The Development of a Hybrid 3DVAR-EnKF Algorithm for Storm-scale Data Assimilation

A dual-resolution hybrid 3DVAR-EnKF data assimilation algorithm is developed based on existing 3DVAR and ensemble Kalman filter (EnKF) programs within the ARPS system. The algorithm uses the extended control variable approach to combine the static and ensemble-derived flow-dependent forecast error covariances. With the dual-resolution implementation, hybrid variational analysis is first performed on the high-resolution grid using a mix of static error covariance and flow-dependent covariance derived from lower-resolution ensemble forecasts. This hybrid analysis serves as the initial condition for the single high-resolution forecast, which can in turn be used to adjust the lower-resolution ensemble forecasts. The lower-resolution forecast perturbations are updated using the standard EnKF method at the lower resolution. In the hybrid method, the relative weights assigned to the static and flow-dependent error covariances can be tuned, and the tuning can be application and ensemble size dependent. The method is first applied to the assimilation of simulated radar data for a supercell storm. Sensitivity experiments are performed with different combinations of the covariance weights. Results obtained using pure-3DVAR (with static covariance entirely), mixed covariance (within the variable framework), and the standard EnKF are compared. The use of dual-resolution has the advantage of reducing computational cost of a standard single-resolution hybrid 3DVAR-EnKF scheme without significantly sacrificing the quality of analysis. Initial tests indicate that the hybrid scheme can reduce the storm spin-up time. Details on the results will be presented at the conference.