Sea surface temperature forecasts using on-line local learning algorithm in upwelling regions

Abstract When forecasting geophysical time series, long records are not always available. Therefore, the possible physical events may not be evenly represented in the available dataset and a unique neural network modelling the phenomenon of interest may not be accurate. As an alternative, a local learning algorithm is presented in this paper. It is applied to the prediction of sea surface temperature in an upwelling area. It provides encouraging results when applied to forecasts for eight days ahead.