HR: 0800h
AN: B51B-0202    [Abstracts]
TI: Errors in Retrieved Land Surface Temperature Due to Anistropic Emissivity Over Structured Vegetation Canopies
AU: Privette, J L
EM: jeff.privette@nasa.gov
AF: NASA Goddard Space Flight Center, Code 614.4, Greenbelt, MD 20771
AU: * Yu, Y
EM: yyu@pop900.gsfc.nasa.gov
AF: George Mason University at NASA Goddard Space Flight Center, Code 614.4, Greenbelt, MD 20771
AU: Pinheiro, A C
EM: ana@hsb.gsfc.nasa.gov
AF: National Research Council, NASA GSFC, Code 614.3, Greenbelt, MD 20771
AU: IP, J
EM: justin.ip@ngc.com
AF: Northrop Grumman Space Technology, One Space Park Dr, Redondo Beach, CA 90278
AB: Split window (SW) land surface temperature (LST) algorithms are widely used for operational satellite products due to their simplicity and robustness. Most of these algorithms assume that the land surface has an isotropic apparent emissivity. That assumption is erroneous for two reasons: 1) homogeneous surface components have intrinsic directional emissivity distribution functions (DEDFs), and 2) most land pixels are composed of multiple components (e.g., trees, grass, soil) with differing emissivities, such that their fractional contribution to the sensor observation varies with the view zenith angle. In this work, we studied the error due to the second effect (only) when isotropic emissivity is assumed in the retrieval algorithm. We simulated DEDFs of structured vegetated surfaces using the modified geometric projection (MGP) model (Pinheiro et al., 2004). We used the resulting DEDFs in the MODTRAN radiative transfer model to estimate top-of-atmosphere brightness temperatures as would be observed by AVHRR, MODIS and VIIRS. Using published SW LST algorithms (including the backup NPOESS VIIRS algorithm), we determined the LSTs that would be retrieved assuming both isotropic and actual anisotropic surfaces. We found that surface emissivity can change significantly from nadir to the edge of the scan. The changes are dominated by the ratio of background to tree crown visible to the sensor, as well as the leaf area index. If the retrieval algorithm does not account for this effect, it can impose LST errors exceeding 1 K. The error increases with view zenith angle, but is independent of solar zenith angle. We describe a practical approach toward accounting for DEDF effects imparted by structured land surfaces in global operational algorithms, including the VIIRS LST EDR.
DE: 0416 Biogeophysics
DE: 0426 Biosphere/atmosphere interactions (0315)
DE: 0430 Computational methods and data processing
SC: Biogeosciences [B]
MN: Fall Meeting 2005