The impact of 4D-Var data assimilation of HF-Radar and SST observations on the surface circulation of the northwestern Mediterranean Sea 

<p>We analyze the effect of the assimilation of surface velocities from two HF-Radars (HFR) and satellite SST in an ocean circulation model covering the northwestern Mediterranean Sea.<br />One HFR observes the area in front of Toulon (France), the other covers the area adjacent to the coast near La Spezia (Italy).<br />The ROMS 4D-Var DA package is used to analyze a three-month period, from September to November 2020.<br />The DA is implemented to have a sequence of 3-day assimilation windows, each followed by a 3-day forecast cycle in which the model evolves without adjustments.<br />Boundary conditions are from the CMEMS Analysis-Forecast product and atmospheric forcing from ECMWF high resolution model (1/10 degree).<br />The impact of the DA is quantified by comparing the analysis and forecast runs with a freerun that is unaffected by any assimilation procedure.<br />The results show that both the analysis and the forecast perform better than the freerun considering the HFR and SST observations used in the assimilation procedure.<br />Moreover, eulerian velocities derived from surface drifters are used as an independent dataset to further validate the quality of the DA output.<br />Within the area covered by the HFRs, both the analysis and forecast surface velocities show greater agreement with the drifter data than the surface velocities from the freerun.<br />However, in the rest of the domain, surface circulation improvements and deteriorations tend to balance out, and mean squared error and correlation values are comparable for analysis, forecast, and freerun.<br />We also analyzed the impact of the observation typology and of the control vector on coastal transport for the upper 50 m at three transects.<br />Two of them are located at the HFR areas and the other in between.<br />Both HFR and SST data have a significant effect on surface transport increment.<br />The HFR observations contribute to the change in the transport magnitude while the SST tends to change the velocity distribution along the transect, leaving the transport magnitude unchanged.<br />Coastal transport is, in general, mostly affected by corrections to initial conditions in all three transects. The changes to the boundary conditions and atmospheric forcings have different importance depending on the transects analysed.</p>