ACSPO (AVHRR Clear-Sky Processor for Oceans) 31 March 2008

1. ACSPO (AVHRR Clear-Sky Processor for Oceans)31 March 2008 ACSPO Evaluation Yury Kihai (STAR/QSS) – Programming Support, System Design, Operations XingMing Liang (STAR/CIRA) – CRTM, Physical SST, Aerosol Boris Petrenko (STAR/IMSG) – Clear-Sky Mask, Physical SST Prasanjit Dash (STAR/CIRA) – Global Cal/Val and QC/QA, MODTRAN RTM Denise Frey (OSDPD/QSS) – Operations, Programming Support Dilkushi De Alwis (STAR/CIRA) – Cal/Val vs. in-situ SST John Sapper and Sasha Ignatov Co-Leads, STAR & OSDPD ACSPO v.1

2. ACSPO Products • Primary: Clear-Sky Radiances (CSR) over Ocean in all sensor’ bands • Can be assimilated in NWP • Can be used for improved SST retrievals (e.g. JCSDA physical SST) • Quality of Clear-Sky Radiances will be monitored • SST (derived from CSR in the Earth emission bands) • Currently, regression SST only (AVHRR, MODIS, VIIRS heritage) • ACSPO is physical SST retrieval-ready (currently tested) • Hybrid Physical/Regression • Aerosol (derived from CSR in solar reflectance bands) • Single-channel Aerosol retrievals (current NESDIS operational algorithm) • Aerosol product will be used to • monitor accuracy of CSR in solar reflectance bands • Explore aerosol correction to SST • Assist in Cloud Mask evaluation and improvement ACSPO v.1

3. Examples of MUT and ACSPOinstantaneous SST anomaly images(“Retrieved SST – Reynolds weekly v.2 SST”) ACSPO v.1

4. METOP-A granule 1 1 January 2008 01-02 GMT (Nighttime) ACSPO v.1

5. METOP-A granule 3 1 January 2008 02-03 GMT (Nighttime) ACSPO v.1

6. 1 January 2008 04-05 GMT (Daytime) METOP-A granule 5 ACSPO v.1

7. Quality of Product / Service • Quality of ACSPO products • Higher density and equal or better accuracy than heritage product (see subsequent slides) • User readiness and acceptance of product • Users have had access to and have evaluated the product ACSPO v.1

8. Improved global coverage:ACSPO vs. MUT ACSPO v.1

9. Daytime SST Composite:3 January 2008 (MetOp-A) Heritage NESDIS SST product ACSPO SST product • Native pixel resolution 8 km • 6.6×104 SST pixels • 8.3% ocean covered at 1day×0.3o resolution • Native pixel resolution 4 km • 2.1×106 SST pixels • 32.7%oceancovered at 1day×0.3o resolution ACSPO v.1

10. Nighttime SST Composite:3 January 2008 (MetOp-A) Heritage NESDIS SST product ACSPO SST product • Native pixel resolution 8 km • 5.3×104 SST pixels • 12.8% ocean covered at 1day×0.3o resolution • Native pixel resolution 4 km • 2.8×106 SST pixels • 35.9%oceancovered at 1day×0.3o resolution ACSPO v.1

11. Global statistics of SST Anomaly“Retrieved minus Reynolds weekly v.2”ACSPO vs. MUT1-8 January 2008NOAA-17, -18, and MetOp-A ACSPO v.1

12. Anomalies in Heritage MUT SST vs. ACSPO, MetOp-A MUT ACSPO Night Day ACSPO v.1

13. Anomalies in Heritage MUT SST vs. ACSPO, NOAA-18 MUT ACSPO Night Day ACSPO v.1

14. Anomalies in Heritage MUT SST vs. ACSPO, NOAA-17 MUT ACSPO Night Day ACSPO v.1

15. Global statistics ofBrightness Temperatures Anomalies“CRTM simulated minus AVHRR measured”1-8 January 2008NOAA-17, -18, and MetOp-A ACSPO v.1

16. ACSPO v.1

17. ACSPO v.1

18. Insufficient statistics ACSPO v.1

19. Insufficient statistics ACSPO v.1

20. Angular Bias in ACSPO SST and BTs at swath edges(NB: MUT makes retrievals within ±54°whereas ACSPO uses full swath out to ±68°) ACSPO v.1

21. Nighttime SST Anomaly (“Retrieved – Reynolds”)3 January 2008 (MetOp-A) Heritage NESDIS SST product ACSPO SST product • Native pixel 8 km – gridded to 0.3o resolution • Covered 13.1% global ocean • Native pixel 4 km – gridded to 0.3o resolution • Covered 35.9% global ocean • Note cold bias in ACSPO nighttime SST at swath edges ACSPO v.1

22. ACSPO v.1

23. Daytime SST Anomaly (“Retrieved – Reynolds”)3 January 2008 (MetOp-A) Heritage NESDIS SST product ACSPO SST product • Native pixel 8 km – gridded to 0.3o resolution • Covered 8.3% global ocean • Native pixel 4 km – gridded to 0.3o resolution • Covered 32.7% global ocean • Note cold bias in ACSPO daytime SST at swath edges ACSPO v.1

24. Insufficient statistics ACSPO v.1

25. Anomaly as a function of Zenith Angle, MetOp-AY-axis: SST anomaly (SST-weekly Reynolds) 0.5 to -1.0 K ; X-axis: Satellite Zenith Angle -70  to 70 MUT ACSPO Night Day ACSPO v.1

26. Nighttime residual cloud(NB: cold bias in the “Roaring forties”in the Southern Hemisphere) ACSPO v.1

27. Anomaly Maps, MetOp-A, Night MUT 8 Days ACSPO 8 Days ACSPO Daily ACSPO v.1

28. Anomaly Maps, MetOp-A, Day MUT 8 Days ACSPO 8 Days ACSPO Daily ACSPO v.1

29. ACSPO version 1 & Ongoing version 2 improvements • Version 1: Create an efficient data infrastructure for new science & technology • Newly available features/functionalities realized in ACSPO version 1 • Fully integrated with NCEP GFS and CRTM • Coverage greatly improved, while maintaining or exceeding heritage accuracy • Extensive Cal/Val QC/QA integrated into ACSPO • Code stable and efficient (room for improvement exist) • Next phase ACSPO v.1 (July 2008) • processing FRAC (1km) MetOp-A data with ACSPO operational • New features/functionalities being explored for ACSPO v. 2 (Dec 2008) • Optimize Clear-Sky masks • Resolve view angle biases • Resolve nighttime cloud contamination • Hi-resolution 1km land-sea mask • Daily Reynolds SST (in place of weekly) to improve ACSPO Clear-Sky Mask • Separate out instrument-related code into a separate module (Flexibility with respect to input sensor data AVHRR 4km/1km, MODIS, VIIRS, ..) • Evaluate accuracy of CRTM Jacobian for Physical SST • Use fully consistent and uniform Fortran 90/95 coding standards, optimize code • Integrate ACSPO products with GHRSST L2P NetCDF ACSPO v.1