Estimation of transport at open boundaries with an ensemble Kalman filter in a coastal ocean model

The performance of the ensemble adjustment Kalman filter (EAKF) to estimating volume transports at open boundaries was investigated for a coastal ocean model of the Palma Bay, Mallorca Island, Mediterranean Sea. The circulation in the Palma Bay is mainly driven by the remotely wind-generated island trapped waves (ITWs) propagating around Mallorca and by a local wind response. Thus, for a model of the bay proper only, the large-scale ITWs must be incorporated into the open ocean boundary conditions. To take into account the effect of ITWs, an EAKF was used to assimilate the moored Acoustic Doppler Current Profiler (ADCP) velocity time series collected in the inner part of the Bay through a simultaneous state and parameter estimation. In this approach, flows at open boundaries were included in the model state as time-dependent parameters and were updated in each assimilation step. The simulation was validated with moored ADCP data as well as with independent towed ADCP spatial velocity survey data. The new results are markedly improved over the model experiments without data assimilation or with state estimation only. In particular, simulation using the estimated values at open boundaries significantly outperforms simulation with ‘a priori’ prescribed values at boundaries. The error reduction is about 45% when ITWs dominate the circulation. Our results present a promising approach in dealing with the open boundary conditions in a coastal ocean model.

► We estimate open boundary conditions in a coastal model of the Palma Bay. ► We use the ensemble Kalman filter through simultaneous state and parameter estimation. ► The estimated boundary conditions account for the large-scale flow. ► The large-scale flow at the open boundaries improves model simulation. ► The error reduction is about 45% with respect to independent data.