Developing a compositing algorithm for retrieval of green vegetation fraction from the Suomi NPP satellite

Real-time weekly global Green Vegetation fraction (GVF) is needed in the numeric weather, climate and hydrological models. The current NOAA operational GVF product is derived from weekly AVHRR NDVI data, which are composited using the maximum-value compositing (MVC) method. MVC is a widely used technique to remove cloud and atmospheric contamination over land surface by selecting the observation of the maximum NDVI in a compositing period. However, it is well documented that the maximum NDVI is often selected from high sensor zenith angles, which may introduce error in GVF retrieval. To reduce the composite sensor zenith angles, a view angle adjusted soil-adjusted vegetation index, instead of NDVI, is proposed as the criterion of compositing in this study. Compared with MVC, the new compositing method reduced sensor zenith angles significantly but high (not necessary maximum) NDVI and EVI values were retained.