Assimilation of Radar Kdp Observations Using an Ensemble Kalman Filter for the 31 May 2013 Oklahoma Storm Event: Investigation of DA Configuration Sensitivity and Simulated Microphysical States

Given the challenges of approximating observed microphysical states, dual-polarization radar variables can approximate hydrometeor properties with high spatiotemporal resolution. In this study, sector-scan specific differential phase (KDP), which is sensitive to liquid water, is directly assimilated alongside horizontal reflectivity ZH using the ensemble Kalman filter (EnKF) and a forward operator with T-matrix scattering amplitudes for the 31 May 2013 El Reno-Stillwater supercells event. Several DA parameter sensitivities are examined to optimize the assimilation of both ZH and KDP, including assimilation threshold, localization radius, and assimilation order. Reducing the KDP assimilation threshold to 0.75° km-1 results in the smallest KDP root mean square innovation (RMSI) among the single-sensitivity DA experiments at the final analysis time, as the analysis better reflects low KDP values downstream of the convective core. While ZH and KDP can be independently optimized, ZH is consistently overestimated near the El Reno supercell’s convective core when KDP is assimilated. Multi-sensitivity DA experiments are designed to mitigate this ZH overestimation, but do not sufficiently improve the accuracy of ZH in the analysis. Simulated rain and rimed ice microphysics variables are investigated to determine their contribution to ZH and KDP. It is found that the rain category alone is not responsible for the ZH overestimation when assimilating KDP observations, and that wet rimed ice is additionally enhancing ZH. These results emphasize the importance of simulated liquid water fraction on ice particles when optimizing the assimilation of multiple dual-polarization radar variables.