NDVI and GEMI comparison using AVHRR-NOAA data to remote sensing the vegetation in Brazil

onditions. Satellite derived vegetation indices have proved to be an outstanding tool for characterizing the surface conditions throughout the last two decades. Since its very first application for crop monitoring during late 1978 with the first TIROS series satellite vegetation indices were found to be related to an increasing number of surface parameters such as leaf area index (LAI), biomass, percentage ground cover among others. In more recent years vegetation indices have also shown to be usefull as an input to surface related schemes in numerical prediction models (such as GCMs) with an improvement of the surface fluxes simulations. Over a number of different vegetation indices approaches NDVI (Normalized Difference Vegetation Index) is probably the most widely used. Exploring the different response of vegetation and bare soil over the visible and near-infrared bands this index is quite sensitive to the green vegetation foliage but is also affected by underlying soil colour changes due to precipitation and by the atmospheric condition that can alter the index value specially due to aerossol scattering on the visible region and to water vapor absorption on the near-infrared region. A more recently introduced index, the GEMI (Global Environment Monitoring Index) was developed in order to use the same two spectral bands as does NDVI and to attend the need to be sensitive to vegetation and insensitive to soil colour changes and to the atmospheric conditions. A comparison between these two indices is carried using AVHRR/NOAA-14 data for the brazilian area. The preliminary results showed that both indices have a quite satisfatory performance on the representation of the continental surface over a wide range of green vegetation presence. GEMI has demonstrated some advantages over NDVI on the separation between less vegetated areas and clouds as well as being less sensitive to atmospheric c