Evaluation of the MODIS vegetation index compositing algorithm using SeaWiFS data

Vegetation index data were composited in space and time to monitor vegetation changes in a spatial continuous fashion. Sixteen days of SeaWiFS (Sea-viewing Wide Field-of-view Sensor) data (surface reflectance data corrected for Rayleigh scattering Ozone and water vapor) were, composited to prototype and test the MODIS (MODerate resolution Imaging Spectroradiometer) algorithm for “standardized” MODIS vegetation indices. The SeaWiFS was tilted 20°, preventing nadir looks of the Earth´s surface. However, the MODIS algorithm applies a pixel-based bidirectional reflectance distribution function model (BRDF; Walthall) to obtain nadir-equivalent reflectance values at the prevalent solar zenith angle. The nadir-equivalent reflectance values were used to compute the normalized difference vegetation index (NDVI) and the enhanced vegetation index (EVI). A back-up algorithm (Maximum Value Composite; MVC) was applied if insufficient data were available to invert the BRDF model. The MODIS compositing algorithm was also executed using a standardization of the reflectance values to 20° view zenith angle (SeaWiFS tilt-angle), and compared to the “nadir-equivalent MODIS” algorithm and the MVC algorithm. To apply the MODIS compositing algorithm, a cloud mask was developed for the SeaWiFS data using a reflectance threshold for the visible bands. A land-water mask was used to retain only the land pixels. Anisotropy was found to be affecting the vegetation indices, although the temporal spectral profiles of cloud-free observations were quite variable, which was likely related to the inclusion of sub-pixel clouds in the “cloud-free” observations. The classical MVC method was computed to quantify the difference between the MODIS and MVC algorithms and establish a relationship for continuity purposes that will allow for comparisons of archived historical MVC-NDVI data with NDVI data from the current sensors