“一带一路”重点地区所处的亚欧大陆是世界上气象灾害种类最丰富、影响范围最大的区域之一，每年因强降水事件造成的人员和财产损失占全球的一半以上。研究“一带一路”重点地区强降水特征及演变趋势、研发高准确率的沿线国家和地区的降强水预报技术，对于保障“一带一路”各国合作与经济建设具有重要的意义。本文聚焦“一带一路”重点地区，系统分析研究了近30年（1991—2020年）强降水的特征及变化趋势，统计计算得出了“一带一路”6个重点区域强降水定义的相对阈值；通过多源融合技术得到了逐日降水实况产品，一定程度上解决预报业务中观测数据不足，时间滞后等实际问题；使用基于动态权重系数的多模式集成方法研发了“一带一路”强降水预报方法，研究成果为我国“一带一路”气象业务建设提供了一定的科学和技术支撑。主要结论如下：
（1）近30年以来，“一带一路”重点地区强降水在频率、强度、持续性均呈现增加的变化趋势，其中春季增速最快，夏秋两季其次，冬季增速最慢。空间分布上，南亚、东南亚、西亚平均日降水量增加的趋势最明显，蒙俄、中亚、中东欧呈现下降趋势。利用百分位法计算确定了“一带一路”重点地区的24小时强降水相对阈值，南亚地区24小时强降水相对阈值为90mm、东南亚地区为150mm、西亚地区为20mm、中东欧地区为70mm、蒙俄地区为30mm、中亚地区为20mm。阈值的确定可以客观分离出对当地造成高影响的降水事件，也可以在预报中快速识别强降水的发生时间和地区。
（2）采用地面自动观测站、Himawari—8卫星反演、FY—4A卫星反演、地面雷达反演与EC数值预报模式零场等多源数据，通过多源融合技术，得到可用于业务使用的逐日降水时间序列产品（多源融合降水产品，China National Meteorological Centre Multiple Sources, CMS）。CMS产品与全球交换站的降水量观测数据相比，偏差小于±2mm的站点数达到99.3%，且离散性低，对雨带位置、降水量级的反应接近于观测实况，较实时地面观测数据仅滞后6小时。CMS的评估结果表明其可以替代地面降水观测数据，满足气象预报业务上对空间分布和时效性的要求。
（3）利用多源融合降水产品与不同数值预报模式以及集合预报结果，研究全球智能网格降水技术方法，基于“一带一路”重点地区强降水定义阈值基础，开发无缝隙精细化的强降水网格预报，增强了中央气象台对该地区降水产品的技术支撑空白。对预报结果的评估为各区域强降水TS评分为0.18～0.35，显著高于数值预报模式的直接预报结果。在短临预报时效内，较数值模式预报准确率平均提升了25%，在短中期时效上，较数值模式预报准确率提升10%以上。
本文研究的“一带一路”重点地区的强降水特征及预报技术，采用融合动态权重方法、滚动更新方法、多模式自适应最优集成方法等，研发全新的全球智能网格降水预报技术，提高了“一带一路”重点区域的强降水预报的精细化和准确率，为重点区域的经济合作项目建设提供重要气象预报信息，减少各类强降雨灾害造成的损失。

The Asia and Europe region, where the "Belt and Road" focus areas are located, is one of the world's richest regions in terms of the variety of meteorological hazards and the scope of impact. The study of heavy precipitation characteristics and trends in key areas of the "Belt and Road", and the development of highly accurate precipitation forecasting techniques for countries and regions along the "Belt and Road", are of great importance in ensuring the cooperation and economic construction of the countries along the "Belt and Road". This paper focuses on the key regions of " Belt and Road ", systematically analyses and studies the characteristics and trends of heavy precipitation in the past 30 years (1991-2020), and statistically calculates the relative thresholds for the definition of heavy precipitation in six key regions of " Belt and Road "; through multi-source fusion technology, it obtains daily precipitation products, which to a certain extent solves the practical problems of insufficient observation data and time lag in forecasting operations; using a multi-model integration method based on dynamic weighting coefficients, it develops a method for forecasting heavy precipitation in "One Belt, One Road", and the research results provide some scientific and technical support for the construction of meteorological operations in " Belt and Road " in China.The main conclusions are:
(1) Over the past 30 years, the frequency, intensity and persistence of heavy precipitation in the key regions of the Belt and Road have shown an increasing trend, with the fastest increase in spring, followed by summer and autumn, and the slowest increase in winter. In terms of spatial distribution, South Asia, Southeast Asia and West Asia show the most obvious trend of increasing average daily precipitation, while Mongolia, Russia, Central Asia and Central and Eastern Europe show a decreasing trend. The relative thresholds for heavy precipitation pre 24 hours in the key regions of the Belt and Road were determined by using the percentile method, with 90mm in South Asia, 150mm in Southeast Asia, 20mm in West Asia, 70mm in Central and Eastern Europe, 30mm in Mongolia and Russia, and 20mm in Central Asia. allows for the objective separation of precipitation events with high local impact, as well as the rapid identification of the time and area of occurrence of heavy precipitation in the forecast.
(2) The multi-source data from ground-based automatic observation stations, Himawari-8 satellite inversion, FY-4A satellite inversion, ground-based radar inversion and EC numerical forecast model zero field are used to obtain an operational daily precipitation time series product (China National Meteorological Centre Multiple Sources, CMS) through multi-source fusion technology. Compared with precipitation observations from global exchange stations, the CMS product has 99.3% deviation of less than ±2 mm and low dispersion, and its response to the location of rainbands and precipitation levels is close to the observed reality, with a lag of only 6 hours compared to real-time ground observations. The results of the CMS evaluation show that it can replace ground precipitation observations and meet the requirements of spatial distribution and timeliness in meteorological forecasting operations.
(3) Using the multi-source fusion precipitation products, different numerical forecasting models and ensemble forecasting results, the study of the global intelligent grid precipitation technology method, based on the " Belt and Road " key areas of heavy precipitation definition threshold basis, the development of seamless refinement of heavy precipitation grid forecasts, enhance the technical support of the Central Weather Bureau for the region's precipitation products gap. The forecast results were assessed to have a TS score of 0.18 to 0.35 for each region, which is significantly higher than the direct forecast results of the numerical forecast model. The forecast accuracy was improved by an average of 25% over the numerical model for the short proximity forecast time period and by more than 10% over the numerical model for the short to medium term time period.
This paper investigates the characteristics and forecasting techniques of heavy precipitation in the key regions of the Belt and Road, using a combination of dynamic weighting methods, rolling update methods and multi-model adaptive optimal integration methods to develop a new global intelligent grid precipitation forecasting technique, which improves the refinement and accuracy of heavy precipitation forecasting in the key regions of the Belt and Road, providing important meteorological forecasting information for the construction of economic cooperation projects in the key regions and reducing the losses caused by various heavy rainfall disasters.