中文摘要：

用高时空分辨率的雷达定量降水估测和预报（RQPE和RQPF）对洪水监测、发布洪水警报起重要指导作用，目前国内外对其研究的方法主要采用概率配对法、Z-I关系法和基于Z—I关系的数学算法校正法，这些方法均存在强调某一因素忽略其它因素的问题，并且对短时强降水，尤其是极强降水低估严重。从统计和个例两方面对比分析了四种雷达定量降水方法的误差，致力于寻找一种将雷达回波反射率预报场反演为降水场时误差较小的技术方法，能改进对短时强降水、特别是极强降水严重低估问题的方法，并且该方法也能在业务应用中方便、快速、精确实现。对比研究结果表明：动态分级Z—I关系法是一种对雷达QPE和回波反射率预报场反演降水场都较好的方法，并且改进了对短时强降水、尤其是大于55mm／h的短时极强降水低估的问题；在业务计算中仅需15秒，能满足6分钟更新一次计算结果的需要，而且不依赖气候统计，便于移植。

英文摘要：

Radar-based quantitative precipitation estimate （RQPE） and forecast （RQPF） with high spatial and temporal resolutions are playing very important role in flood monitoring and warning. Nowadays, researches on RQPE and RQPF are focused on probability-fitting, Z-I relations and mathematic algorithms for correction based on Z-I relations. Deficiencies somehow were found in the above methods, of which all underestimate precipitation amount in nowcasting of torrential rain, and especially seriously in extremely heavy downpour. In order to find a better way to calculate precipitation from the radar echo reflectivity, statistics and case studies are made in this article on errors of four methods on radar-based quantitative precipitation estimate and forecast. The better way will alleviate underestimation of precipitation and can be applied with easiness, readiness and improved accuracy in operational forecasting. The results show that the method of categorical Z-I relations is the best one among the four available on RQPE and RQPF. It improves the estimation of short-range precipitation for torrential rain, especially precipitation greater than 55 mm/h. The elapsed time in calculating RQPE and RQPF is only 15 seconds, thus satisfies requirement of updated observation every six minutes. In particular, the calculation doesn＇t depend on climatic statistics, therefore makes it easy to be implemented elsewhere.