Hyperspectral IR clear and cloudy sounding retrieval study Jun Li @, Timothy J. Schmit #, Chian-Yi Liu @, Elisabeth Weisz @, Li Guan @, and Allen Huang @ @Cooperative Institute for Meteorological Satellite Studies University of Wisconsin-Madison, Madison, Wisconsin #ASPT, Office of Research and Applications, NESDIS/NOAA, Madison, Wisconsin 4 th Workshop on Hyperspectral Science of UW-Madison MURI and GOES-R 17 28 April 2004 The Pyle Center Madison, WI UW-Madison
IR Spectral Coverage (DS or SW/M) HES Example 1 Example 2 0.625 cm-1 0.625 cm-1 HES 0.625cm -1 HES 1.25cm -1 2.5cm -1 0.6 cm -1 0.6 cm -1 C O 2 (T) O z o n e Traditional Side of H2O absorption Important lines for cloud emissivity and cloud type CO N2O Temperature CO 2 weak H 2 O
FAST FM INVERSE
Hyperspectral clear/cloudy sounding retrieval Clear sounding retrieval Regression and Regularization (physical retrieval) Cloudy Sounding retrieval Hole hunting for single FOV Cloud clearing using imager/sounder (ABI/HES, MODIS/AIRS are be used for testing the algorithm) Cloudy regression
HES GOES
HES GOES
Clear sky HES sounding retrieval RMSE Regression versus Physical
Hyperspectral cloudy sounding retrieval Hole hunting for single FOV requires a very good cloud-detection algorithm (high spatial resolution imager data can help) Cloud clearing using imager/sounder - base on Bill Smith concept (ABI/HES, MODIS/AIRS are be used for testing the algorithm) Cloudy regression requires a realistic cloudy radiative transfer model
TPW 02km/ABI on GOES-R
TPW 04km HES-SW/M
TPW 10km/HES DS on GOES-R
TPW ~14km/AIRS resolution
BT(K)
AIRS cloud detection from MODIS cloud mask Confident clear partly cloudy full cloudy
Hyperspectral cloudy sounding retrieval Hole hunting for single FOV requires a very good cloud-detection algorithm (high spatial resolution imager data can help) Cloud clearing using imager/sounder - base on Bill Smith concept (ABI/HES, MODIS/AIRS are be used for testing the algorithm) Cloudy regression requires a realistic cloudy radiative transfer model
where N * (ν)=ε 1 (ν)n 1 /ε 2 (ν)n 2. it follows that estimates of N* for window spectral regions, N * (W), can be calculated as ) ( * 1 ) ( 2 ) ( * ) ( 1 ) ( Q Q N v R N v R v clear R = ) ( ) ( ) ( ) ( ) * ( 2 1 W R W R W R W R W N clear clear = SRF(Rairs) Rmodis N* can be calculated from two adjacent AIRS cloudy footprints!!
Clear neighbor Cloudy FOV (Line 74, column 55) Cloudy neighbor for N*
Radiance difference between cloud-cleared FOV and clear neighbor FOV
AIRS single FOV profile retrieval versus ECMWF analysis
Temperature RMS difference between AIRS and ECMWF~ 250 thin cloud FOVs Clear Neighbor Cloud contaminated Cloud-cleared
Hyperspectral cloudy sounding retrieval Hole hunting for single FOV requires a very good cloud-detection algorithm (high spatial resolution imager data can help) Cloud clearing using imager/sounder - base on Bill Smith concept (ABI/HES, MODIS/AIRS are be used for testing the algorithm) Cloudy regression requires a realistic cloudy radiative transfer model
Simulation with MM5 Cube data 5.2 by 5.2 km spatial resolution 0.5 hour temperature resolution MM5 output include T/Q/O3/Ts and LWP, IWP, CTP, etc. Fast efficient cloud radiative transfer model (Yang et. al) accounting for cloud absorption and scattering is used
MM5 data - truth (T, Q, O3, Ts, LWP, IWP, CTP) Radiative transfer Model (clear and cloudy) HES clear and cloudy radiances Add HES TRD noise comparison Simulated HES observations Perform regression approach Sounding retrieval clear and cloudy skies (Tr, Qr, Tsr, etc.)
HES is not alone, together with high temporal resolution ABI, weather system evolution will be well captured!
30-minute step ABI 5-minute step
How much current GOES sounder can improve forecast? Study showed (Ma et al. 1999) current GOES does not change forecast temperature over CONUS statistically, however, it substantially improves forecast water vapor
Ma et al. 1999; JAM
Ma et al. 1999; JAM
How much GOES sounder can improve forecast? Question: Do we use the current GOES sounder information optimally?
GOES-10 Sounder
GOES-10 Sounder
Two points (1) Make better use of the current GOES sounder data (2) Prepare Optimal use of GOES-R HES data
Summary Hyperspectral IR data provide much better temperature and moisture information than current GOES sounder Hole hunting requires high spatial resolution and good cloud detection technique Cloud-clearing using Imager/Sounder data gives single FOV cloudy sounding in some partly cloudy cases. This technique needs to be further investigated Simulation with cube data from MM5 shows cloudy cloudy retrieval in some thin cloud cases are possible. Sounding retrieval should have positive impact on NWP if use the information correctly.