourly cycled runs (real-time RAP uses 1-h cycle with partial cycle 2x per day) • Evaluate impact of AIRS data (radiance and SFOV retrieval) relative to control experiment (standard observations only) • Examine bias correction, data quality control, channel selection AIRS brightness temperature for channel 791 AIRS SFOV temperature for 500 hPA AIRS BT SFOV T 2100 UTC 8 May 2010 HRRR Temperature bias SFOV – raob comparison • 53 matched profiles (clear sky, 15-km horizontal distance, 3-h time window) • Three SFOV observation sets provided by U. Wisc. CIMSS, improvement shown V1 – first set V2 – improved V3 – best set Moisture bias SUMMARY AND FUTURE WORK Results: AIRS radiance data and SFOV with bias correction yield small positive impacts for short-term forecasts in the Rapid Refresh with the full mix of observations. Positive impact for radiance data with standard GSI bias correction with spin-up SFOV: Dry moisture bias, diurnal temperature bias pattern indentified and corrected. Plans: Examine AIRS impact for hurricane and other cases; cloud contamination investigation for radiance data; work toward use of AIRS data in operational RAP Radiance Assimilation Results 22 UTC RAP HRRR FIM ESRL - GSD Assimilation and Modeling Branch 00 UTC Evaluation of AIRS Radiance Bias Correction (BC) Forecast Verification AIRS Impact with other Satellite Data BC and O-A SFOV Retrieval Assimilation Results Histogram of moisture innovations [ specific humidity (O-B) normalized by background saturation specific humidity ] for the data in 400-800 hPa before and after BC from 9-day control run with SFOV in monitoring mode CNTL SFOV WITH BC SFOV NO BC CNTL SFOV WITH BC SFOV NO BC Analysis 12-h forecast 0-h analysis and 12-h forecast relative humidity bias for control run (no SFOV, black), SFOV with NO bias correction (blue) and SFOV with bias correction (red) SFOV Temperature Bias Correction Sample SFOV profiles compared with raobs Diurnal variation of horizontal avg. SFOV temperature innovations (O-B) Pressure vs. time of day cross- section of horizontally averaged SFOV temperature innovations (O-B). SFOV observations between 400 and 800 hPa are assimilated. Compare sample SFOV vs. raob. temperature and moisture profiles Forecast Verification Moisture innovations +15% bias correction Dry Moist Dry Moist Moisture innovations No bias correction 3-h forecast temp. bias (left) and 0-h temp. analysis bias (right) valid at 12Z for control run (black), SFOV temp. with NO BC (blue) and SFOV with BC (red) SFOV WITH BC CNTL SFOV NO BC SFOV NO BC SFOV WITH BC CNTL Assimilated West of AK --- (north) Eastern NA Central NA Western NA/AK AK / Grnlnd Eastern NA Western NA/AK 00z 03z 06z 09z 12z 15z 18z 21z 00z West of AK 9z + 3h fcst bias 12z analysis bias Tucson, AZ 12z 11 May 2010 Raob Warm Dry Cold SFOV 9-day retro average 400-800- hPa layer 9-day retro average Histogram of BT O-B before BC and after BC for AIRS channel 252 (left, CO2 channel , PWF ~672 hPa) and channel 1382 (right, water vapor, PWF ~866 hPa) from AIRS experiment with BC spin-up before BC BC and Quality Control (QC) BT Diff. & RMS Errors before and after Assimilation Histogram of BT O-B after BC for AIRS channel 252 (left) and channel 1382 (right) from AIRS experiments without BC spin-up (blue) and with BC spin-up (red) Without BC spin-up After BC With BC spin-up GSI calculated cloud top pressure (CTP) from experiments without BC spin-up (left) and with BC spin-up (right) on 0600 UTC 8 May 2010 RAP 0-h analysis CTP on 06z 8 May 2010 Total number of used observations with and without BC spin-up o Without BC spin-up * With BC spin-up Water vapor Carbon dioxide Surface (left) BT O-A difference and (right) O-A standard deviation (STD) for experiments without BC spin-up (circles) and with BC spin-up (stars) averaged over entire retrospective period. Channels are vertically arranged by each channel’s peak weighting function (PWF) height. o WITHOUT BC spin-up * WITH BC spin-up (left) BT difference and (right) RMS errors before (circles) and after (stars) assimilation vertically arranged by channel’s peak weighting function height Water vapor Carbon dioxide Surface Normalized error reduction relative to the control run [(CNTL-EXPT)/CNTL] (%) for AIRS assimilation experiments with (red) and without (blue) BC spin-up for (left) temperature, (middle) relative humidity, and (right) wind. Stats against raobs for 1000-100-hPa layer ). WITH BC spin-up WITHOUT BC spin-up WITH BC WITHOUT BC Normalized error reduction relative to the control run [(CNTL-EXPT)/CNTL] (%) for SFOV assimilation experiments with (red) and without (blue) BC for (left) temperature, (middle) relative humidity, and (right) wind (800-400-hPa layer vs. raobs over CONUS). HRRR Case Study (initialized from RAP) o O-B * O-A Observed composite reflectivity (left); HRRR 6-h forecast reflectivity, initialized from CNTL RAP (middle) and initialized from SFOV RAP (right), valid 15Z 13 May 2010. Adding only AIRS data Relative Humidity Wind Normalized error reduction relative to the control run for experiments with (red) and without (blue) AIRS data in conjunction with other satellite data (amsua/hirs/mhs) for (left) temperature , (middle) relative humidity, and (right) wind (100-1000-hPa layer). WITH AIRS plus other satellite WITH other satellite data CNTL 6h fcst SFOV 6h fcst 1500 UTC 13 May 2010 Wind Temperature Temperature 9-day retro average Without BC spin-up With BC spin-up RAP CTP SFOV moisture bias correction (BC) SFOV moisture innovation before and after BC o O-B * O-A Rel. Hum. Temperature Relative Humidity Wind 1000-100- hPa layer 9-day retro average 9-day retro average 1000-100-h Pa layer Water vapor Carbon dioxide Surface
