The mathematical models are made for physical variables (water height, currents, waves, temperature, salinity), chemical variables (mainly hydrocarbons), biological variables (limited to the first links in the food chain), sediment transport and transport of other (potentially polluting) substances such as oil slicks, and for the calculation of changes in the form and composition of the sea bed. These variables are inter-related, the level of complexity and uncertainty increasing as you move from the physical to the biological and to the sediment.

The results of the models are then carefully compared with all the available observations made on trips to sea, by measuring instruments on the sea surface or the sea bed and satellite observations. This is the “validation phase”. If the results of the validation are insufficient, this means that the mathematical formulas selected to describe the processes are inadequate. If the validation results are sufficient, this means that the models can be used for sea management and in predicting parameters. This enables the reduction of uncertainties in the results calculated by the model by, for example, better adjusting the parameters. A more advanced technique is data assimilation whereby forecasts made by the model are adapted to the observations during the actual simulation. The method also enables the calculation of a margin of error.

The models are, above all else, research instruments with which to test scientific hypotheses in a virtual manner. In that case we speak of models in development. As soon as these models show an acceptable reliability we speak of operational models. These models can provide a variety of information on the sea (forecasts of water height, currents, waves, etc.); just as mathematical models can be used to forecast weather.