A Multi‐Model Ensemble Kalman Filter for Data Assimilation and Forecasting

Abstract Data assimilation (DA) aims to optimally combine model forecasts and observations that are both partial noisy. Multi‐model DA generalizes the variational or Bayesian formulation of Kalman filter, we prove it is also minimum variance linear unbiased estimator. Here, formulate implement a multi‐model ensemble filter (MM‐EnKF) based on this framework. The MM‐EnKF can multiple ensembles for forecasting in flow‐dependent manner; uses adaptive error estimation provide matrix‐valued weights separate models observations. We apply methodology various situations using Lorenz96 illustration purposes. Our numerical experiments include with parametric error, different resolved scales, fidelities. results significant reductions compared best model, as well an unweighted ensemble, respect probabilistic deterministic metrics.