Contribution de Joaquim BALLABRERA, Jacques VERRON, and Giovanni RUGGIERO:

Assimilation of SWOT-type altimetry in a high-resolution ocean model

One of the most successful components of the ocean observing system is satellite altimetry. Since its beginning, altimetry has been a rich source of information about a wide set of temporal and spatial scales of climate interest. Altimeter data is currently being assimilated in operational oceanography. A foreseeable new technology to further improve our knowledge about the sea surface topography is the Surface Water Ocean Topography (SWOT) mission, a cooperative effort between NASA and CNES. By its ability to provide maps of the sea topography at scales below 10 km, SWOT will become the first altimetry mission able to capture sub-mesoscale features. The capacity of continuously monitoring this new range of scales will allow a better comprehension about the kinetic energy of the ocean circulation, the role of small-scale processes in the transport of tracers (as temperature, salinity or carbon), coastal tides, and even internal tides, which have not been well sampled by traditional altimetry. It can be expected that the amount of information that data assimilation will be able to infer from this new stream of data will be directly linked to the ability of the numerical models to represent sub-mesoscale features. However, the computational cost of models with a horizontal resolution fine enough (1 to 10 km) to permit sub-mesoscale variability, the use of the most advanced data assimilation methods (variational or ensemble-based methods) could reach prohibitive computational costs, except when applied to small domains. This work aims to investigate the use of new data assimilation approaches in which the model is used as a dynamical interpolator, both forward in time and backwards (as the Backward and Forward Nudging, BFN), in non-linear simulations. Although the backward integration of a numerical model has been shown to be useful in both variational (replacing the adjoint model) and nudging approaches, the actual ability of the backward simulation to constrain the initial and final states of the model is investigated as a function of the dynamical regime of the flow, the length of the assimilation period, the parameters being observed, and the nature of the observational noise (amplitude and presence of unresolved processes). The data assimilation experiments are going to be carried out in two simulations using the NEMO-OPA ocean model. In a first instance, an idealized subtropical gyre (Square Box SQB model) is being used to investigate the ability of the BF to constrain surface variability from a sampling simulating the swath characteristics of SWOT (120 km wide, and 22-day repeat orbit). The second set of experiments will use a regional simulation of the Solomon Sea.

up arrow