Forecasting non-linear systems using the ensemble Kalman filter and related methods

The project will develop new assimilation techniques related to the Ensemble Kalman Filter and apply them in various research fields in view of operational forecasting of non-linear dynamical systems. The disciplines targetted are the following: - Physic al oceanography (MSC: TOPAZ sea-ice): An parameter estimation extension of the EnKF will be developed in the context of assimilation of sea-ice drift in the Arctic Ocean. - Biological oceanography (MSC: TOPAZ-ECO system): An extension of the EnKF for ass imilation of non-Gaussian variables will be developed for a 3D coupled physcial-ecosystem model of the Atlantic and Arctic Oceans. - Weather forecasting (Met.no: HIRLAM model, Storm: MM5/WRF model). Met.no will conduct EnKF-related modifications of an ex isting variational data assimilation system, Storm will directly use the EnKF for forecasting targetted local weather phenomena in Northern Norway. - Oil reservoir forecasting (MSC/Hydro: Eclipse). Develop new EnKF-based methods general problems of oil r eservoir models. An internal project at Hydro will complement with practical applications and demonstrate the use of probabilisitc production forecasts with oil fields in operations. - Bioeconomic management (NHH) sustainable management of marine resourc es with a stochastic market and stochastic biology. - Health (Ecole des Mines de Paris): Analysis or early warning of epidemic outbreaks in France based on the Sentinelles network. These disciplines share some challenging aspects related to the non-linea rity of the system (biases, non-Gaussian distributions). Some of them are high-dimensional and some are low-dimensional, the latter will therefore be used as a laboratory for assessing new developments against advanced Monte Carlo methods.