On the convergence of the ensemble Kalman filter. (Q664408): Difference between revisions

One of the most successful recent data assimilation methods for high-dimensional problems is the ensemble Kalman filter. The present analysis does not assume that the ensemble members are independent or normally distributed. A discrete-time filtering problem is introduced. It is based on the Bayes theorem. The main theorem establishes the convergence of the ensemble Kalman filter in the Lebesgue space \(L^p\) for all \(p\in [1,\infty)\). As a consequence the authors derive that the ensemble mean and covariance converge to the filtering mean and covariance.