Spatially and temporally varying adaptive covariance inflation for ensemble filters

摘要: Ensemble filters are used in many data assimilation applications geophysics. Basic implementations of ensemble trivial but susceptible to errors from sources. Model error, sampling error and fundamental inconsistencies between the filter assumptions reality combine produce assimilations that suboptimal or suffer divergence. Several auxiliary algorithms have been developed help tolerate these errors. For instance, covariance inflation combats tendency ensembles insufficient variance by increasing during assimilation. The amount is usually determined trial error. It possible, however, design Bayesian determine adaptively. A spatially temporally varying adaptive algorithm described. normally distributed random variable associated with each element model state vector. Adaptive demonstrated two low-order experiments. In first, dominant source small second, dominant. better mean estimates than other methods.