1 / 22

Effect of consistent CRTM coefficients on M-O bias and DDs in MICROS

Effect of consistent CRTM coefficients on M-O bias and DDs in MICROS. Xingming Liang and Alexander Ignatov. Motivation. Cross-platform inconsistencies are at least in part due to inconsistent CRTM coefficients for different sensors/platforms

gerd
Download Presentation

Effect of consistent CRTM coefficients on M-O bias and DDs in MICROS

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. Effect of consistent CRTM coefficients on M-O bias and DDs in MICROS Xingming Liang and Alexander Ignatov CRTM Coefficients in MICROS

  2. Motivation • Cross-platform inconsistencies are at least in part due to inconsistent CRTM coefficients for different sensors/platforms • Sensitivity analysis of CRTM Coefficients is critically important for understanding and minimizing model effect on MICROS double differences MODIS CRTM Coefficients Recalculated Metop-B CRTM Coefficients: Calculated Correctly? from www.star.nesdis.noaa.gov/sod/sst/micros CRTM Coefficients in MICROS

  3. CRTM Coefficients Analyzed in MICROS • 4 consistent CRTM Coefficients datasets were provided by Yong Chen and analyzed in MICROS using 1 day of global data (15 Jan 2013): • ODAS-ORD, ODAS-PW, ODPS-ORD, and ODPS- PW • Based on the same baseline LBLRTM v11.7 • All include CFC absorption CRTM Coefficients in MICROS

  4. Methodology • Compare 4 different results using 3 metrics • Global M-O biases. The expectation is that for better coefficients, global M-O biases are smaller • Global SDs of M-O biases. The expectation is that for better coefficients, STDs are smaller • Double differences for two pairs of platforms flying close orbits (For Hi-res, it’s Metop-A and -B, and NPP and Aqua. For GAC, it’s Metop-A and -B and NOAA-18 and -19). The expectation is that for better coefficients, DDs are smaller and more consistent across different pairs CRTM Coefficients in MICROS

  5. AVHRR GAC (IR37) Δ1 = N18 minus N19 Δ2 = M-A minus M-B σ >> IR37, ODAS_ORD IR37, ODPS_ORD Δ1=+0.133 K Δ1=+0.113 K Δ2=+0.082 K Δ2=+0.080 K σ > σ > σ >> IR37, ODAS_PW IR37, ODPS_PW Δ1=+0.127 K Δ1=+0.111 K Δ2=+0.101 K Δ2=+0.093 K CRTM Coefficients in MICROS

  6. Δ1 = Aqua minus S-NPP Hi-Res Sensors (IR37) Δ2 = M-A minus M-B σ ~ IR37, ODAS_ORD IR37, ODPS_ORD Δ1=+0.237 K Δ1=+0.223 K Δ2=+0.085 K Δ2=+0.086 K σ > σ > σ ~ IR37, ODAS_PW IR37, ODPS_PW Δ1=+0.155 K Δ1=+0.141 K Δ2=+0.105 K Δ2=+0.099 K CRTM Coefficients in MICROS

  7. Observation in IR37 • Global M-O biases • Generally smaller for PW by ~0.06 K (except for VIIRS, where it’s larger by ~0.05 K) • For GAC, generally smaller for ODPS • N-16 ODAS has a large anomaly up to 1K, looks much better for ODPS • SDs • SDs significantly smaller for ODPS for all GAC AVHRRs. Hi-res sensors are only minimally affected. • If this accuracy is deemed sufficient, then any combination of these options is good • DDs • Are insensitive to the choice of ORD-PW, ODAS-ODPS to within 0.02K for all platforms, except Aqua and NPP. For Aqua-NPP, DDs are very sensitive to the combination. This sensitivity is alarming (should we understand why??) smallest for ODPS-PW • If this accuracy is deemed sufficient, then any combination of these options is good • Recommendation • Overall, between SST team, CRTM team and sensor calibration team is needed to resolve cross-platform bias using ODPS-PW • Understand large anomaly in N-16 ODAS CRTM Coefficients in MICROS

  8. CRTM Coefficients in MICROS

  9. AVHRR GAC (IR11) Δ1 = N18 minus N19 Δ2 = M-A minus M-B σ > IR37, ODAS_ORD IR37, ODPS_ORD Δ1=+0.077 K Δ1=+0.074 K Δ2=+0.000 K Δ2=-0.003 K σ ~ σ ~ σ > IR37, ODAS_PW IR37, ODPS_PW Δ1=+0.071 K Δ1=+0.076 K Δ2=+0.013 K Δ2=-0.007 K CRTM Coefficients in MICROS

  10. Δ1 = Aqua minus S-NPP High Resolution Sensors (IR11) Δ2 = M-A minus M-B σ > IR37, ODAS_ORD IR37, ODPS_ORD Δ1=+0.052 K Δ1=+0.015 K Δ2=+0.005 K Δ2=0.000 K σ ~ σ ~ σ > IR37, ODAS_PW IR37, ODPS_PW Δ1=+0.006 K Δ1=+0.047 K Δ2=+0.017 K Δ2=-0.002 K CRTM Coefficients in MICROS

