In order to optimize the assimilation scheme, a new auto-correlation function is obtained by analyzing the forecasted errors on wave parameters from the WAve Model (WAM) driven by several forecasted wind fields (one and three days). Simulation tests with synthetic wave spectra are performed with the new correlation model. By comparing to the old function the statistical analysis showed that the assimilation index is improved in both analysis and forecast periods. However, the effect of this change in the forecast period is very small. In other respects, the assimilation scheme has been tested for real ASARENVISAT wave spectra provided by the European space agency. The scheme is adapted to the limitation in frequency related to the ASAR radar. The preliminary results show that the assimilation scheme works correctly and gives a good estimate of the mean wave parameters. Also the impact of the assimilat ion is more significant for the low frequency part of the wave spectrum. The comparison with existing altimeter data from ERS2 has permitted on the one hand to validate the real ASAR directional wave spectra, and on the other hand to estimate the efficiency of the assimilation scheme with spectral information.
Better understanding and prediction of sea state are an essential need for many application domains, such as navigation and security of ships, offshore industrial activities and the survey of the coastal zone…etc. In the framework of SWIMSAT (Surface Wave Investigations and Monitoring from SATellite, proposition to spatial mission to European Space Agency), we have shown the considerable contribution induced by the assimilation of directional wave spectra in wave prediction model WAM (Aouf et al, 2002). These results can be summarized on one hand by the improvement of the estimation of wave parameters (significant wave height, wave period and direction).