Ensemble Kalman filter assimilation of Doppler radar data with a compressible nonhydrostatic model: OSS experiments

A Doppler radar data assimilation system is developed based on an ensemble Kalman filter (EnKF) method and tested with simulated radar data from a supercell storm. As a first implementation, it is assumed that the forward models are perfect and that the radar data are sampled at the analysis grid points. A general purpose nonhydrostatic compressible model is used with the inclusion of complex multiclass ice microphysics. New aspects of this study compared to previous work include the demonstration of the ability of the EnKF method to retrieve multiple microphysical species associated with a multiclass ice microphysics scheme, and to accurately retrieve the wind and thermodynamic variables. Also new are the inclusion of reflectivity observations and the determination of the relative role of the radial velocity and reflectivity data as well as their spatial coverage in recovering the full-flow and cloud fields. In general, the system is able to reestablish the model storm extremely well after a number of assimilation cycles, and best results are obtained when both radial velocity and reflectivity data, including reflectivity information outside of the precipitation regions, are used. Significant positive impact of the reflectivity assimilation is found even though the observation operator involved is nonlinear. The results also show that a compressible model that contains acoustic modes, hence the associated error growth, performs at least as well as an anelastic model used in previous EnKF studies at the cloud scale. Flow-dependent and dynamically consistent background error covariances estimated from the forecast ensemble play a critical role in successful assimilation and retrieval. When the assimilation cycles start from random initial perturbations, better results are obtained when the updating of the fields that are not directly related to radar reflectivity is withheld during the first few cycles. In fact, during the first few cycles, the updating of the variables indirectly related to reflectivity hurts the analysis. This is so because the estimated background covariances are unreliable at this stage of the data assimilation process, which is related to the way the forecast ensemble is initialized. Forecasts of supercell storms starting from the best-assimilated initial conditions are shown to remain very good for at least 2 It.