Combining a Model with Observations: Data Assimilation in ROMS

Brian S. Powell
University of Hawaii at Manoa, USA

Ocean models provide an estimate of the ocean state filtered by discrete dynamical equations. Observations provide sparse information about the ocean at a variety of temporal and spatial length scales. Using one to inform the other allows us to understand and estimate the ocean more fully, but accomplishing this requires proper formulation of the problem. Data assimilation is the procedure of using the observed data to improve the model's estimate of the ocean, and there are a variety of methods available. In this talk, we will discuss the philosophy of combining the two estimates and how the advanced, state-of-the-art tools that ROMS provides to solve the problem.