Assimilation of Directional Wave Spectra in the Wave Model WAM: An Impact Study from Synthetic Observations in Preparation for the SWIMSAT Satellite Mission

Abstract

Abstract Within the framework of the Surface Waves Investigation and Monitoring from Satellite mission (SWIMSAT) proposed to the European Space Agency, an assimilation scheme has been implemented in the Wave Model (WAM) in order to estimate the impact of spectral information on wave prediction. The scheme uses an optimal interpolation and the “spectral partitioning” principle. The synthetic wave spectra are located along a SWIMSAT orbit track and are assimilated in a 4-day-period simulation. Random errors are included to simulate the uncertainties of SWIMSAT instrumentation. The sensitivity of the scheme to background and observational errors and the correlation length is examined. The assimilation impact is investigated for two cases of moderate and large errors of the first guess. The results show that the assimilation scheme works correctly and the rms errors of significant wave height, mean period, and direction are significantly reduced for both periods of analysis and forecast. The impact on significant wave height is noticeable during the period of analysis and stays efficient for 2-day forecasts. For a large error in the first guess, the impact increases and remains significant for 3-day forecasts. Statistical analysis of mean wave parameters clearly shows that the use of spectral information yields a better estimate of wave frequency, direction, and low-frequency wave height in comparison with the results based upon assimilation of wave heights only. However, total significant wave height is less sensitive to the addition of spectral information in the assimilation scheme. The use of correlation length depending on the latitude of grid points leads to a better spread of incremental observations and, hence, to better skills in terms of the rms errors of mean wave parameters. The use of several wavelength cutoffs concerning the SWIMSAT synthetic wave spectra suggests that the “assimilation index” of mean wave parameters decreases with the increasing wavelength cutoff.