Terrestrial biosphere analysis of SeaWiFS data over the Amazon region with MODIS and GLI prototype vegetation indices

A stream of multitemporal SeaWiFS (sea-viewing wide field-of-view sensor) data was extracted and analyzed over the Amazon region and surrounding land-community validation sites representing a wide range of vegetation conditions. The raw data was atmospherically-corrected for molecular scattering and ozone absorption and the resulting bidirectional reflectances were used to compute various ratio-based, orthogonal-based, aerosol resistant, and enhanced vegetation indices. Results were analyzed for both the tilted 20/spl deg/ look angle of SeaWiFS and the extrapolated nadir (0/spl deg/) look using a BRDF model. Various vegetation indices were analyzed to evaluate their usefulness for carbon modeling studies, discriminating canopy structure variations, and in depicting land-use/land cover changes in semi-arid cerrado, and high biomass rainforests (primary and secondary). Previous studies have indicated problems with vegetation indices in the Amazon region with respect to chlorophyll-saturation and asymptotic vegetation index behavior. The objectives of this work were to (1) prototype gridded vegetation map products for the EOS-MODIS (MODerate resolution Imaging Spectroradiometer) and ADEOS-Global Imager (GLI) sensors; (2) assess the utility of multiple vegetation indices over continental data fields; (3) analyze sensitivities over a global range of vegetation conditions (cerrado to rainforest); and (4) assess the relative usefulness of SeaWiFS data in depicting canopy structural and compositional differences across the Amazon region.