A performance evaluation of the MODIS vegetation index compositing algorithm

The Moderate Resolution Imaging Spectroradiometer (MODIS) vegetation index (VI) products are aimed at providing consistent, spatial and temporal comparisons of global vegetation conditions, which will be used for operational monitoring of the Earth's photosynthetic vegetation activity in support of land cover classification, change detection, and biophysical interpretations. Standard processing in generating the MODIS VI products include the temporal compositing process in order to minimize cloud and atmosphere contamination, and standardize sun/view angles. In this study, we evaluated the performance of the MODIS VI compositing algorithm as to whether the algorithm is selecting the cloud-free observations and whether it is producing and selecting observations with view zenith angles closest to nadir. The analyses were conducted over a set of globally-distributed sites covering a wide range of biome types over dry and wet seasons. We found the MODIS VI compositing algorithm to work fairly well over arid to semi-arid to temperate zones in both dry and wet seasons, in which nearly all of the composited pixels were cloud-free with view zenith angles closer to nadir than the conventional MVC results. On the other hand, the MODIS algorithm did not improve upon the MVC results well over tropical rain forest areas as cloud covers persisted throughout the period analyzed in these regions.