MO4.L10 - EVALUATION OF CRIS/ATMS PROXY RADIANCES/RETRIEVALS WITH IASI RETRIEVALS, ECMWF ANALYSIS AND RAOB MEASUREMENTS
1. Evaluation of CrIS/ATMS Proxy Radiances/Retrievals With
IASI Retrievals, ECMWF Analysis and RAOB Measurements
Murty Divakarla1, Chris Barnet2, Mitch Goldberg2,
Xu Liu3, Bill Blackwell4, Eric Maddy5, Guang Guo5, Susan Kizer3,
Tom King5, Walter Wolf2, Antonia Gambacorta5, and Kexin Zhang5
1IM Systems Group Inc., 6309 Executive Blvd, Rockville, MD, 20852 USA.
2STAR, NOAA/NESDIS, Camp Springs, MD , 20746 USA.
3NASA Langley Research Center, Hampton, VA 23681, USA.
4MIT Lincoln Laboratory, Lexington, MA 02420 USA.
5PSGS, Fairfax, VA 20706 USA.
Paper Number: GR-10 1973
GR-
2010 IEEE International Geoscience and Remote Sensing Symposium
July 25-30, Honolulu, Hawai
25-
Acknowledgements:
Kevin Garrett, Tony Reale, Frank Tilley
Camp Springs, MD 20746.
1
2. Hyper-
Hyper-Spectral IR Sounders and MW Instruments
AIRS/IASI/CrIS and AMSU-A/MHS/ATMS
AMSU-
AIRS 9FOVs
1 2 3
4 5 6
7 8 9
8461
IASI 4FOVs
CrIS 9FOVs
1 2 3
4 5 6
7 8 9
Aqua-AIRS/AMSU-A 1:30 AM/PM Atmospheric Infrared Sounder (AIRS) – 2378 IR Channels
MetOp-IASI/AMSU-A/MHS 9:30 PM/AM Infrared Atmospheric Sounder Interferometer (IASI) – 8461 IR Channels
NPP-C1 C3: CrIS/ATMS 1:30
NPP C1 & C3 C IS/ATMS 1 30 AM/PM Cross-track I f
C t k Infrared Sounder (C IS) 1317 IR Channels
dS d (CrIS) Ch l
Advanced Microwave Sounding Unit (AMSU-A15 CH MW temperature sounder - 55 GHz Oxygen band)
Microwave Humidity Sounder (MHS 5 CH ~ 183 GHz)
Advanced Technology Microwave Sounder (ATMS – 22 CH Temperature and Moisture sounder)
AMSU/MHS 2
4. Outline for Presentation
CrIS/ATMS Proxy SDRs and CrIMSS EDRs
1. Product Retrieval Algorithm(s)/Retrieval Products (EDRs)
» NGAS CrIS and ATMS EDR Product Algorithm (Science Interface)
– Crimss_Larc_v1.5, Developed by Xu Liu and Kizer
Implemented at NOAA/STAR
» NOAA-IASI Operational Retrievals from NOAA/NESDIS/NUCAPS
NOAA-
2. Data Sets for CrIS/ATMS Proxy Data Generation and NGAS-
NGAS-
CrIMSS EDR Product Evaluation
» MetOp Global Data (MGD) Matches of ECMWF/RAOBs/GFS
» Focus Day (10/19/2007) Data MetOp IASI/AMSU-A/MHS
IASI/AMSU-A/MHS,
ECMWF, AVN
3. Proxy Data Generator Algorithms from LaRC and MIT
– C IS P
CrIS Proxy – X Li and Ki
Xu Liu d Kizer (LaRC)
(L RC)
– ATMS Proxy - Bill Blackwell (MIT)
Implemented at NOAA/STAR
4. Evaluation of Proxy Data Sets (CrIS/ATMS) and NGAS EDRs
– With Corresponding IASI/AMSU-A/MHS SDRs
IASI/AMSU-
4
– With ECMWF/RAOBs and IASI EDRs.
