Huang_IGARSS2011_VIIRS_Aerosol_JH_20110727.ppt

  • 410 views
Uploaded on

 

More in: Technology
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
    Be the first to like this
No Downloads

Views

Total Views
410
On Slideshare
0
From Embeds
0
Number of Embeds
0

Actions

Shares
Downloads
27
Comments
0
Likes
0

Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
    No notes for slide
  • Comparison of VIIRS(ABI) and MODIS aerosol AOT retrievals with those from AERONET. The VIIRS(ABI) retrieval is an adaptation of the VIIRS aerosol algorithm for the Advanced Baseline Imager (ABI) to be flown on GOES-R. (Over water the VIIRS and ABI algorithms are (almost) identical. The VIIRS retrievals use gas-corrected 10-km “aerosol” reflectances from the standard MODIS aerosol product. This makes comparison of MODIS and VIIRS AOT straightforward since sampling and spatial resolution of the products are the same, differences in cloud masks and gas absorption are avoided. The MODIS and VIIRS(ABI) retrievals are compared to AERONET Level 2 data using the matchup procedure of Ichoku et al. (satellite retrievals from a 50-km square area around the AERONET site, and 1-hour average of AERONET measurements centered on the satellite overpass). The comparisons cover the period from 2000 to current date from Terra and from 2002 to current date from Aqua. All available ocean/island sites are used. Top row: scatter plot of satellite vs. AERONET AOT, color coded according to the fine mode fraction; MODIS on the left, VIIRS(ABI) on the right. The two heavy dashed lines represent the expected envelope of . The shorter dashed line is the linear fit; the thin solid line is the 1:1 to line. Bottom row: comparison of binned AOT, whiskers represent the standard deviations. MODIS and VIIRS(ABI) retrievals are very similar. VIIRS(ABI) has somewhat smaller overall bias, possibly due to better modeling of wind speed dependence of surface reflectance (fixed in MODIS, variable in VIIRS). Outliers at small AOT include both fine and coarse mode aerosols.

