Application of the Back and Forth Nudging Data Assimilation to the Navy Coastal Ocean Model

The Back and Forth Nudging (BFN) is an easy to implement iterative data assimilation method that has performance similar to that of 4DVAR. The BFN is less sensitive to the initial guess, requires less computational resources and does not require a background covariance matrix like most data assimilation methods. The BFN has been succesfully applied to various models including but not limited to shallow water, multi-layer quasi-geostrophic and the primitive equation ocean models. However, these applications have mostly been limited to twin experiments in rectangular bassins with flat bathymetry.

In this presentation, we investigate the application of the BFN data assimilation to an operational model, namely the Navy Coastal Ocean Model (NCOM), in a realistic setting of the gulf of Mexico using actual observations. Results will be compared to those obtained from a 4DVAR assimilation experiment.