1 / 26

Towards including M*D28 in SQUAM: Preliminary test using 1 day of data

M*D28 vs. ACSPO. Towards including M*D28 in SQUAM: Preliminary test using 1 day of data Sasha Ignatov, Xinjia Zhou, Prasanjit Dash. Objective. Include M*D28 SST in SST Quality Monitor http ://www.star.nesdis.noaa.gov/sod/sst/squam /

selia
Download Presentation

Towards including M*D28 in SQUAM: Preliminary test using 1 day of data

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. M*D28 vs. ACSPO Towards including M*D28 in SQUAM: Preliminary test using 1 day of data Sasha Ignatov, Xinjia Zhou, Prasanjit Dash M*D28 vs ACSPO

  2. Objective • Include M*D28 SST in SST Quality Monitor http://www.star.nesdis.noaa.gov/sod/sst/squam/ • Monitor in near-real time and back-fill from http://oceandata.sci.gsfc.nasa.gov/ • 1 day of global data (1 Oct 2013) processed in SQUAM and compared to ACSPO (stratified by Day/Night) • SST domain • Performance statistics relative to Reynolds SST and in situ SST from http://www.star.nesdis.noaa.gov/sod/sst/iquam/ M*D28 vs ACSPO

  3. Daytime Analyses M*D28 vs ACSPO

  4. Day: ACSPO L2 minus Reynolds L4 1 October 2013 • Deviation from Reference SST is flat & close to 0 • Residual Cloud/Aerosol leakages seen as cold spots M*D28 vs ACSPO

  5. Day: MOD28L2 minus Reynolds L4 1 October 2013 • Narrow swath width • More Cloud leakages than in ACSPO • Larger bias in high latitude M*D28 vs ACSPO

  6. Day: ACSPO L2 minus Reynolds L4 1 October 2013 • Shape close to Gaussian • Domain & Performance Stats close to expected M*D28 vs ACSPO

  7. Day: MOD28 L2 minus Reynolds L4 1 October 2013 • MOD28 sample -14% smaller compared to ACSPO • Comparable Mean/Med, 0.5K larger STDV/RSD M*D28 vs ACSPO

  8. Day: ACSPO L2 minus Reynolds L4 1 October 2013 • Shape close to Gaussian but there is a cold tail – residual cloud • Performance Stats better than specs M*D28 vs ACSPO

  9. Day: MOD28 L2 minus Reynolds L4 1 October 2013 • MOD28 match-up data set -16% smaller compared to ACSPO • increased Min/Max, STDV/RSD & Larger fraction of outliers M*D28 vs ACSPO

  10. Day 1 October 2013 – Summary ΔT = “MODIS minus Reynolds” SST (expected ~0) • MOD28 SST domain is -14% smaller, all Stats are degraded, compared to ACSPO ΔT = “MODIS minus in situ” SST (expected ~0) • IDPS SST in situ match-up is -16% smaller, all Stats compared to ACSPO M*D28 vs ACSPO

  11. NightAnalyses M*D28 vs ACSPO

  12. Night: ACSPO L2 minus Reynolds L4 1 October2013 • Deviation from Reference SST is flat & close to 0 • Residual Cloud/Aerosol leakages seen as cold spots • Warm spots show areas with diurnal warming during daytime M*D28 vs ACSPO

  13. Night: MOD28(SST4) L2 minus Reynolds L4 1 October 2013 • Narrow swath width • More Cloud leakages than in ACSPO • Larger bias in high latitude M*D28 vs ACSPO

  14. Night: ACSPO L2 minus Reynolds L4 1 October2013 (repeat) • Deviation from Reference SST is flat & close to 0 • Residual Cloud/Aerosol leakages seen as cold spots • Warm spots show areas with diurnal warming during daytime M*D28 vs ACSPO

  15. Night: MOD28(SST) L2 minus Reynolds L4 1 October 2013 • More coverage in high latitude than MOD SST4 M*D28 vs ACSPO

  16. Night: ACSPO L2 minus Reynolds L4 1 October2013 • Shape close to Gaussian but skewed positively due to diurnal warming • Domain & Performance Stats close to expected M*D28 vs ACSPO

  17. Night: MOD28(SST4) L2 minus Reynolds L4 1 October 2013 • MOD28 sample -34% smaller compared to ACSPO • Degraded STDV/RSD & Larger fraction of outliers M*D28 vs ACSPO

  18. Night: ACSPO L2 minus Reynolds L4 1 October2013 (repeat) • Shape close to Gaussian but skewed positively due to diurnal warming • Domain & Performance Stats close to expected M*D28 vs ACSPO

  19. Night: MOD28(SST) L2 minus Reynolds L4 1 October 2013 M*D28 vs ACSPO

  20. Night: ACSPO L2 minus Reynolds L4 1 October2013 • Shape close to Gaussian • Domain & Performance Stats better than spec M*D28 vs ACSPO

  21. Day: MOD28(SST4) L2 minus Reynolds L4 1 October 2013 • MODS28 match-up data set -43% smaller compared to ACSPO M*D28 vs ACSPO

  22. Night: ACSPO L2 minus Reynolds L4 1 October2013 (repeat) • Shape close to Gaussian • Domain & Performance Stats better than spec M*D28 vs ACSPO

  23. Day: MOD28(SST) L2 minus Reynolds L4 1 October 2013 M*D28 vs ACSPO

  24. Night 1 October 2013 – Summary ΔT = “MODIS minus Reynolds” SST (expected ~0) • MOD28 SST domain -35% smaller, SST4 is better than SST • Gap between Conventional and Robust stats wider in MOD28- More outliers ΔT = “MODIS minus in situ” SST (expected ~0) • MOD28 SST domain is -45% smaller, SST4 is better than SST • Gap between Conventional and Robust stats wider in IDPS - More outliers M*D28 vs ACSPO SST feedback to VCM 24

  25. Summary • Observations for both day and night • Degraded performance is in part due to narrower swath width • Day • MOD28 domain -14% smaller than ACSPO • Night • MOD28 domain -30% smaller than ACSPO, all SST stats degraded M*D28 vs ACSPO

  26. Questions • Sample number ratio between day time and night time is not close to 1. • Two night time products, SST and SST4, have different cloud mask and sample number. • Choose QL=0 only, is other QLs usable? M*D28 vs ACSPO

More Related