1 / 19

Preliminary Comparisons of ACSPO v2.21 vs. v2.20 in SQUAM

Preliminary Comparisons of ACSPO v2.21 vs. v2.20 in SQUAM www.star.nesdis.noaa.gov/sod/sst/squam/. Background. Cloud Mask in ACSPO 2.20 GAC was observed to be overly conservative Much more voids in GAS coverage, compared to FRAC

lorne
Download Presentation

Preliminary Comparisons of ACSPO v2.21 vs. v2.20 in SQUAM

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. Preliminary Comparisons of ACSPO v2.21 vs. v2.20 in SQUAM www.star.nesdis.noaa.gov/sod/sst/squam/ ACSPO v2.21 vs. v2.20

  2. Background • Cloud Mask in ACSPO 2.20 GAC was observed to be overly conservative • Much more voids in GAS coverage, compared to FRAC • # of clear-sky GAC pixels disproportionately low compared to FRAC (less than the expected ratio of ~15) • Despite smaller sample, GAC SST stats are inferior to those of FRAC • To address these observations, v2.21 was developed • Team discussed in mid-August • Boris made fixes to GAC CM (but not FRAC CM I understand) • John Stroup made a build • Xingming has implemented first with data of Aug 26. 2013 • Here only one day of 2.20 is shown (24 Aug 2013) and one day of 2.21 (3 Sep 2013) • data from 2.21 have been displayed in MICROS and SQUAM since 27 Aug 2013, so anyone interested can go there and see more ACSPO v2.21 vs. v2.20

  3. ACSPO v2.20 24 August 2013 ACSPO v2.21 vs. v2.20

  4. FRAC: ACSPO v2.20 (Night) - OSTIA 24 August 2013 • Many areas of the ocean are covered with data, in 1deg maps ACSPO v2.21 vs. v2.20

  5. GAC: ACSPO v2.20 (Night) - OSTIA 24 August 2013 • More data voids in GAC compared with FRAC ACSPO v2.21 vs. v2.20

  6. FRAC: ACSPO v2.20 (Night) - OSTIA 24 August 2013 • Shape close to Gaussian; Residual cloud – skewed left tail • Domain and Performance Stats close to expected ACSPO v2.21 vs. v2.20

  7. GAC: ACSPO v2.20 (Night) - OSTIA 24 August 2013 • GAC sample is ~20% smaller, compared to scaled FRAC • Shape & GAC SST performance stats degraded, compared to FRAC ACSPO v2.21 vs. v2.20

  8. FRAC: ACSPO v2.20 (Night) – in situ 24 August 2013 • Shape close to Gaussian; left tail indicates residual cloud • Domain and Performance Stats close to expected ACSPO v2.21 vs. v2.20

  9. GAC: ACSPO v2.20 (Night) – in situ 24 August 2013 • GAC sample smaller by factor of ~2 compared to FRAC • Shape & GAC SST performance stats degraded compared to FRAC ACSPO v2.21 vs. v2.20

  10. ACSPO 2.20 Night 24 Aug 2013 – Summary ΔT = “VIIRS minus OSTIA” SST (expected ~0) • (Scaled) GAC SST domain +81% of FRAC • Nevertheless, all SST Stats degraded ΔT = “VIIRS minus in situ” SST (expected ~0) • # of matchups expected to be comparable for GAC and FRAC • However, # of FRAC matchups is larger by a factor of ~2 • Nevertheless, all GAC SST Stats degraded, compared to FRAC ACSPO v2.21 vs. v2.20

  11. ACSPO v2.20 03 September 2013 ACSPO v2.21 vs. v2.20

  12. FRAC: ACSPO v2.21 (Night) - OSTIA 03 September 2013 • Daily coverage fairly complete, except some persistent cloudy areas ACSPO v2.21 vs. v2.20

  13. GAC: ACSPO v2.21 (Night) – OSTIA 03 September 2013 • GAC coverage seemed to have been improved compared to 2.20 • Still more data voids in GAC compared to FRAC ACSPO v2.21 vs. v2.20

  14. FRAC: ACSPO v2.21 (Night) - OSTIA 03 September 2013 • Shape close to Gaussian; left tail heavier due to residual cloud • Domain and Performance Stats close to expected ACSPO v2.21 vs. v2.20

  15. GAC: ACSPO v2.21 (Night) - OSTIA 03 September 2013 • GAC sample 102% of scaled FRAC • SST performance statistics further degraded from v2.20 as expected ACSPO v2.21 vs. v2.20

  16. FRAC: ACSPO v2.21 (Night) – in situ 03 September 2013 • Shape close to Gaussian; left secondary peak suggests residual cloud • SST performance statistics seems deteriorated from 2.20 (maybe a different day?) ACSPO v2.21 vs. v2.20

  17. GAC: ACSPO v2.21 (Night) – in situ 03 September 2013 • GAC match-up size very comparable to FRAC • GAC SST statistics degraded more from FRAC and both degraded from v2.20 ACSPO v2.21 vs. v2.20

  18. ACSPO 2.21 Night 03 Sep 2013 – Summary ΔT = “VIIRS minus OSTIA” SST (expected ~0) • (Scaled) GAC SST domain 102% of FRAC • As expected, all SST Stats degraded more than in 2.20 ΔT = “VIIRS minus in situ” SST (expected ~0) • # of matchups comparable for GAC and FRAC, as expected • All GAC SST Stats are more degraded, compared to FRAC ACSPO v2.21 vs. v2.20

  19. GAC sample scales better with FRAC now Confirmed by L4 analyses Confirmed by validation against drifters Still, more data voids in GAC maps compared to FRAC (although the difference in coverage somewhat reduced) SST GAC statistics degraded more compared to FRAC Is it OK to trade for coverage? Overall ACSPO 2.21 SST stats degraded from 2.20 for both FRAC and GAC Maybe just the effect of using different day May need to re-run 2.20 and 2.21 for the same day Boris will be back the week of 16 Sep 2013 – can get together and discuss as needed Preliminary Observations ACSPO v2.21 vs. v2.20

More Related