Changes in the Global Vegetal Cover Through a Phenological Analysis of GIMMS Data

This study uses the 22 years of GIMMS (Global Inventory Modelling and Mapping Studies) NDVI (Normalized Difference Vegetation Index) data for the determination of land surface phenology in order to characterize changes in the global vegetal cover. This dataset has a global bi-monthly coverage since mid-1981 to 2003 with a spatial resolution of 8 km at the equator. The applied methodology involves, for each pixel, a fitting procedure of yearly NDVI data to a double logistic function, describing most of the NDVI behaviours of the biomes of our planet. This fitting procedure summarizes in seven parameters the NDVI yearly cycle, two of them dealing with land surface phenology, namely spring and autumn dates from which growing season length is retrieved. The accuracy of this method for land surface phenology estimation is of 5.5 days. To validate the results, trends are retrieved pixel by pixel, then averaged on geographical ensembles and confronted with their confidence intervals. The resulting trends compare well with previous studies, and give a supplementary insight of the changes that the global vegetation has suffered since the early eighties.