Assimilation of wide swath satellite altimetry to map geostrophic and internal tide signals of the ocean dynamics

Mapping Sea Surface Height (SSH) from satellite altimetry is key to many scientific and operational applications. At the fine scales observed by SWOT, SSH variations are primarily driven by two types of dynamics - nearly geostrophic, Balanced Motion (BM) and wavy motion due to the Internal Tide (IT). These two processes differently affect ocean dynamics. The IT does not affect the surface horizontal transport of passive tracers (oil spills, plastics, algae…) unlike BM, but has a significant contribution to the vertical transport of heat, salt and nutrients. Separating BM and IT contributions to SSH variations will be essential in the mapping process. To a different extent, this separation is now common practice with high-frequency outputs of numerical simulations. Yet it is still an unresolved challenge for SSH maps computed with satellite observations like SWOT, which are sparse in space and time. 

This study introduces an innovative method to separate BM and IT components from SSH altimetric observations including SWOT. The method is based upon a data assimilation system combining two models - quasi-geostrophic for BM and linear shallow-water for IT. The inversion is performed with a weak-constraint four-dimensional variational (4DVar) approach, with two different sets of control parameters adapted to each component. A major, expected benefit of this approach lies in the potential to capture the non-stationary part of the internal tide component. The method outputs hourly SSH and surface velocity fields over a domain for both components. 

The work focuses on the North Pacific Ocean, because this zone has a strong mesoscale and submesoscale activity, including the two dynamics of interest. Observing System Simulation Experiments (OSSEs) are carried out over 20°x20° domains surrounding the SWOT crossovers. The experiments include both conventional Nadir and wide-swath SSH measurements, that are interpolated from the LLC4320 MITgcm simulation. The mapping algorithm performances are evaluated by comparing the outputs with the MITgcm referenced fields. The first results indicate that the assimilation system is able to separate the two targeted components, including the non stationary part of the internal tide. As this study is part of a PhD, the latest results available will be presented at the conference.