Transcript

  • 1. Characterization and Evaluation of NPP VIIRS Aerosol EDR Performance Based Upon Prelaunch Analysis N. Christina Hsu (NASA/GSFC) Istvan Laszlo (NOAA/STAR) Jingfeng Huang (NASA/GSFC, Morgan State University)* Myeong-Jae Jeong (Gangneung-Wonju National University, Korea) David O. Starr (NASA/GSFC) Heather Q. Cronk (IMSG at NOAA/STAR) Hongqing Liu (Dell Services at NOAA/STAR) Robert Holz (UW/PEATE) Min Oo (UW/PEATE) Acknowledgement to: Sid Jackson (NGAS) and other VIIRS Aerosol/Cloud Cal/Val Team Members for valuable feedbacks and discussions, and Raytheon algorithm conversation team for additional data support Email: Jingfeng.huang@nasa.gov
  • 2. VIIRS – Visible Infrared Imager Radiometer Suite VIIRS 24 EDRs Land, Ocean, Atmosphere, Snow * Product has a Key Performance attribute NPP Satellite CERES VIIRS CrIS ATMS OMPS Limb OMPS Nadir
  • 3. Outline
    • Part I. Sensor Performance:
    • 1.1 What is optical crosstalk?
    • 1.2 Method to evaluate VIIRS sensor crosstalk
    • 1.3 VIIRS Crosstalk Impact on Aerosol Retrieval
    • 1.4 Crosstalk Summary
    • Part II. Algorithm Performance:
    • 2.1 VIIRS-MODIS Comparison
    • 2.1.1 MODIS-Like NOAA/STAR VIIRS Product vs. MODIS
    • 2.1.2 MODIS-Like PEATE VIIRS Product vs. MODIS
    • 2.1.3 VIIRS-Like IDPS Product vs. MODIS
    • 2.2 VIIRS-AERONET Comparisons
    • Summary
  • 4. Quality of Coatings, Top of IFA From XtalkTim5_Mills.ppt by Steve Mills (NGST); Courtesy: Chris Moeller. Optical : Dominant VISNIR crosstalk. Linear with signal, spectral and spatial component. Believed to be largely due to filter defects. Magnitude primarily depends on number of defects in spectral filters (‘spits’). Defects from coatings of Integrated Filter Assembly cause large angle scattering. Part I: Sensor Performance 1.1 What is Optical Crosstalk? Quality of Coatings, Bottom of IFA
  • 5. 1.2 Crosstalk Impact - Evaluation Method
    • Assess the impact of VIIRS VisNIR xtalk on aerosol EDR using MODIS Deep Blue algorithm and MODIS standard Dark Target algorithm.
    • Investigate how aerosol property statistics changes using VIIRS proxy (MODIS) data with and without xtalk, based upon TV influence coefficients. Effects of out-of-band response and polarization sensitivity are not considered in this analysis.
    • Two scenes are selected to test aerosol EDR for a range of aerosol loadings from low to heavy plumes over both water as well as vegetated and arid land surfaces.
  • 6. Arabian Desert Scene 1.3 Crosstalk Impact on VIIRS AOT (Land, Deep Blue)
    • Crosstalk impact on Deep Blue AOT retrieval (land) is within 0.01.
    (a) RGB (b) AOT w/o XTalk (c) AOT with XTalk (d) Difference: (c)-(b) 5% 25% Mean 50% 75% 95%
  • 7. 1.3 Crosstalk Impact on VIIRS AOT (Ocean & Land, Dark Target) China Scene
    • Crosstalk impact on Dark Target AOT retrieval is within 0.005, for both land and ocean.
    (a) RGB (b) AOT w/o XTalk (c) AOT with XTalk (d) Difference: (c)-(b) 5% 25% Mean 50% 75% 95%
  • 8.
    • Without considering out-of-band response and polarization sensitivity, the impact of xtalk is moderate on SDR and aerosol EDR performance over land and ocean for both test granules.
    • In general, the effects of xtalk on SDR are within 0.1% for most of the wavelengths. However, they are larger for 470 nm (mostly contributed by the 443 nm channel) over land and ocean (~0.2%), as well as for 865 nm band over ocean (~0.4-0.6%).
    • The resulting errors in AOT due to xtalk are within 0.01 over land and ocean for both darker surface and bright desert scenes retrieved from Deep Blue and Dark Target algorithms, which meet the NPP requirements of aerosol EDR performance.
    1.4 Crosstalk Impact on VIIRS Aerosol - Summary
  • 9. Part II: Algorithm Performance 2.1 VIIRS-MODIS Comparison 2.1.1 MODIS-Like NOAA/STAR Product ( Science code Drop 4.9.3 ) (Note: MODIS data run through VIIRS algorithm with MODIS LUT at NOAA/STAR; close to final drop of 4.9.4) 2.1.2 MODIS-Like PEATE/LEOCAT Product ( IDPS build 1.5.0.48 ) (Note: MODIS data run through VIIRS algorithm with MODIS LUT at Atmosphere PEATE; It is a relatively old ops code build from 2009 based on Drop 4.9 ) 2.1.3 VIIRS-Like IDPS Product ( Final Drop 4.9.4 ) (Note: a quick glance of VIIRS-Like granules)
  • 10. 2.1.1 VIIRS (STAR) vs. MODIS: 2010.214-218,LAND 214 215 216 217 218 218, all QA
    • VIIRS (land) aerosol retrieval (drop 4.9.3) achieves comparable performances to MODIS, except some low biases when AOT is higher than 1.5;
    • Some high biases when AOT is less than 0.5 is also observed with all QA, but not significant with best QA.
  • 11. 2.1.1 VIIRS (STAR) vs. MODIS: 2010.214-218,OCEAN 214 215 216 217 218 218, all QA
    • VIIRS (ocean) aerosol retrieval (drop 4.9.3) achieves comparable performances to MODIS overall;
    • There are some systematic low biases from VIIRS to MODIS with both all QA and best QA.
  • 12. 2.1.1 VIIRS (STAR) vs. MODIS: 2010.218 (08/06) VIIRS 2010218 MODIS 2010218
    • Top panel: with all QA; bottom panel: with best QA.
