Data-Adaptive Prediction of Sea-Surface Temperature in the Arabian Sea

A nonlinear data-adaptive approach known by the name of genetic algorithm has been proposed for predicting satellite-observed sea-surface temperature (SST) in the Arabian Sea. A preliminary empirical orthogonal function (EOF) analysis has been carried out to separate the temporal variability from the spatial variability, and the algorithm has been applied to the time series of the principal components (PCs). The algorithm finds explicit analytical forecast equations that are later used to forecast the PCs. Afterward, predicted SSTs have been reconstructed using the predicted PCs and precomputed EOFs. Performance of the forecast has been evaluated by comparing it with persistence forecast, and it has been found that the algorithm is able to improve upon persistence forecast for the lead times of two to four weeks.