Abstract Radar reflectivity contains information about hydrometeors and plays an important role in the initialization of convective-scale numerical weather prediction (NWP). In this study, a new background-dependent hydrometeor retrieval method is proposed and retrieved hydrometeors are assimilated into the Weather Research and Forecasting model (WRF), with the aim of improving short-term severe weather forecasts. Compared to traditional approaches that are mostly empirical and static, the retrieval parameters for hydrometeor identification and reflectivity partitioning in the new scheme are extracted in real-time based on the background hydrometeor fields and observed radar reflectivity. It was found that the contributions of hydrometeors to reflectivity change a lot in different reflectivity ranges and heights, indicating that adaptive parameters are necessary for reflectivity partitioning and hydrometeor retrieval. The accuracy of the background-dependent hydrometeor retrieval method and its impact on the subsequent assimilation and forecast were examined through observing system simulation experiments (OSSEs). Results show that by incorporating the background information, the retrieval accuracy was greatly improved, especially in mixed-hydrometeor regions. The assimilation of retrieved hydrometeors helped improve both the hydrometeor analyses and forecasts. With an hourly update cycling configuration, more accurate hydrometeor information was properly transferred to other model variables, such as temperature and humidity fields through the model integration, leading to an improvement of the short-term (0−3 h) precipitation forecasts.

中文翻译：

---

基于背景水汽反演的雷达反射率数据同化方法：观测系统模拟实验

摘要 雷达反射率包含有关水凝物的信息，在对流尺度数值天气预报（NWP）的初始化中起着重要作用。在这项研究中，提出了一种新的背景相关水凝物反演方法，并将反演的水凝物同化到天气研究和预测模型（WRF）中，旨在改进短期恶劣天气预测。与传统的多为经验和静态的方法相比，新方案中的水汽识别和反射率划分的反演参数是基于背景水凝物场和观测到的雷达反射率实时提取的。发现水凝物对反射率的贡献在不同的反射率范围和高度上变化很大，表明自适应参数对于反射率划分和水凝物反演是必要的。通过观测系统模拟实验 (OSSE) 检验了依赖背景的水凝物反演方法的准确性及其对后续同化和预报的影响。结果表明，通过结合背景信息，检索精度大大提高，特别是在混合水汽区。同化回收的水凝物有助于改进水凝物分析和预测。通过每小时更新的循环配置，更准确的水凝物信息通过模型集成被适当地传输到其他模型变量，例如温度和湿度场，从而改善了短期（0-3 小时）降水预测。