The impact of ocean data assimilation on the simulation of mesoscale eddies at São Paulo plateau (Brazil) using the regional ocean modeling system

Abstract

This work presents simulations of ocean circulation in the southwestern South Atlantic Ocean with emphasis on the São Paulo Plateau region, a unique physiographic feature at the foot of the continental slope offshore southeast Brazil, where the flow is dominated by ocean eddies and dipoles. Surface dynamics south of 23°S presents a heterogeneous circulation pattern intimately related to the presence of mesoscale features, such as meanders of the Brazil Current along the continental slope, and eddies that interacts with deep ocean features. The reproduction of these eddies at the correct time and spatial structure are quite challenging in the context of a non-assimilative model, considering the limitations imposed by the model’s numerical schemes, physical parameterizations, boundary conditions and surface forcing. In this sense, the Regional Ocean Modeling System (ROMS) 4-dimensional Variational Data Assimilation (4D-Var) module was used to increase model skill and minimize deviations from oceanic observations. Weekly observations of Sea Surface Height (SSH), daily Sea Surface Temperature (SST) fields and in situ vertical profiles of temperature and salinity from multiple sources were used in consecutive assimilation cycles to produce an accurate representation of ocean features and their variability. Both the assimilative and free run were evaluated against observations, including in situ current measurements. Results showed significant improvements of the assimilation procedure on both assimilated and non-assimilated variables, with RMSE reductions of 27% and 47% for SST and SSH, respectively, and 51% for current speed measurements during a dipole event in the São Paulo Plateau. The assimilative run was able to reproduce a remarkable event of mesoscale eddies interaction (dipole) and consequent current intensification, leading to important improvements on model skill. The correct use of observational information of open ocean surface features through data assimilation is a key factor for a proper simulation of such eddy dipole events.