Mapping savanna ecosystems of the Llanos del Orinoco using multitemporal NOAA satellite imagery

Abstract

Monthly composites NOAA-advanced very high resolution radiometer (AVHRR) normalized difference vegetation index (NDVI) images (June 1992–October 1993) were used to analyze the phenology of savanna ecosystems in the Llanos del Orinoco region. The objective was to elaborate an ecosystems map of the area based on the temporal pattern analysis. Expert knowledge of savannas allowed the selection of representative ecosystems to monitor changes in NDVI. From 54 satellite images, model curves were created for each selected ecosystem. They were analyzed with the objective to characterize and identify each ecosystem. From these temporal patterns and statistical analyses, six satellite images were chosen to carry out a supervised classification using Mahalanobis distance methodological approach. Validation of the map was implemented using ground control points. Differences in phenology between ecosystems are strongly related to the environmental climatic conditions, especially to seasonal rainfall. Therefore, the phenology for each ecosystem can be explained based on water availability. A methodological approach, which reveals the phenology of the most important savanna ecosystems in Venezuela, was applied in this work. This study represents an improvement in mapping of ecological processes.