5. 1. Product Retrieval Algorithms /Retrieval Products
CrIMSS_LaRC_1.5 Version Ported from
NPP-
NPP-NGAS IDPS CrIMSS EDR Product Algorithm g
Simultaneous Retrieval
T(p), q(p), O3(p),Tskin
Spectral Emissivities
OSS-RTM
Use of EOFs to
Characterize and Measure
Retrieved Parameters
a-priori Constraints
Derived from a Blended
Training Data Set
5
6. 1. Product Retrieval Algorithms /Retrieval Products
NOAA Retrievals of
Core Products and Trace Gas Products
1 :3 0 P M /A M
A q u a -A IR S
M e tO p -IA S I
9 :3 0 A M /P M
CrIS/ATMS
R e trie v a l
Code
ECM W F
NCEP
SO N DES
R e trie v a l ATOVS
p ro d u c ts
The NOAA level 2 retrieval processing system was developed during
the Aqua mission (AIRS/AMSU-A)
(AIRS/AMSU-
Expanded to retrieve MetOp (IASI/AMUS-A/MHS) T(p), q(p), O3(p) core
(IASI/AMUS-
products, and trace gas products (CH4, CO, CO2 etc.)
Emerging as NOAA-Unique CrIS/ATMS Product System (NUCAPS)
NOAA-
Identical systems one for research and the other for operations
– Reprocessing Options with Algorithm Upgrades, New Data
– Emulate Various IASI/ AIRS Retrieval Algorithms
http://www.orbit.nesdis.noaa.gov/smcd/spb/iosspdt/iosspdt.php#1 6
7. Data Sets for CrIS/ATMS Proxy SDR Generation
and EDR Evaluation
MetOp-IASI/Aqua-AIRS
Match-up Database
Aqua-AIRS: (2002- Current)
MetOp-IASI: January 2008 - C
M tO IASI J Current)
t)
RAOB Measurements Matched to Aqua (1:30 AM/PM)
and MetOp(9:30 AM/PM) Satellite Observations
MetOp-IASI/AMSU-A/AMSU-B Level1B Radiances
IASI Level-2 Retrievals
Aqua-AIRS/AMSU-A Level1B Radiances
AIRS Level-2 Retrievals
NCEP-GFS (AVN) Level2-Forecast/Analysis
ECMWF Level-2 Forecast/Analysis
NOAA-18 ATOVS/M2-ATOVS Level-2 Retrievals
Ability to Get Matched O3SNDs and TO3 (BD
Measurements) From WOUDC
Collocated Within ±3 Hrs. & 100 Km Radius
~ 100,000 Matches for IASI (2008 …)
~ 200,000 Matches for AIRS (2002 …)
Source of RAOB Data : OSDPD MDB NOAA/NESDIS
OSDPD, MDB,
Many Publications - AIRS Website, NOAA/STAR/ Website
http://www.orbit.nesdis.noaa.gov/smcd/spb/iosspdt/iosspdt.php#1 7
8. Data sets Proven Valuable for Global Validation
Aqua-
Aqua-AIRS T(p), q(p) Validation
8
9. IASI Retrieval Validation with RAOBs, ECMWF etc.
Accepted Samples : 11,400 (Sea), 67,434 (Global)
p p , ( ), , ( )
ALL RAOBs and ALL Instrument Types Were Selected (Total Number of Site/Type:1093)
(Results will Improve Further If we do STATS with Selected RAOBs, Inst Types)
Land Sea Coast ALL
22876/41182 CLDCLR (56%) 11400/18811 CLDCLR (60%) 33158/61914 CLDCLR (54%) 67434/121907 CLDCLR (55%)
237/372 CLR (63%) 1995/2369 CLR (84%) 2874/4068 CLR (70%) 5106/6809 CLR (74%)
9
10. IASI Statistics – RMS - Global - Sea
Yield : 60% , NSAMP: 11,400
50
100
100
200
Pressure (hPa)
(
400
300
600
1000
0
1000
0 1 2 3 4 0 20 40 60 RMS
RMS Difference: Left Panel for Temperature (K), Right Panel for Water Vapor (%)
(K)
RAOB vs. IASI-RET ECMWF, NCEP-GFS, ATOVS
IASI- ECMWF, NCEP-GFS,
_______ ----------- -------------- ______ 10
11. 2. Data Sets for CrIS/ATMS Proxy Data
Generation and EDR Product Evaluation
(b) Focus Day Data Sets
Consists of
236 Granules of Matched Datasets for the
‘Focus Day’
» October 19, 2007 (10/19/2007)
» Each Granule Contains 22 or 23 Scan
Lines of
– CrIS/ATMS Proxy SDRs
– IASI/AMSU-A/MHS SDRs
IASI/AMSU-
– CrIMSS EDR products
– IASI EDR Products from NOAA IASI
Operations (NUCAPS)
– NCEP-GFS and ECMWF Analysis
NCEP- Approximate Granule Locations
Fields Size Not to Scale
Two More Focus Days (10/19/2008,
05/11/2010) of Data is on the Way
12. IASI–to-
IASI–to-CrIS
IASI-to-CrIS: Proxy Data Algorithm(Xu Liu and Kizer, SOAT Meeting, 2009)
12
13. Proxy Data Verification – IASI-> CrIS
IASI-
Figure 4. IASI observed brightness temperature spectra (top) and the corresponding CrIS proxy brightness temperature spectra (bottom).
