New Concepts In Retrieving Aerosol Properties Using MISR

Since March 2000 the nine camera Multi-angle Imaging SpectroRadiometer (MISR) aboard NASA's EOS Terra platform has been providing information about aerosols over both land and ocean. During this period many incremental improvements to the individual ocean and land aerosol retrieval algorithms have been made but the fundamental ideas behind each have remained essentially unchanged. Here we explore some new algorithmic concepts, multiangular in nature, which may provide a considerable increase in the accuracy of retrieved aerosol properties from space. The current MISR retrieval algorithm over ocean nominally utilizes only the red (672 nm) and near IR (866 nm) spectral bands, assuming that neither band has any significant contamination from water-leaving radiance (WLR). This approach provides a good determination of aerosol optical depth but the retrieved Angstrom exponent is subject to much more uncertainty because of the relatively small wavelength separation of the red and near IR bands. The concept being explored for improving the ocean algorithm is to also include the remaining blue (446 nm) and green (558 nm) MISR bands under the assumptions that 1) only the near IR band has near-zero WLR and 2) the WLR in the remaining three bands is isotropic. An algorithm with these conditions should provide a more accurate retrieval of aerosol properties and, simultaneously, the retrieval of WLR (ocean color). Over land the current aerosol retrieval algorithm is composed of two parts. The first is an angular shape comparison of the directional surface reflectance among the four MISR spectral bands, testing for similarity, a constraint that filters out the least probable aerosol models in the retrieval process. This procedure is then followed by a principal component analysis of the change in surface contrast with view angle and the final selection of retrieved aerosol models. This algorithm has produced high quality retrievals of aerosol optical depth over a wide variety of terrain including bright deserts. To obtain greater sensitivity to other aerosol properties, such as type and particle size, we are experimenting with incorporating an albedo effect on the shape comparison test using a new semi-empirical surface model and having better determinations of the surface directional reflectance which are less corrupted by diffuse sunlight. We will show results from experimental algorithms incorporating these new ideas. Comparisons will be made with retrieved aerosol properties from the current MISR operational algorithms, supplemented with aerosol results from AERONET when available. For both ocean and land, it is expected that there will be better discrimination among different aerosol types (e.g., smoke and dust) and greater sensitivity to particle size (i.e., more accurate Angstrom exponents). This work was performed at the Jet Propulsion Laboratory, California Institute of Technology under contract with the National Aeronautics and Space Administration.