NOAA Assessment of the Oceansat-2 Scatterometer

1. NOAA Assessment of the Oceansat-2 Scatterometer Seubson (Golf) Soisuvarn, ZoranaJelenak and Paul S. Chang NOAA/NESDIS/StAR

2. OceanSat-2 Scatterometer- OSCAT • OSCAT is a Ku-band rotating pencil beam scatterometer with similarities to QuikSCAT • Developed by ISRO • Launched Oct 2009 on board of OceanSat-2 satellite

3. NRT OSCAT Data via ISRO and EumetSat

4. OSCAT Data Availability @ NOAA • NOAA has been receiving delayed (~24 hours) OSCAT data via ISRO FTP server since September 2010 • Timely OSCAT data flow from ISRO to EUMETSAT commenced in February 2011. • Since then EUMETSAT has been receiving approximately 10-14 orbits per day in a timely fashion • In March 2011 timely OSCAT data flow began at NOAA via EUMETSAT dedicated FTP server • ISRO currently working on providing all orbits in a timely fashion and working out procedures to operationally support the data flow • NOAA is currently receiving all three levels of OSCAT data: L1B, L2A and L2B • L2A and L2B (gridded @ 50 km WVC) 180min 150min Courtesy of T. Heinemann EumetSat

5. NRT OSCAT North Atlantic Data Gaps • Data gap in North Atlantic observed in ascending nodes • This data gap problem is a result of the definition of data-orbits from North Pole to North Pole. • The reception stations in Svalbard and Tromsø are not at the North Pole and due to technical constraints at the Oceansat-2 satellite, the satellite dump needs to be cut-off before the transmission starts. Therefore, the measurements between the reception stations and the North Pole are missing when the satellite reaches the reception station from the Atlantic side of the globe. • ISRO and EUMETSAT try to find a solution for this problem. Courtesy of T. Heinemann EumetSat NRT L2B data coverage for all received orbits in May 2011. High coverage indicated by red and yellow colors, low by blue and violet colors

6. Ongoing Efforts(In collaboration with JPL, KNMI and ISRO) Understand and characterize the L1B performance Develop and implement sigma0 corrections as appropriate Develop and implement NOAA unique L2 products to meet NOAA user requirements

7. Data Analysis Overview • The two latest L1B data were used in this analysis • February 2011 processing (OLD) • Implemented in current NRT processor • July 2011 processing (NEW) • There is currently no overlap period between the two data sets • Two different days were used in analysis, but approximately equal amount of data • Day 094, 2011 (OLD) – 14 orbits • Day 030, 2010 (NEW) – 15 orbits • All L1B QC flags were used except bit:4 (in order to show all slices) and bit:9 negative sigma0 flag • L1B latest known Discrepancy • L1B was matched up with NCEP model winds and NSCAT2 GMF was used to compute modeled sigma0 • Noise Subtraction Algorithm • Calculate Signal power and Noise power • Scan Angle Dependence • Ascending Descending Dependence

8. Noise Subtraction Power after noise subtraction Noise subtraction inconsistent among slices for pre-July ISRO OSCAT processing version Noise power OLD – Pre July 2011 NEW = July 2011 delivery zero line zero line

9. INNER BEAM OUTER BEAM OLD OLD Signal Power NEW NEW

10. Noise Subtraction – 2 Ps • Signal-to-noise ratio are improved at low wind speeds • In both versions signal power appear to saturate at ~ 3-4 m/s Pn OLD Ps Pn NEW

11. Negative Sigma0 • Low value of sigma0 anomalies with wind speed is now gone. OLD NEW

12. Sigma0 bias vs Scan Angle OLD – Pre July 2011 OUTER BEAM INNER BEAM Large biases between ascending & descending are still observed: ~ 1 dB for inner beam (H-pol) and ~ 0.5 dB for outer beam (V-pol) NEW Biases with respect to NSCAT-2 model function operating on GDAS winds.

13. Xfactorvs Scan Angle OLD – Pre July 2011 INNER BEAM OUTER BEAM

14. Noise Power vs Scan Angle OLD – Pre July 2011 OUTER BEAM INNER BEAM

15. ISRO L2B Data Analysis(Pre July 2011 Data Set )

16. ISRO OSCAT L2B Selected Wind Vectors Wind Speeds Wind Directions 4≤ws<24m/s ECMWF rms 2.08m/s GDAS rms 2.64m/s 4≤ws<24m/s ECMWF rms 18.24º GDAS rms m/s 20.42º 6<ws<24m/s ECMWF rms 18.04º GDAS rms m/s 19.84º 6<ws<24m/s ECMWF rms 1.76m/s GDAS rms 1.89m/s

17. Average zonal winds bias vs ECMWF for March-July 2011 ISRO OceanSAT-II Near Real Time Winds. Pattern likely due to cross track biases. Courtesy of SvetlaHristova-Veleva JPL

18. NOAA OSCAT Wind Data Processor • Process OSCAT data from Level 1B • Grid Sigma0 @ 25 km WVC (L2A) • MLE based wind retrieval • Median filter based ambiguity removal

19. ISRO orbit (50 km) NOAA orbit (25 km) ISRO derived WVC index (i,j) from satellite position and velocity vectors (not currently available in routine L1B processing ) Derived WVC index (i,j) by approximation from orbital elements given in L1B attribute parameters, i.e. Inclination = 98.28 deg Semi-major axis = 7098.14 km Eccentricity = 0.00113 Equator crossing longitude (descending node) = varied orbit-by-orbit -New version of OSCAT L1B files contain state vectors

20. NOAA 25km OSCAT L2A Product We calculate standard deviation of each Sigma0 from the following formulation σ01 σ03 σ02

21. TC Bret07/20/2011

22. TC Cindy07/21/2011

23. Hurricane Dora07/20/2011

24. Summary and Conclusions • OSCAT data has been flowing to NOAA in near real-time via EUMETSAT since March 2011 (not all orbits yet) • OSCAT L1B/L2A investigation shows: • High wind speed retrievals from OSCAT should definitely be possible • Sigma0 residual biases are significantly high at low wind speeds • Sigma0 are dependent on antenna scan angle • Sigma0 are also exhibit an Ascending & Descending pass dependence • Development of enhanced L2A product from ISRO’s L1B • 25 km WVC grid • L2A product will contain standard deviation of composite Sigma0 • Is proving to be useful parameter in definition of objective function normalization during retrieval process • OSCAT has the potential to provide a high utility product to largely mitigate the impacts due to the loss of QuikSCAT in NWS weather forecasting and warning products and services 24