One of the most valuable applications of dual-Doppler radar data is the retrieval of the three-dimensional wind field within convective storms. Since radars generally provide the only dense observations of storms above the ground, dual-Doppler retrievals and the various analyses derived from them (e.g., parcel trajectory and vorticity budget calculations) are critical to illuminating storm kinematics and dynamics. Thorough understanding of the errors associated with different wind retrieval techniques is therefore critical to maximizing their use in advancing understanding of convective storms.

This study uses an observing system simulation experiment (OSSE) framework to assess the accuracy of supercell kinematical and dynamical retrievals obtained from ensemble Kalman filter radar data assimilation (EnKF-RDA) versus dual-Doppler analysis (DDA). The truth simulation and data assimilation experiments are performed using the National Severe Storms Laboratory Collaborative Model for Multiscale Atmospheric Simulation (NCOMMAS) and its EnKF-RDA system. The dual-Doppler analyses are obtained using a variational approach that weakly satisfies observational, mass conservation and smoothness constraints. Of particular interest is the accuracy of single-radar EnKF-RDA retrievals, and whether EnKF-RDA substantially improves upon DDA in the two-radar case. In initial experiments, a “perfect” forecast model (except that it uses a coarser grid than the truth simulation) is used to place an upper limit on the expected improvement in retrievals from EnKF-RDA versus DDA. In subsequent experiments, one or more parameterization schemes in the EnKF forecast model are varied from those used in the truth simulation to obtain a more realistic assessment of the errors that can be expected in practice. The impacts of decreasing cross-beam angle and decreasing observational resolution on the relative performance of the two methods are also examined. Implications for mobile radar deployment strategies are discussed.