Satellite based forecasting of sea surface temperature in the Tuscan Archipelago

Abstract

The system described employs a nonlinear forecasting technique based on a combination of genetic algorithms and empirical orthogonal function (EOF) analysis. The genetic algorithm identifies the equations that best describe the behaviour of the different temporal orthogonal functions in the EOF decomposition and therefore, enables global forecasting of future time variability. The method is applied to obtain a one-month ahead forecast of the monthly mean space-time variability of the sea surface temperature (SST) of the Tuscan Archipelago, northwest coast of Italy. The system performance has been validated comparing forecast fields with real satellite observations. Results indicate that the system provides better predictions than those based on climatology. Future research is oriented to make the system applicable to military operations, environmental control and fisheries activities.