The Inverse Regional Ocean Modeling System (IROMS): development and application to data assimilation of coastal mesoscale eddies.

Emanuele Di Lorenzo
Georgia Institute of Technology

We describe the development and application of the Inverse Regional
Ocean Modeling System (IROMS), a 4D-variational data assimilation system for high-resolution basin-wide and coastal oceanic flows. IROMS makes use of the recently developed perturbation tangent linear (TLM), representer tangent linear (RPM) and adjoint (ADM) models of the Regional Ocean Modeling System (ROMS) to implement a representer-based generalized inverse modeling system. This modeling framework is modular. The TLM, RPM and ADM models are used as stand-alone sub-models within the Inverse Ocean Modeling (IOM) system described in Chua and Bennett (2001). The system allows the assimilation of a wide range of observation types and uses an iterative algorithm to solve nonlinear assimilation problems. The assimilation is performed either under the perfect model assumption (strong constraint) or by also allowing errors in the model dynamics (weak constraints). For the weak constraint case the TLM and RPM models are modified to include additional forcing terms on the right hand side of the model equations. These terms are needed to account for errors in the model dynamics. Posterior error statistics, term balances and array assessment are computed using separate diagnostic tools provided by the IOM.

After testing IROMS in an idealized 3D double gyre circulation we present a realistic application for the Southern California Bight (SCB), a region characterized by strong mesoscale eddy variability. The SCB model domain geometry is derived using real coastlines and a smooth version of satellite bottom topography. We assimilate synthetic data for sea surface height, upper ocean (0-500m) temperatures, salinities and currents over a period of 3 days. The spatial distribution of the synthetic observations follows the California Cooperative Oceanographic Fisheries Investigations (CalCOFI) sampling grid. The model first guess, prior to assimilation, is initialized using climatological conditions. The assimilation solution for the strong constraint experiment successfully reduces the initial model-observation misfit by 75% and improves the model fields also at locations where observations are not assimilated. In the weak constraint experiment the model-observation misfit is reduced by 89%. To verify the quality of the assimilation solution we integrate the model beyond the assimilation window for additional 3 days and measure the predictive skill against independent observations. Both the strong and weak constraint case show forecast skill greater than persistence and climatology.