1 / 30

GEO-CAPE Ocean Science: Discussions on Uncertainties and Error Tolerance

This NASA GEO-CAPE Working Group Meeting aims to discuss the uncertainties in satellite-derived products and field-measured data, as well as the capability of current algorithms to meet GEO-CAPE mission goals. Various studies and observations will be presented to analyze the uncertainties and error tolerances in ocean science.

dglenn
Download Presentation

GEO-CAPE Ocean Science: Discussions on Uncertainties and Error Tolerance

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. GEO-CAPE Ocean Science: Discussions on Uncertainties and Error Tolerance NASA GEO-CAPE Working Group Meeting, NASA AMES, 21-23 May 2013

  2. Questions • What are the uncertainties in the satellite-derived products • What are the uncertainties in the field-measured data • Can current algorithms and processing approaches meet the requirements GEO-CAPE mission goals, e.g., resolving diurnal changes?

  3. Questions • What are the uncertainties in the satellite-derived products • How do we define uncertainties • RMS difference? Mean absolute difference? Spatial/Temporal scales • Rrs uncertainties • Global, regional, fixed locations (numerous papers) • Typically ~>10% between satellites and field Rrs for blue bands • Other products • Global ocean Chl: >50% (Gregg & Casey, 2004) • IOPs?

  4. Rrs uncertainties estimated from satellite data alone From Hu et al. (2013, RSE)

  5. Rrs uncertainties estimated from satellite data alone MODIS Rrs uncertainties at Chl = 0.1 mg/m3 412: 0.0065 sr-1 443: 0.0040 sr-1 488: 0.0035 sr-1 531: 0.0025 sr-1 547: 0.0020 sr-1 667: 0.0005 sr-1

  6. Rrs and IOP uncertainties from field measurements • Above-water and below-water Rrs • Could differ by > 20% (Toole et al., 2000) • Best estimates: 3% (Hooker, Chang, etc.) – do we have a consensus? • Chl and IOPs • 35% in Chl? • IOPs?

  7. Questions • Can current algorithms and processing approaches meet the requirements GEO-CAPE mission goals, e.g., resolving diurnal changes?

  8. Beam-C and Chl-a fluorescence in Tampa Bay (Chunzi Du MS thesis) Short-term changes of bio-optical properties

  9. Backscattering and Chl-a in Tampa Bay (Chen et al., 2010) Short-term changes of bio-optical properties December 2004 January 2005

  10. Backscattering and Chl-a in Tampa Bay (Chen et al., 2010) Short-term changes of bio-optical properties December 2004

  11. MVCO Time Series Observations in situ Hydroscat-6 AERONET-OC / SeaPRISM, IOPs from QAA algorithm Sosik, Feng, et al. 2010 3- 3 days

  12. MVCO in situ Hydroscat-6 AERONET-OC / SeaPRISM, IOPs from QAA algorithm 3-day period SeaPRISM Sosik, Feng, et al.

  13. Diurnal variability of water leaving radiance data and uncertainty in satellite retrievals in coastal locations Soe Hlaing, Alex Gilerson Optical remote-sensing lab, Electrical Engineering Department, NOAA-CREST, City College of the City University of New York

  14. AERONET-OC sites used in the study • AERONET-OC standard multi-spectral SeaPRISM instrument • Measurement sequences are executed every 30 minutes 12:00 PM ±4 hrs of local time • at OC bands: 413, 442, 491, 551 & 667nm. • The Long Island Sound Coastal Observatory (LISCO) • Located at approximately 3 km from the shore of Long Island near Northport, NY, USA. • Particulate backscattering coefficient at 551 nm is in the range of 0.01 to 0.03 m-1. • Total absorption coefficient at 442 nm varies from 0.38 to 1.2 m-1. • The absorption due to Colored Dissolved Organic Matter (CDOM) at 442 nm is typically close to 0.4 m-1 and in few cases can be as high as 1 m-1. • The WaveCIS site • Locate at approximately 18 km from the shore of Timbalier Bay area, MS, USA. • Particulate backscattering coefficient at 551 nm for WaveCIS water is usually around 0.01 m-1 but, in some rare cases, it reaches up to 0.04 m-1. • Unlike LISCO water, total absorption of the water body is low with its seasonal average value equal to 0.31 m-1 at 442 nm of which ~0.15 m-1 is attributed to CDOM.

