Abstract
Ocean data assimilation (DA) has widely been used for various engineering purposes including monitoring, forecast, and port operations. However, in semi-enclosed estuaries such as the Seto Inland Sea (SIS), Japan, DA has rarely been adapted because satellite data and Argo floats are desperatively unavailable, albeit being essential to successful oceanic predictions. In the present study, we develop a DA system toward rigorous predictions optimal for relatively small coastal and estuarine settings. The system consists of a three-dimensional variational (3DVAR) assimilation scheme being made compatible with ROMS using an in-situ dataset available for the SIS. The DA system significantly improves the reproducibility of water temperature and salinity fields from those without DA. We show an evidence that detailed terrestrial freshwater discharge data is inevitable for more accurate salinity predictions in the estuary.
