近实时卫星降水的误差解析、数据融合和水文集合模拟研究

项目名称： 近实时卫星降水的误差解析、数据融合和水文集合模拟研究

项目编号： No.51479118

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 水利工程

项目作者： 胡庆芳

作者单位： 水利部交通运输部国家能源局南京水利科学研究院

项目金额： 84万元

中文摘要： 近实时卫星降水数据的不断涌现为地面降水资料稀缺地区的水文研究与实践提供了新机遇。科学认识这些数据的误差特征，充分发挥其水文应用潜力是一个开放性的科学问题。鉴于此，本项目拟在汉江、岷江和青海湖流域，开展TRMM 3B42RTV7、CMORPH、GSMAP_NT和PETSIANN四种代表性近实时卫星降水资料的误差综合解析、数据融合和水文集合模拟研究。主要内容：（1）完善综合评价指标，在多种时空尺度上阐明卫星降水的误差特征与影响因素；（2）针对不同地面雨量资料条件，发展集成多种信息的卫星降水融合新方法，辨识其对提高降水估计精度的增益；（3）基于Bayesian模型平均，采用卫星降水及其融合数据，开展径流、干旱集合模拟，揭示降水误差导致的不确定性并提出有效的削减方法。项目成果对于合理使用近实时卫星降水数据，提高资料稀缺流域的水旱监测预报能力具有重要意义，还将为卫星降水反演算法的改进提供反馈信息。

中文关键词： 近实时卫星降水融合；水文集合模拟；空间降尺度；地理加权回归；贝叶斯模型平均

英文摘要： In recent, various developed real-time satellite rainfall datasets and their sustaining improved quality provides new opptrunities both for hydrology research and practice, especially in those areas with scare gauge mesurements. Hence, how to anlyze the accuracy of these datesets and fully tap into their potentiality has become a siginificant and open problem for hydrology communicty. Motivated by this, choosing three important areas in China, the Hanjing River Basin, Mingjiang River Basin and Qinghai Lake Basin, this project plans to conduct a deep research for the error characteristics analysis, data merging and hydrologic esemble simualtions using the four representative real-time satellite rainfall datasets, namely TRMM 3B42RTV7, CMORPH, GSMAP_NT and PETSIANN. Main proposals include the following items: (1) Using improved assesing indices, to compresneivey illustrated the error distrutions rules for various real-time satellite rainfall estimates and their related influencing factors; (2) For various surface rainfall gauge density and configuration, to propsed new rainfall merging alogrithms capable of combining various satellite rainfall information with gauge measurements and elucidate their gain for improve the rainfall estimation. (3) Based on the Baysian model averaging framework, runoff and drought emsemble simualtion will be conducted both using purely near real-time satelllite rainfall and their merged rainfall. The uncertainty of simulated runoff and drought process caused by rainfall error will be anlyzed. Also, measurements to reduce the uncertainty will be considered. The expected achievements of the project are significant for rational utlization of real-time satellite rainfall datasets to improved our capability for rainfall flood and drought hazard monitoring&mitigation. At the same time ,new findings will be fed back into the development of satellite rainfall retrieving alogrithms.

英文关键词： Near real-time satellite rainfall merging;Hydrologic ensemble simulations;Spatial downscaling;Geographically weighted regression;Bayesian model averaging