Surface Stokes drift in the Baltic Sea based on modelled wave spectra

The Stokes drift is an important component in the surface drift. We used the wave model WAM to evaluate the mean values and exceedance probabilities of the surface Stokes drift in the Baltic Sea. As there is no direct way to verify the accuracy of the modelled Stokes drift, we compared the bulk parameters calculated by the wave model against buoy measurements to ensure the quality of the wave hindcast. Furthermore, we evaluated the surface Stokes drift from measured wave spectra to assess the accuracy of the modelled surface Stokes drift. The importance of the Stokes drift as a component of the total surface drift was evaluated by calculating the hindcast mean values and percentiles of the surface Stokes drift. The mean values were between 0.08 and 0.10 ms^?1 in the open sea areas, thus being of the same order of magnitude as the mean wind shear currents. The highest values of the surface Stokes drift were slightly larger than 0.6 ms^?1. The comparison of modelled Stokes drift values to estimates obtained from measured spectra suggests that the mean values are well represented by the model. However, the higher modelled values are most likely slightly too large because the wave energy was overestimated during high wind situations in some of the sub-basins, such as the Gulf of Finland. A comparison to a drifter experiment showed that use of the Stokes drift improves the estimate of both the drift speed and the direction in the Gulf of Finland. Parameterised methods to evaluate the Stokes drift that are used, e.g. in currently available Baltic Sea drift models, overestimate the smaller values (under 0.3 ms^?1) and underestimate the larger values of the Stokes drift compared to the values calculated by the wave model. The modelled surface Stokes drift direction mostly followed the forcing wind direction. This was the case even in the Gulf of Finland, where the direction of the wind and the waves can differ considerably.