Data Assimilation into a Coupled Physical-Biochemical Ocean Model: Towards the Application of Optimal Nonlinear Filtering

The talk will present our efforts to apply the optimal nonlinear filtering theory to estimate the state of a coupled physical-biochemical ocean model. The talk is divided into two parts.
In the first part, I will present a discrete Gaussian-mixture-based solution of the optimal nonlinear filter that generalizes the optimality of the analysis step of the ensemble Kalman filters to nonlinear systems. We refer to this approach as "particle Kalman filter (PKF)" because it applies two types of corrections: a Kalman-type correction to the particles and a particle-type correction to the particles weights. I will also discuss the implementation of the PKF with state-of-the-art atmospheric and oceanic data assimilation problems. In the second part, I will describe a marine ecosystem data assimilation system that was designed to simultaneously assimilate physical and biological observations. The system is composed of a coupled Princeton Ocean Model (POM) - Biochemical Flux Model (BFM) and a square-root ensemble Kalman filter.
Assimilation results from regional and coastal configurations of the Mediterranean Sea will be presented.