Assimilating polarimetric radar data with an ensemble Kalman filter: OSSEs with a tornadic supercell storm simulated with a two‐moment microphysics scheme

Abstract

AbstractThe impact of assimilating differential reflectivity ZDR in addition to reflectivity (ZH) and radial velocity (Vr) from a polarimetric radar on the analysis of a tornadic supercell storm using an ensemble Kalman filter (EnKF) is studied in an observing system simulation experiment (OSSE) framework assuming a perfect forecast model. A double‐moment microphysics scheme is used to allow for proper simulation of polarimetric signatures. Root‐mean‐square errors of analysed state variables are calculated and the structure and intensity of analysed fields and derived quantities are examined. Compared to the baseline experiment assimilating radial velocity and reflectivity only, the assimilation of additional ZDR further reduces the errors of all state variables. The analysed hydrometeor fields are improved in both pattern and intensity distributions. Polarimetric signatures including ZDR and KDP columns, and ZDR arc in the supercell, are much better reproduced. Sensitivity experiments are performed that exclude the updating of hydrometeor number concentrations by ZDR or of state variables not directly linked to ZDR via observation operators. The results show that if number concentrations are not updated together with the mixing ratios, most of the benefit of assimilating ZDR is lost. Among other state variables, the updating of water vapour mixing ratio qv has the largest positive impact while the impact of updating vertical wind w comes in second. The updating of horizontal wind components or temperature has a much smaller but still noticeable impact. Reliable flow‐dependent cross‐covariances among the state variables and observation prior as derived from the forecast ensemble and used in EnKF are clearly very beneficial.