Propagation of Ensemble Kalman Filter corrections into nested ocean model domains: A Norwegian coast case study

As ocean observational systems expand their spatiotemporal coverage and models in ocean forecasting systems grow more complex and simulate at higher resolutions, the computational cost of assimilating observations using tools such as the Ensemble Kalman Filter increases, and at the finest model grid scales suitable observations are often too sparse or too coarse for direct assimilation. Consequently, investigating optimal strategies for data assimilation becomes imperative, particularly in nested ocean model setups. These setups present an opportunity to assimilate new observations primarily in the coarser ’mother’ domains provided that one can still capitalize on the benefits of observational data in nested domains through data propagation across boundary conditions. In this study, we assimilate sea surface temperature and salinity observations into a regional domain and examine their propagation to a nested local coastal domain, analyzing 12 distinct subdomains that capture various inflow and outflow regimes. Our findings reveal considerable spatiotemporal variability across these subdomains, with areas under the influence of strong inflows, notably along the Norwegian Coastal Current, benefiting the most from observational data assimilated in the mother domain. This study contributes to efforts to balance computational cost and forecast accuracy in ocean forecasting systems, setting the stage for further advancements in efficient and accurate ocean predictions. • Strongest observational data propagation in coastal and inflow subdomains. • More temporal lag in propagation of temperature corrections vs salinity corrections. • Potential computational cost reduction for small orgs using high-accuracy models.