1. Characterizing and comparison of uncertainty in the AVHRR Pathfinder Versions 5 & 6 SST field to various reference fields Robert Evans Guilllermo Podesta’ RSMAS Mar 1, 2011 with special thanks to R. Reynolds for the provision of AVHRR OI reference fields and Chelle Gentemann for provision of AMSR fields

2. Description of LATBAND formulation and Matchup criteria LATBAND Latitude increments Pathfinder Version 6 (PF6) based on LATBAND formulation – 6 zonal bands 20 degrees wide centered on the Equator 2.5 degree wide transition band, linearly interpolated Coefficients estimated monthly, e.g. all January sat-in situ observations grouped, for each month of a year for a given sensor Matchup criteria – within 2km and ±30 minutes for buoy-satellite observation Skin temperature product SST retrievals validated against radiometer matchups (MAERI)

3. Use of homogeneity and channel difference to develop a probability tree to assign quality level – PF6 Q=0Q=1Q=4Q=3 Example of using a probability distribution approach to assign quality. Implementation of this approach will require exploration of multiple categories, e.g. small; large residuals

4. Rejection of cold retrievals by quality level Green=sat, Orange buoy Need to be able to identify buoys with small offset

5. Pathfinder Progress Pathfinder validated through comparison with buoys, Reynolds OI, AMSR Pathfinder Version 6 will be generated using LATBAND approach (six 20 degree wide zonal bands) and referenced to Reynolds daily, ¼ degree OI Quality tests have been improved over Version 5.0 Files will be in GHRSST format A Version 5.2 will be released in the near future using the Version 6 code base and the Version 5 SST retrieval equation coefficients Satellites to be processed: N7, 9, 11, 14, 16, 17, 18 Excluded Satellites: N12 (bad scanner), N13 (boom), N15 (out of family) Future added Satellites for Pathfinder V6: N19, Metop

6. Buoy Comparison of Pathfinder Versions 5 & 6

7. Pathfinder V6 compared to Buoys Nighttime only, 2005- 2009, N18 ± 0.1 0.4< 180 matches/mo < Median Residual Number of Residuals Std. Dev. Residual

8. Examples of Pathfinder 5 & 6 vs. reference fields Path V6 in red, Path V5 in blue N16 – Buoy N16 began to degrade in late 2003. Ability to process multi-year time series will support alternate calibration approaches (Mittaz-Harris) N18 – Reynolds OI (AVHRR only) N18 consistent with time until late 2009 Pathfinder 6 reduces seasonal anomalies wrt the reference field

9. Pathfinder Version 5.2 progress, N7-> N18 N15, Not used, out of family NIGHT N16 N18 N7 N9 N11 N14 Reload Missing orbits N7 & N9 for 1985 N11 to N14 gap 3 month N17 Start N16 8688

10. Comparison Products Processed ~5.4 years of NOAA-18, 2005-2010 ¼ degree daily Pathfinder – Reference field (DT) Comparison with each of 5 reference fields Three types of comparison Pathfinder – Buoys, using buoys not included in coefficient estimation Daily DT fields, to show coherent regions of large difference (order 0.5K) Latitude vs. Time plots, based on best quality zonal average of each DT field, to show persistent zonal differences and temporal trends

11. SST Reference File – 5 versions Richard Reynolds provided 3 versions of the ¼ degree, daily V2 OI OI + AVHRR (from NAVO, day + night satellite data) OI + AVHRR + AMSR (day + night satellite data) OI + AVHRR + AMSR (night only satellite data, minimize possible impact of residual diurnal warming) 3 Day AMSR composite (day + night), daily, ¼ degree field (RSS) AATSR (night only), based on 0.1 degree night only, 3 channel, dual view product (processing version: May, 2010) R. Reynolds processed AATSR daily fields into monthly, ¼ degree maps to fill gaps due to combination of narrow swath and cloudy observing conditions

12. 6 Year Night Pathfinder V6-Reference Comparisons 3day AMSR (day+night) Little difference at high lat, minimal zonal and temporal oscillation OISST: NAVO AVHRR High north lat summer, mid lat zonal oscillation (N) OISST: NAVO AVHRR+ AMSR Similar comparison for all OI versions Middle grey band ± 0.1K AATSR : Monthly Average Night – Dual view, 3 channel High lat not available in summer Pronounced seasonal zonal oscillation (N+S) N equatorial aerosol more pronounced

13. Impact of diurnal warming on comparison to reference fields Anomaly fields show significant difference when a single reference field is used for both day and night satellite fields Anomaly fields are very similar when reference field is temporally matched to satellite observation time Path V6 – 3day AMSR composite (single field for 24 hours) Path V6- Reynolds OI (single field for 24 hours) Path V6 – AMSR 3 day composite, separate day and night Day Night

14. Comparison of 1 day Pathfinder vs ref fields 3day AMSR reference OISST (NAVO AVHRR ) reference Monthly AATSR reference Night, Dual view, 3 channel July 19, 2009 N18 night Note differences, Med, high lat

15. Anomalies reasonably consistent across satellites although diurnal variability is present Monthly DT field, Pathfinder V6 – 3 day AMSR (separate Day and Night reference) for January 2006 N18-AMSR N Night N18-AMSR D Day N17-AMSR D Morning N17-AMSR N Evening Color step = 0.2K -.2 0.2

16. N18 PF6-AMSR (night only) monthly, best quality, 2006 Residual patterns evolve month to month MarJan MayJul SepNov

17. Conclusions Comparison with 1 field/day 3day AMSR fields shows smallest residuals, OI and AATSR show larger seasonal, zonal oscillation Use of reference fields temporally matched to the satellites minimizes the magnitude of residual patterns A reference field reflecting the diurnal cycle possibly the preferred approach Comparison with July night field shows differences between reference fields (High Lat, western N Pacific, Mediterranean Sea) Improved quality test reduced cloud and aerosol impact However, dust aerosol impact remains an issue

18. END

19. n18 06080 test. The left is avhrr-only-v2, the center is atsr, and the right is the sst.The northern coast and lakes, near the bottom, are there, but too cold to show them and the warmer water at the top at the same time. AVHRR ATSRReynolds OI