Abstract : The coastal ocean is characterised by a rich spectrum of interacting scales and processes, and is subject to heavy societal pressure and demand. Our understanding of this complex area needs to be improved, through modelling and data gathering. Realistic numerical models are potentially useful to design observational arrays. Ensemble methods provide a proxy for model errors, which may be detected or constrained by an efficient array.
In a first study, an ensemble Kalman filter is used to assimilate simulated data of a wide swath altimeter in a barotropic model of the North Sea. Measurements of the cross-track slope of the sea surface lead to a better representation of the currents, and to a wider domain of influence of observations. The roll of the satellite induces along-track correlation of the observation errors; when it is taken into account in the data assimilation process, it does not jeopardize the improvement potentially brought by the instrument.
In a second study, a new analysis technique for assessing the performance of observational networks is proposed. It implements a quantitative criterion to discriminate between various networks, and a qualitative one which allows understanding the impact of the considered network on all the model dimensions. Using a 3D model of the Bay if Biscay, this technique illustrates (1) the capacity of a wide swath altimeter to retrieve useful coastal ocean mesoscale dynamical signals, (2) the positive contributions of current measurements of currents in the design of a coastal cruise, and (3) the contribution of tide gauge data at detecting errors on the shelf.