The IASI instrument has 8461 IR channels spanning the IR spectrum 645-2760 cm-1. The instrument has 4 fields of view (FOVs) for each
Field f R
Fi ld of Regard (FOR) and the radiances are Gaussian apodized. The CrIS instrument has a total of 1315 IR channels in 3 bands covering
d d h di G i di d Th C IS i h l f h l i b d i
longwave (655-1095 cm-1), midwave (1210-1750 cm-1), and shortwave (2155-2550 cm-1) channels with spectral gaps between the bands.
The instrument has 9 FOVs for each FOR and the radiances are Hamming apodized. The plot is generated from the first FOR of the granule
139 SDR file. Similar plots for each granule (the first FOR in the SDR files) can be viewed at the web site that is currently under 13
construction (http://www.star.nesdis.noaa.gov/smcd/spb/iosspdt/iosspdt.php).
14. IASI–to-
IASI–to-CrIS Xu Liu’s Proxy Data Algorithm
Results for the Focus Day, 10/19/2007
962.5 cm-1 1232 cm-1
IASI IASI
CrIS CrIS
DIF MAP
DIF MAP
Similar figures for many other frequencies over the CrIS spectrum are available 14
15. AMSU/MHS To ATMS Proxy Data
/ y
MHS
AMSU-A/MHS -> ATMS Proxy Algorithm(Blackwell, SOAT Meeting, 2009)
15
16. AMSU-
AMSU-A/MHS to ATMS
Results for the Focus Day, 10/19/2007
23GHz and 52.48 GHz
23 GHz 52.48 GHz
AMSU CH 1
AMSU CH 5
A
A
ATMS CH 1
ATMS CH 6
H
H
DIF MAP
DIF MAP
F
F
Similar figures for all the AMSU-A/MHS and ATMS channels are available 16
17. AMSU vs. ATMS Proxy Data Stats for
“The Focus Day”, 10/19/2007
y / /
QV vs.QH
17
18. Microwave Integrated Resource System (MIRS)
Retrieval QC – ATMS Proxy vs. Real Observations
NOTE: MIRS Retrievals perform empirical bias-tuning to proxy-ATMS using
ECMWF before using them to generate MIRS-NPP-ATMS Proxy Retrievals. 18
Thanks to : Kevin Garrett, NOAA/STAR
19. CrIMSS Retrieval Comparisons with
IASI Retrievals and ECMWF
( )
(1)In p
(1)In performing these comparisons We haven’t still
g p
analyzed the Quality Flags for the CrIMSS
Retrievals.
• IASI QC flag was used to pick the corresponding
CrIMSS retrieval.
• This could be slightly ‘un-fair’ for CrIMSS EDR product
‘un-
in the
i th sense th t a profile accepted b th IASI system
that fil t d by the t
might have got rejected by the CrIMSS EDR QC.
Nevertheless, this analysis provides some in-sight into
in-
CrIMSS retrievals.
retrievals
(2)The
(2)The CrIMSS Algorithm may require ‘empirical bias
corrections’. We haven’t yet done that.
19
20. CrIMSS Retrieval Comparisons with
IASI Retrievals and ECMWF (STDEV)(Granule 139, SH_TR, Sea)
(STDEV)(Granule
‘Using IASI-QC and CrIMSS Not Tuned’
IASI-
50
100
00
10
200
Pressur (hPa)
re
400
300
00
60
000
000
10
10
0 1 2 3 4 0 20 40 60
Left Panel for Temperature, Right Panel for Water Vapor
STDEV :ECMWF vs.. IASI (IR + MW) (Solid Red)
( ) ( )
SDDEV : ECMWF vs.. CrIMSS (IR+ MW) (Dotted Red)
STDEV : ECMWF vs.. IASI (MW) (Solid Green)
STDEV : ECMWF vs.. CrIMSS (MW) (Dotted Green 20
N: 588/660 %Accepted (CLDCLR) : 89%, % “CLEAR”:27% (IASI Minimum CLD Amount)
22. Synergetic Use of MIRS Retrievals with CrIMSS (MW)
EDRs Could Improve/Resolve some Issues.
