Forecasting the interannual trends in terrestrial vegetation dynamics using time series modelling techniques

The sensitivity and response of vegetation to climatic perturbations operates over a range of temporal scales. Provision of information on the state of vegetation prior and post climatic perturbations is a critical starting point for forecasting the dynamics of terrestrial systems. A simplistic methodology for generating forecasts that integrate remotely sensed data using time series modelling techniques is discussed in this paper. We focus on forecasting the behaviour of terrestrial South American vegetation in the presence of interannual perturbations by the ENSO using simplistic time series modelling techniques. Coarse resolution (1 degree x 1degree) NDVI data products, derived from the NOAA AVHRR which are available for the period of 1981 to 1992 were used as the primary input for the simulations. Forecasts were generated using ARIMA modelling methodology. The model outputs were validated against the finer resolution 8 km x 8 km AVHRR NDVI data set, which extends from 1981 to 1999. Considering the spatial scales involved and the relative simplicity of the techniques, this methodology provides a rapid and effective means of analysing vegetation dynamics over the spatio-temporal domain.