【摘要】 针对雷达降水估测产品在复杂地形区域强降水监测能力不足的问题，利用风云4号B星（FY-4B）先进静止轨道辐射成像仪红外辐射数据，广西地面气象观测站雨量数据和柳州雷达降水估测产品，基于深度神经网络模型与订正方法，实现雷达与卫星降水信息的动态优化融合。结果表明，基于站点观测的雨量和GPM降水数据评估，融合降水产品相对雷达降水估测的暴雨落区更加精准，对大于1 h的30 mm雨量的强降水低估现象改善明显。融合FY-4B卫星红外资料能有效弥补雷达在复杂地形下监测强降水能力不足的问题。更多还原

【Abstract】 In response to the problem of insufficient capabilities of radar precipitation estimation products in monitoring heavy rainfall over complex terrain areas, using the infrared radiation data of FengYun-4B(FY-4B)advanced geostationary orbit Radiation Imager(AGRI), rainfall data from Guangxi ground meteorological observation stations and the precipitation estimation products of Liuzhou radar, this study achieved dynamic optimization merging of radar and satellite precipitation information based on a deep neural network model and correction methods. The results show that, as evaluated against station-observed rainfall and GPM precipitation data, the fusion precipitation products are more accurate in identifying the rainfall area compared to radar-only estimates, and significantly reduces the underestimation of heavy precipitation exceeding 30 mm within 1 h. The inclusion of FY-4B infrared data effectively compensates for the inadequate capability of radar in monitoring heavy rainfall over complex terrain.更多还原