Multiplicative Spatio-Temporal Models for Remotely Sensed Normalized Difference Vegetation Index Data

Abstract

Vegetation forecasting is closely tied to many important international concerns, including: monitoring the impacts of global climate change and energy usage, managing the consumption of natural resources, predicting the spread of invasive species, and protecting endangered species. In light of these issues, this article develops vegetation forecasting models for normalized difference vegetation index (NDVI) data recorded by remote sensing via satellites in East Africa. Spatio-temporal auto-regressive moving average (STARMA) is a class of models that can be used in monitoring and forecasting, but it must be modified for highly seasonal processes with temporal trends. We propose to use multiplicative STARMA models to estimate and forecast NDVI values for sub-regions that have previously been detected to have statistically significant temporal trends. For illustration, we select a few East African sub-region’s NDVI series to apply the proposed models and demonstrate the advantages over traditional modeling techniques.