Soft project: a new ocean forecasting system based on satellite data

The aim of the SOFT project is to develop a new ocean forecasting system by using a combination of satellite data, evolutionary programming and numerical ocean models. To achieve this objective two steps are proposed: (1) to obtain an accurate ocean forecasting system using genetic algorithms based on satellite data; and (2) to integrate the above new system into existing deterministic numerical models. Evolutionary programming will be employed to build "intelligent" systems that, learning from the past ocean variability (provided by satellite data) and considering the present ocean state, will be able to infer near future ocean conditions. Validation of the forecast skill will be carried out by comparing the forecasts fields with satellite and in situ observations. Validation with satellite observations will provide the expected errors in the forecasting system. Validation with in situ data will indicate the capabilities of the satellite based forecast information to improve the performance of the numerical ocean models. This later validation will be accomplished considering in situ measurements in a specific oceanographic area at two different periods of time. The first set of observations will be employed to feed the hybrid systems while the second set will be used to validate the hybrid and traditional numerical model results. Keywords: forecasting, satellite data, empirical orthogonal functions, numerical models, genetic algorithms, neural networks, Mediterranean Sea.