High-Resolution Ocean Prediction with the NCOM 4DVAR

The NCOM-4DVAR is an ocean assimilation system that employs the weak-constraint representer method to assimilate a wide variety of temperature, salinity, velocity, and sea surface height observations with the Navy Coastal Ocean Model (NCOM) using 4-dimensional variational assimilation (4DVAR). This assimilation system has been tested and validated in a number of regional applications, and has been shown to outperform the currently operational 3dvar assimilation system, in terms of analysis and forecast skill. Up to now the NCOM-4DVAR has only been demonstrated for relatively large domains with horizontal resolutions of >= 3 km, therefore restricting the resolution to primarily mesoscale features. We know that submesoscale ocean processes with scales less than 10 km have a significant impact on the overall ocean environment, and it would be advantageous to have prediction skill of these types of processes. To do this, though, the NCOM-4DVAR has to operate at very-high resolution (~1km). The forward ocean model (NCOM) is capable of operating at these high resolutions, and it has shown to be able to resolve and propagate submesoscale features. Operating the NCOM-4DVAR at these high resolutions, however, creates some difficult challenges that need to be overcome; primarily the computational cost and the nonlinear interactions that develop in the model limiting the validity of the tangent linear approximation, especially in the presence of strong nonlinear flows that typically exist at high resolutions. This study aims to demonstrate that given enough computational resources, 4DVAR can successfully be applied to two high-resolution (1 km) domains in the North Arabian Sea and the Western Pacific Ocean.