基于多源降水观测的全球高分辨率再分析降水融合方法研究

Accurate quantitative global high-resolution fine-scale precipitation is critical information for advancing our ability to recognize the laws of current atmospheric motion, understand the characteristics of past weather and climate evolutions, and predict the trends of future weather and climate changes. Given the limitations inherent in the individual single-source estimates of global precipitation, the methods to merge them appropriately so as to take advantage of the strengths of each source has become the necessary way to produce reliable global precipitation analyses. Using objective statistical analysis methods and optimization interpolation algorithms, this study intent to carry out a new merged method for producing global high-resolution reanalysis precipitation based on multi-source fine-scale precipitation observations. The new merged method which based on cumulative distribution function matching technique and optimal time smoothing interpolation scheme is suitable for combing global fine-scale precipitation data from rain gauge observations, satellite observations, and numerical model outputs, and will eventually produce a global hourly high-resolution gridded precipitation reanalysis dataset. This study will not only be used to provide accurate global precipitation data for weather and climate researches and operational services, but will also be applied to test simulation performances of numerical models, deepen the understanding of model uncertainties, and improve the skill of weather forecasts and short-term climate prediction. This study has important scientific significance and excellent application prospect.

准确的全球高分辨率精细尺度降水资料是认识当前大气运动规律，了解过去天气气候演变特征以及预测未来天气和气候发展趋势的一个重要信息来源。鉴于单源降水资料在全球降水估计时的局限性，能充分结合不同来源资料优势的资料融合方法成为获取可靠全球降水信息的必要途径。本项目拟结合客观统计分析方法和最优插值方法，开展基于多源降水观测的全球高分辨率再分析降水融合方法研究。在我国已有的高分辨率多源降水资料基础上，发展一种基于累积概率密度函数匹配技术和最优时间平滑插值技术的精细尺度降水资料融合方法，对全球台站降水、卫星降水和模式降水资料进行合理融合，最终生成一套全球高分辨率网格化降水再分析资料集。本项目的研究，不仅能为我国天气气候研究和业务提供准确的全球降水基础数据，还能进一步检验数值模式的模拟性能，深化对模式不确定性的认识，提高天气预报和短期气候预测水平，具有重要的科学价值和应用潜力。

为解决我国全球尺度高分辨率格点降水资料过度依赖国外的问题,本项目结合定量客观统计分析方法和最优插值方法，开展了基于多源降水资料的全球高分辨率格点降水融合方法研究。通过发展一种基于累积概率密度函数匹配技术和最优时间平滑插值技术的精细尺度降水资料融合策略，实现对全球尺度的台站降水、卫星降水和模式降水资料的最优融合，生成国内首套全球精细尺度多源格点融合降水资料BMEP（BCC Merged Estimation of Precipitation），BMEP逐日降水在全球尺度整体质量与国际先进的GPCP同类产品相当，在中国区域的质量则要优于GPCP，且实时性明显优于GPCP。该数据集填补了我国在全球尺度格点降水产品研制上的空白，并已经应用于国家气候中心次季节-季节尺度（S2S）气候预测业务模式中，取得良好的业务应用效果。本项目工作对我国当前气候业务模式初始化和全球降水监测业务有明显贡献。