Impact of sea level data assimilation on the storm surge and seiche forecasts

This work assesses the impact of assimilating the sea level data, with an Ensemble Kalman Filter, on storm surge and seiche modelling. The study area is the Adriatic Sea, where seiches are always present after a storm surge, and often overlap on a new storm surge with a possible amplification of the total sea level. Due to errors in the wind and pressure forcing, the forecast of such extreme events is rather challenging in the Adriatic Sea, and a wrong reproduction of such pre-existing seiches reflects on a bad sea-level forecast. Here we show, by two case studies, that the assimilation of sea-level data along the coasts of the Adriatic basin is able to correct the initial state of the hydrodynamic model. Since the initial state is particularly important in the case of pre-existing seiches, the reduction of the initial error propagates several days into the forecast. The two cases here presented were between the most extreme storm surge events in the last years and they both included the pre-existing seiches. The initial forecast was very poor, due to the fact that the wind was underestimated. The assimilation of 3-day long hourly sea level data at eleven stations distributed along the Adriatic coasts produces a better forecast in both cases. Moreover, the ensemble spread allows the uncertainty of the forecast to be estimated, even if the estimate should be calibrated over time in order to be more reliable.