  11. Observation in IR11 • Global M-O biases • Larger in ODPS than ODAS by ~0.1 K for MODIS, by ~0.05 K for VIIRS, and by ~0.03 K for AVHRR. (why are opposed w/rt IR37 ?) • Comparable between PW and ORD • SDs • Smaller in ODPS than ODAS • Comparable between PW and ORD • DDs • Are insensitive to the choice of ORD-PW, ODAS-ODPS to within 0.02K for all platforms, except Aqua and NPP. • Have 0.4 K inconsistencies between AVHRR and MODIS&VIIRS. • Recommendation • Since the inconsistencies between AVHRR and MODIS&VIIRS are within 0.1 K in official MICROS, further analysis for new coefficients is needed. Specially for new addition – CFC absorption. CRTM Coefficients in MICROS

  12. CRTM Coefficients in MICROS

  13. AVHRR GAC (IR12) Δ1 = N18 minus N19 Δ2 = M-A minus M-B σ > IR12, ODAS_ORD IR12, ODPS_ORD Δ1=+0.020 K Δ1=+0.010 K Δ2=-0.227 K Δ2=-0.241 K σ ~ σ ~ σ > IR12, ODPS_PW IR12, ODAS_PW Δ1=+0.015 K Δ1=+0.003 K Δ2=-0.239 K Δ2=-0.226 K CRTM Coefficients in MICROS

  14. Δ1 = Aqua minus S-NPP High Resolution Sensors (IR12) Δ2 = M-A minus M-B σ > IR12, ODPS_ORD IR12, ODAS_ORD Δ1=+0.060 K Δ1=+0.071 K Δ2=-0.237 K Δ2=-0.225 K σ ~ σ ~ σ > IR12, ODPS_PW IR12, ODAS_PW Δ1=+0.067 K Δ1=+0.079 K Δ2=-0.234 K Δ2=-0.223 K CRTM Coefficients in MICROS

  15. Observation in IR12 • Global M-O biases • Larger by ~0.04 K in ODPS than ODAS • Generally smaller in PW compared to ORD • SD • Smaller for ODPS. • Comparable between PW and ORD • Consistencies: • All AVHRRs are consistent within 0.1 K, except for Metop-B, out of AVHRR family by ~0.25 K. • Inconsistencies between AVHRR and MODIS&VIIRS are ~0.4 K. • Recommendation • Since the inconsistencies between AVHRR and MODIS&VIIRS are within 0.1 K in official MICROS, further analysis for new coefficients is needed. Specially for new addition – CFC absorption. • Joint work between SST team, CRTM team and sensor calibration team is needed to resolve cross-platform bias between Metop-A and –B. CRTM Coefficients in MICROS

  16. CRTM Coefficients in MICROS

  17. Δ1 = N18 minus N19 AVHRR GAC (SST) Δ2 = M-A minus M-B σ ~ SST, ODAS_ORD SST, ODPS_ORD Δ1=-0.112 K Δ1=-0.111 K Δ2=+0.028 K Δ2=+0.028 K σ ~ σ ~ σ ~ SST, ODPS_PW SST, ODAS_PW Δ1=-0.111 K Δ1=-0.111 K Δ2=-0.028 K Δ2=+0.028 K CRTM Coefficients in MICROS

  18. Δ1 = Aqua minus S-NPP High Resolution Sensors (SST) Δ2 = M-A minus M-B σ ~ SST, ODAS_ORD SST, ODPS_ORD Δ1=+0.013 K Δ1=+0.014 K Δ2=+0.025 K Δ2=+0.024 K σ ~ σ ~ σ ~ SST, ODPS_PW SST, ODAS_PW Δ1=+0.013 K Δ1=+0.014 K Δ2=+0.025 K Δ2=-0.024 K CRTM Coefficients in MICROS

  19. Observation in SST • M-O mean biases, SDs and DDs are insensitive to the choice of ORD-PW, ODAS-ODPS for all platforms. CRTM Coefficients in MICROS

  20. CRTM Coefficients in MICROS

  21. Conclusion • 4 CRTM Coeffs. data set were used to check the sensitivities on the three metrics: M-O mean bias, SDs, and cross-platform consistency. • SDs are generally smaller in ODPS than ODAS, and in PW than ORD. • Overall, ODPS-PW is recommended to use for model simulation. • N-16 ODAS has a large anomaly up to 1K, looks much better for ODPS. Understanding this issue is needed. • Inconsistencies between AVHRR and MODIS&VIIRS are ~0.4 K. further analysis is needed. • Inconsistencies between Metop-A and –B are ~0.25 K in IR12. Joint work between SST team, CRTM team and sensor calibration team is needed to resolve cross-platform bias. CRTM Coefficients in MICROS

  22. Observation in IR37 - Old • Using ODPS instead of ODAS, • Mean biases were changed largest for High Res. for the PW cases (increased by 0.03 K, except for NOAA16), and slightly changed for other cases (changed by ±0.01 K) • SDs were improved for all GAC AVHRRs, but not affect high Res. Sensor significantly. • Using PW instead of ORD, • Mean biases were reduced by ~0.06 K for all cases, except for VIIRS, which increased by 0.05 K. • SDs were slightly improved for all sensors • NOAA16 increased by -1.0 K when ODAS_PW instead of ODAS_ORD, and by 0.2 K when ODPS_PW instead of ODPS_ORD due to out of band effect issue in Ch3B. • Consistencies: • All AVHRRs (except for NOAA16), MODIS/Terra, and VIIRS are consistent within ±0.1 K for all cases. • Inconsistency between Terra and Aqua are ~0.3 K, due to sensor calibration issue. CRTM Coefficients in MICROS

More Related