    • 131 granules on 08/06/2010
    ALL QA ALL QA QA=3 QA=3
  • 13. 2.1.1 VIIRS (STAR) vs. MODIS: 2010.218 (08/06)
    • AOT Difference between VIIRS and MODIS are different for land and ocean;
    • Significant AOT discrepancy were found for high AOT values over land;
  • 14. 2.1.1 VIIRS (STAR) vs. MODIS: Heavy Aerosol Events Observation of the huge wildfire event over West Russia in August 2010: VIIRS (QA3) is significant lower than MODIS (QA3) when AOT > 1.5. VIIRS MODIS VIIRS MODIS 08/05/10 08/06/10
  • 15. 2.1.2 VIIRS (PEATE) vs. VIIRS (STAR) vs. MODIS VIIRS (STAR) vs. MODIS (Land) VIIRS (PEATE) vs. MODIS (Land) VIIRS (STAR) vs. MODIS (Ocean) VIIRS (PEATE) vs. MODIS (Ocean)
    • Data from 15 collocated Terra Granules
    LAND OCEAN 4.9  4.9.3 4.9  4.9.3
  • 16. 2.1.2 VIIRS (PEATE) vs. VIIRS (STAR) vs. MODIS MODIS VIIRS(STAR), Drop 4.9.3
    • There are significant discrepancy between MODIS and both drop versions of VIIRS (with best QA): Heavy aerosol and cloud screening
    VIIRS(PEATE), Drop 4.9
  • 17. 2.1.2 VIIRS (PEATE): Amazon Smoke, 10/2007 MODIS VIIRS MODIS VIIRS 2007274.1705 2007275.1750
    • More challenging heavy aerosol cases: Amazon smoke, 10/2007
  • 18. 2.1.3 VIIRS-Like (IDPS, Drop 4.9.4) vs. MODIS 09/06/2002, DOY249 Land Ocean
    • There is wavelength difference issue in this comparison: MODIS B3 is 470 nm, and VIIRS M3 is 488 nm, resulting in a possible 0.2 AOT difference over land (Sid Jackson, NGAS).
    80 second 400*96 6km resolution 5 minute 203*135 10km resolution
  • 19. 2.1.3 VIIRS-Like (IDPS, Drop 4.9.4) vs. MODIS 01/25/2003, DOY025 Land Ocean
    • There is wavelength difference issue in this comparison: MODIS B3 is 470 nm, and VIIRS M3 is 488 nm, resulting in a possible 0.2 AOT difference over land (Sid Jackson, NGAS).
  • 20.
    • Over water, the Advanced Baseline Imager (ABI) adopts VIIRS aerosol algorithm and uses MODIS VIS-NIR reflectances from 10-km MODIS aerosol product
    • VIIRS aerosol retrieval over ocean achieves comparable performances to MODIS
    Part 2: Algorithm Performance 2.2 VIIRS – AERONET Comparison VIIRS(ABI) AOT vs. AERONET AOT @ 126 AERONET ocean/island sites 2000-2009 (Results provided by Istvan Laszlo NOAA/NESDIS/STAR)
  • 21. SUMMARY
    • Sensor performance
    • VIIRS sensor optical crosstalk does not have significant impact on VIIRS aerosol EDR
    • Algorithm performance:
    • Use of latest VIIRS aerosol algorithm is essential in making the VIIRS-MODIS comparisons to evaluate algorithm performance (drop 4.9 vs. 4.9.3 vs. 4.9.4);
    • VIIRS (drop 4.9.3) achieves comparable results to MODIS over land and ocean when AOT<1.5; VIIRS seems lower than MODIS when AOT > 1.5.
    • There are still challenging issues with VIIRS algorithm, such as heavy aerosol observations and cloud screening;
    • The VIIRS(ABI, Ocean) algorithm also produces comparable AOT to the AERONET AOT observations.
  • 22. THANK YOU!
    • AQUA MODIS AOT Seasonal Climatology, to be continued by VIIRS
    VIIRS +MODIS Jingfeng Huang@NASA/GSFC: [email_address]
  • 23. Q: How comparable are the MODIS Dark Target AOT retrievals to the AERONET AOT retrievals? A: Aqua MODIS DT AOT (Best QA) vs. AERONET AOT (2002-2010). The expected error ranges are ±0.05±0.15AOT for Land, and ±0.03±0.05AOT for Ocean. Land Ocean Backup slide
  • 24. Backup slide: Algorithms VIIRS MODIS Aerosol Models 5 Static Aerosol Models (Dust, Smoke-High Absorption, Smoke-Low absorption, Urban-high absorption, Urban-low absorption), each with 1 fine and 1 coarse mode, model determined from observation Dust in combination with regional fine mode models,13 fine mode fraction Dynamic selection of 2020 possible over-ocean aerosol models, with 4 fine and 5 coarse modes and 101 fine mode fractions 200 combinations of 4 fine model, 5 coarse mode and 10 fine mode fractions, same models Input Bands 412, 445, 488, 672, 2250 nm 470, 667, 2130 nm 672, 746, 865, 1240, 1610, 2250 nm 466, 553, 667, 855, 1240, 1640, 2130 nm Cloud Mask Internal + VCM Internal Aggregation Conduct aerosol retrieval at IP level first, then aggregate AOT to EDR level (40% top, 20% bottom) Aggregate L1B surface reflectance to L2 resolution first, then do aerosol retrieval (50% brightest and 20% darkest for land, 25% and 25% for ocean)
  • 25. Backup slide: VIIRS Bands
  • 26. MODIS-Like VIIRS (10km) QA=best (3), using best QA (0) only from IP (1km) MODIS (10km) QA=3 VIIRS(PEATE): Heavy Aerosol Case on 2009.182.1500 (Aqua) VIIRS IP (1km), QA=best (0),degraded (1): It seems the heavy aerosol pixels were degraded. VIIRS IP Cloud Confidence Flag: It seems the region is fairly confidently cloud free. VIIRS IP bad SDR qualify flag: All seems good SDR. VIIRS IP AOT out of range quality flag: Only a small portion seems to have AOT out of range. The question is: why those heavy aerosol pixels were degraded from QA=0 to QA=1 at IP level? It seems VCM is not the cause, bad SDR is not the cause either, nor the AOT range, BUT THE VOLCANIC ASH! VIIRS IP VOLCANIC ASH