Time-Extended Bayesian Retrieval of Dual-Polarization Radar Data Enhancing Short-Term Precipitation Forecasts

In numerical weather prediction (NWP) models, radar data are commonly utilized to retrieve relative humidity fields, thereby mitigating forecast errors arising from uncertainties in the initial moisture field. This study addresses the degradation in convective structure retrieval capability caused by temporal biases in NWP when using spatial neighborhood sampling methods with radar data for relative humidity field retrieval. We developed a time-extended Bayesian retrieval method and constructed a dual-polarization radar data assimilation framework compatible with the China Meteorological Administration Mesoscale Model (CMA-MESO). The core of this approach lies in expanding the Bayesian retrieval sample library by integrating forecast fields from two hours before and after the assimilation time, thereby improving the inadequate performance of traditional spatial sampling under model temporal bias. When applied to a heavy rainfall event in North China in July 2023, this method effectively reduced retrieval errors. The subsequent assimilation of retrieved humidity fields enhanced the Threat Scores for 0–6 h precipitation forecasts and demonstrated improvement in overprediction bias. This confirms that the time-extended strategy can effectively enhance moisture field analysis and nowcasting accuracy by overcoming the inherent limitations of spatial-only sampling.