Simultaneous estimation of inflation and observational errors in Ensemble Kalman Filter

Based on statistics of “observations minus forecasts”, “observations minus analysis” and “analysis minus forecasts” (Desroziers et al., 2006), we develop a method to estimate online simultaneously observation error covariance and the optimal inflation for the Local Ensemble Transform Kalman Filter (LETKF). Experiments with perfect model simulations with the Lorenz (1996) and the global SPEEDY model show very promising results.