同化稀疏化处理后的雷达径向风资料有助于提升模式对短时降水的预报能力，但稀疏化方式影响得到的雷达径向风超级观测分布特征，进而影响同化及预报效果。本文基于一次华北区域暴雨天气过程，对雷达径向风观测资料开展两组不同单元格分辨率（改变径向间距或方位角间隔）的稀疏化，并进行同化模拟试验，探讨雷达径向风稀疏化方式对暴雨预报的影响。结果表明：改变稀疏化单元格的径向间距或方位角间隔，均使超级观测极值及出现位置发生变化，进而影响急流强度和位置，径向间距还通过影响超级观测极值出现的高度，进一步影响急流高度，且对获得的资料量和同化分析误差影响相对显著。同化不同分辨率的径向风超级观测对风场的调整相似，均能增加河北中南部切变线的气旋性和山东中部低空急流中的偏南分量，而超级观测的分辨率主要影响气旋性切变的曲率和南风急流的强度。对降水预报，同化雷达径向风超级观测可提升模式前6 h降水预报的整体性能，尤其是对25 mm以上强降水的捕捉能力，同时对24 h小雨和中雨量级的降水有更好的评分，并能抑制部分虚假降水预报；当采用的径向风超级观测分辨率较高时，12 h内的强降水预报性能提升显著。降水的预报技巧、空报率和命中率对稀疏化中的径向间距变化更加敏感，而方位角间隔的变化对预报偏差影响相对显著。

Assimilation of thinned radar radial wind data can help improve the model's forecasting capability for short-term precipitation. However，the thinning method affects the distribution characteristics of the radar radial wind super-observations（SOs），which in turn affects the assimilation and prediction results. This study investigates the impact of radar radial wind thinning method on rainfall forecasting through sensitive experiments. Based on a heavy rain event in North China，two experimental groups were conducted，employing radar radial wind SOs with varying grid resolutions（by altering radial spacing or azimuthal interval）. The results show that changing the radial spacing or azimuthal interval of super-observation box alters the extremes of the SOs and their locations，thereby influencing both the intensity and position of the jet stream. The radial spacing additionally affects the jet height by influencing the altitude at which the extremes of the SOs occur，and it has a relatively significant impact on the quantity of data obtained and the analysis error. The assimilation of radial wind SOs with different grid resolution similarly adjusts the wind field，which can increase the cyclone of the shear line in central and southern Hebei and the southerly component of the low-level jet in central Shandong. The grid resolution of the SOs mainly affects the curvature of the cyclonic shear and the intensity of the southerly jet. For precipitation forecasting，assimilation radar radial wind SOs can improve the overall performance of precipitation fore‐ casts for the first 6 h of the model，particularly in capturing heavy precipitation events exceeding 25 mm. Mean‐ while，it provides better scores for 24 h forecasts of light and moderate rainfall and can restrain some false precipitation forecasts. When a higher resolution of radial wind SOs is adopted，the forecasting performance for heavy precipitation within 12 h improves significantly. The threat score，false alarm ratio，and probability of detection （POD）of precipitation are more sensitive to changes in radial spacing during thinning，while variations in azimuthal interval have a relatively more pronounced impact on forecasting bias.