Quality assurance of global vegetation index compositing algorithm using AVHRR data

In this study the quality control aspects of the vegetation index produced by the MODIS (MODerate resolution Imaging Spectroradiometer) algorithm were evaluated and compared with the results of the currently used maximum value composite (MVC), which chooses the highest NDVI (normalized difference vegetation index) over a certain time interval. The composite scenarios were evaluated with respect to: 1) quality flags related to data integrity, cloud cover and composite method, 2) temporal evolution of the VI for different continents and vegetation cover types, and 3) accuracy of the standardization of reflectance values to standard view angles (nadir). On a continental scale the composited NDVI results from the MODIS algorithm were 1 to 24% lower than the off-nadir NDVI results based on the MVC criteria. A simple BRDF (Bidirectional Reflectance Distribution Function) model was adequate to produce nadir equivalent reflectance values from which the NDVI could be computed. The temporal evolution of the NDVI derived with the MODIS algorithm was similar to or smoother than the NDVI derived from the MVC algorithm. The composite method applied in the MODIS algorithm, was dependent on the global position and season, e.g. the BRDF interpolation was most frequently applied in arid and semi-arid regions and during the dry season over tropical rain forests. Examples of a global NDVI and quality flags were displayed using a pseudo color bit mapping scheme.