  15. LISCO and WaveCIS sites Multispectral SeaPRISMinstrument LISCO site SeaPRISM instrument Instrument Panel Retractable Tower 12 meters • Sea Radiance • Direct Sun Radiance and Sky Radiance • Bands: 413, 443, 490, 551, 668, 870 and 1018 nm. WaveCIS site 15

  16. AERONET-OC retrieved nLw (λ) spectral in mW/cm2/µm/sr for February 5, 6, 7 & 9 of 2012 for LISCO location are shown in black dots. nLw of 413, 551 & 667nm wavelengths are shown. • Uncertainty (Ê) in VIIRS data is displayed as shaded box around the average nLw value of the day. • Vertical height of the shaded boxes represents the 2 times the uncertainty level (Ê) of the VIIRS data for LISCO location. • Horizontal width is roughly equivalent to SeaPRISM data acquisition time (±4hours of local noon). These shown nLw values and variability ranges are typical for LISCO location. Note: Level 1.5 AERONET-OC (SeaPRISM )nLw data shown here has been filtered for glint, cloud etc. for quality assurance. Thus not all measurements made within the day are included.

  17. AERONET-OC retrieved nLw (λ) spectral in mW/cm2/µm/sr for March 4, 5 & 6 of 2012 for WaveCIS location are shown in black dots. nLw of 413, 551 & 667nm wavelengths are shown. • Uncertainty (Ê) in VIIRS data is displayed as shaded box around the average nLw value of the day. • Vertical height of the shaded boxes represents the 2 times the uncertainty level (Ê) of the VIIRS data for LISCO location. • Horizontal width is roughly equivalent to SeaPRISM data acquisition time (±4hours of local noon). These shown nLw values and variability ranges are typical for WaveCIS location – low nLw. Note: Level 1.5 AERONET-OC (SeaPRISM )nLw data shown here has been filtered for glint, cloud etc. for quality assurance. Thus not all measurements made within the day are included.

  18. AERONET-OC retrieved nLw (λ) spectral in mW/cm2/µm/sr for November 14, 15 & 16 of 2012 for WaveCIS location are shown in black dots. nLw of 413, 551 & 667nm wavelengths are shown. • Uncertainty (Ê) in VIIRS data is displayed as shaded box around the average nLw value of the day. • Vertical height of the shaded boxes represents the 2 times the uncertainty level (Ê) of the VIIRS data for LISCO location. • Horizontal width is roughly equivalent to SeaPRISM data acquisition time (±4hours of local noon) . • These shown nLw values and variability ranges are occasionally observed for WaveCIS location- (high nLw). • Variability in 667nm appears larger than uncertainty level (especially for Nov 14 case) Note: Level 1.5 AERONET-OC (SeaPRISM )nLw data shown here has been filtered for glint, cloud etc. for quality assurance. Thus not all measurements made within the day are included.

  19. Summary from Hlaing and Alex Gilerson • Diurnal variability ranges of water leaving radiance data of LISCO & WaveCIS locations mostly fall within the VIIRS’s uncertainty levels of the respective sites. • However, it should be noted that VIIRS nLw data used in derivation of its uncertainty is average of 3x3 pixels box (~5-6 km2 area) whereas AERONET-OC measurements are collected from a few m2 area. • AERONET-OC nLw data are also averaged over ± 2 hours of satellite overpass time. • Thus, spatial & temporal variability in the satellite and in-situ data may significantly attribute to the derived uncertainty in VIIRS data.

  20. Error propagation and error tolerance From ACE whitepaper by Menghua Wang et al.

  21. Error propagation and error tolerance From ACE whitepaper by Menghua Wang et al.

  22. Error propagation and error tolerance From ACE whitepaper by Menghua Wang et al. Green: Chl; Red: CDM; Black: Bbp

  23. Error propagation and error tolerance New algorithms may help Daily SeaWiFS data from N. Atlantic gyre

  24. Error propagation and error tolerance Rrs errors are spectrally related

  25. Error propagation and error tolerance New algorithms may help SeaWiFS global statistics, and algorithm sensitivity to Rrs error

  26. Some remarks to stimulate discussion • Satellite Rrs uncertainties approaching the Gordon & Wang algorithm limit. These represent the lower bounds for coastal oceans • Field Rrs and IOP uncertainties – community consensus? • Diurnal changes in the examples are typically 50% - 100%. • Field Rrs can resolve these changes with existing algorithms. How about satellite Rrs? • Algorithm improvement in both atmospheric correction and bio-optical inversion – need to consider error tolerance • How do we move forward from now?

  27. Short-term changes in cyanobacteria bloom size

  28. Tidal variation of sediment A B Dec. 27, 2004 A B C C

  29. Rrs(645) (×10-2) (sr-1) Tidal variation of sediment Terra 10:46 am A B Dec. 27, 2004 A B C C

  30. Rrs(645) (×10-2) (sr-1) Tidal variation of sediment Terra 10:46 am Aqua 13:55 pm A B Dec. 27, 2004 A B C C

More Related