An Application of NEMOVAR for Regional Wave Model Data Assimilation

The benefits of applying data assimilation to a regional wave model for the northwest European continental shelf have been assessed. The assimilation method uses a 3D variational scheme commonly used in ocean model data assimilation, NEMOVAR, coupled to a regional configuration of WAVEWATCH III deployed on a Spherical Multiple-Cell grid with rotated pole. Significant wave height forecast skill was positively impacted over lead times of up to 12 hours, in both offshore and coastal areas of the model. Assimilation experiments tested the impacts of applying both satellite altimeter and in-situ platform observations of significant wave height as the state variable. Addition of in-situ data introduced an improvement in model skill versus assimilating altimeter measurements only, but these improvements were relatively short-lived in lead time (on the order of a few hours). Extending observation coverage into the coastal zone introduced a benefit for only the first 1-2 hours of forecast lead time at coastal sites. Although impacts on significant wave height forecasts were substantial at short lead times, the relatively short ‘system memory’ of the assimilation suggests that, in the case of the northwest European shelf region, the best application of wave data assimilation is in a rapidly cycling short range forecast system used to generate nowcasts, or for a reanalysis product.