MIRS Retrievals Surrounding Granule 238
ATMS Proxy Data – MIRS Retrieval
MetOP AMSU/MHS
MetOP-AMSU/MHS Obs. MIRS Retrieval
Proxy ATMS Ret.
Takes More
k
Iterations at L/S
Boundary 22
23. IASI/CrIMSS Retrievals - RAOB/ECMWF Locations
(I have STATS for TROP/Mid-LAT/POLAR and for land/sea/all, etc..)
TROP/Mid-
‘Using IASI-QC and CrIMSS Not Tuned’
IASI-
Y ield w ith E very M atch of R A O B (R S -90,R S -80, etc.) and A L L IN S T. T ypes
Land S ea C oast A LL
832/1355 = 61% 364/618=59% 1051/1982=53% 2247/3955=57%
7/31 C lear (0 % ) 71/99 C lear (20% ) 104/138 C lear(9% ) 182/317 C lear(8% )
About 12 days of RAOB matched data sets were processed to
generate IASI and CrIMSS EDRs Yield shown here are IASI-
EDRs.
Ret System Yield. We are still implementing CrIMSS QC Flags
to generate corresponding statistics. 23
24. IASI & CrIMSS STATS : Sea D+N – STDEV
‘Using IASI-QC and CrIMSS Not Tuned’
IASI-
50
100
00
10
200
Pressur (hPa)
re
400
300
600
0
000
000
10
10
0 1 2 3 4 0 20 40 60
STDEV Difference: Left Panel for Temperature, Right Panel for Water Vapor
RAOB vs.. IASI-RET IASI-MIT CrIMSS-(IR+MW) CrIMSS (MW)
vs.. IASI- IASI- CrIMSS-
_______ ______ ----------------------- ------------------ 24
25. IASI & CrIMSS STATS: CLDCLR vs.. “CLR”
‘Using IASI-QC and CrIMSS Not Tuned’
IASI-
50
The fact that CrIMSS is coming closer in ‘CLR’
100
Cases with IASI suggest that in so-called ‘clear
cases’ the algorithm is not dependent on ATMS
temperature sounding channels. Also, suggests f
p g gg for
00
10
ALL Cases, ATMS may require bias corrections.
200
Pressur (hPa)
re
400
300
600
0
000
000
10
10
0 1 2 3 4 0 20 40 60
STDEV Difference: Left Panel for Temperature,
Temperature Right Panel for Water Vapor
RAOB vs.. IASI-RET(CLR),CrIMSS-(CLR) ,IASI-RET(CLDCLR) ,CrIMSS-RET(CLDCLR)
vs.. IASI-RET(CLR),CrIMSS- ,IASI- ,CrIMSS-
25
____________ ------------------- _____________ , ------------------------
26. Conclusions
1. Proxy Data is Good.
Evaluation of proxy data sets reveals that the proxy data have
reached a scale of perfection on ‘where they need to be’ for EDR
p y
product generation, and evaluation of EDR products with truth
measurements.
2. Need bias-tuning for MW (ATMS) component (CrIS as well).
bias-
The first MW retrieval is the basis for generating initial cloud-cleared
cloud-
radiances. We believe that the biases observed with the MW retrievals
are propagating into the cloud-cleared radiances and making the IR+MW
cloud-
retrievals biased with respect to ECMWF and RAOB measurements
3. The IR and MW emissivity Verification
Emissivity t i
E i i it retrieval i an i t
l is intermediate product i th C IMSS EDR
di t d t in the CrIMSS
algorithm. However, this is an important product that could characterize
AVTP and AVMP products.
4. CrIMSS Meeting the Specs – (Future Directions)
» We
W expect the CrIMSS EDR algorithm t meet the AVTP and AVMP
t th C IMSS l ith to t th d
product specifications with the updated version (LaRC 1.5.1.2)
• Latest IR and MW emissivity/ LUTs
• Improved CrIS Noise characteristics
p
• Empirical Bias Corrections for the ATMS (and CrIS)
– We have the required data sets to perform such analysis.
analysis. 26
27. Backup Slides
Thank You f
Th k Y for your Attention
Att ti
Suggestions/Questions/Comments/
Data Availability/Queries
y
Chris.Barnet@noaa.gov
Murty.Divakarla@noaa.gov
The NOAA NPOESS/IASI/AIRS Team
Integrated Observing System Science & Product Development Team (IOSSPDT)
NOAA/NESDIS Camp Spring, MD, USA
C S i MD
Overseen By
Mitch Goldberg and Chris Barnet