A Study of Surface Directional Reflectance Properties To Enhance Aerosol Retrieval Capability Over Land Using MISR Data

The nearly-simultaneous multiangle, multispectral,radiometrically calibrated imagery of the Multi-angle Imaging SpectroRadiometer (MISR) has a nominal spatial resolution of 1.1 km and covers the globe in about 9 days. Once the imagery is co-located and co-registered, an aerosol retrieval is performed, over both land and ocean, using an aerosol model look-up database. The technique for aerosol retrieval over ocean is conventional, namely assuming that measurements in the red and near-IR spectral bands are measurements of radiance scattered only within the atmosphere. Over land, however, the radiance measurements generally are a combination of atmosphereric and surface scattering events, the proportions which vary with wavelength and usually are not known a priori. This makes the retrieval of aersosls over land a much more intractable process. In fact any retrieval of aerosol properties over land from space with a passive instrument requires some constraints to be placed on the surface reflectance properties so that atmospheric radiance can be effectively separated from surface reflected radiance in the measurements. To facilitate the MISR standard aerosol retrieval process over land, it is assumed that the surface directional reflectance at any given location has the same (or very similar) angular form or shape in the different spectral bands. There is some theoretical basis for this assumption, especially when the surface spectral albedos have similar values, but an empirical verification in the context of multiangle remote sensing data is necessary if further progress in aerosol retrieval quality over land is to be made. This poster presents some results of a study to test the surface directional reflectance spectral similarity assumption. It focuses on MISR data taken over a number of AERONET sunphotometer sites with different surface conditions, ranging from urban areas to forested regions, at a spatial scale of 1.1 km. In contrast to MISR data alone, the AERONET data provide an independent and better contrained determination of the aerosol properties at a site during the overpass, which then is used to correct the associated MISR top-of atmosphere imagery for atmospheric effects, resulting in the best estimates of the AERONET site surface spectral directional reflectance at 1.1 km resolution. To understand how the similarity of the angular shape depends on spatial scale, the directional reflectance was retrieveded at a variety of spatial resolutions, starting at 1.1 km pixel centered at the AERONET site and was systematically increased by pixel averaging around the site to 17.6 km resolution, the spatial scale used by the current MISR operational aerosol retrieval. A wide variety of AERONET sites were analyzed to provide information on how the degree of spectral reflectance similarity may relate to surface type. Because MISR data has been available since early 2000 to the present, seasonal and secular trends in surface reflectance variability also were investigated. The similarity condition was quantified at each site by the use of various semi-empirical directional reflectance models which allowed spectral albedo effects to be explicitly taken into account. It is expected that the results of this study will improve the current capability of the MISR aerosol retrieval algorithm over land. This work was performed at the Jet Propulsion Laboratory, California Institute ofTechnology under contract with the National Aeronautics and Space Administration.