新一代多卫星遥感反演降水的流域水文模拟和预报能力研究

Precipitation is a critical variable to hydrologic simulation and flood prediction. As a prelude to NASA's planned Global Precipitation Measurement (GPM), current Multi-satellite Precipitation Retrievals are intended to provide the best real-time precipitation estimates with higher spatiotemporal resolution at quasi-global scale. The integration of multi-satellite precipitation estimates to distributed hydrological models provides hydrologists with an opportunity to improve hydrological process simulation and flood prediction capability for large river basins, especially in the remote regions where in-situ precipitation and stream gauge networks are sparse. In this proposal, two tested-grids (approx. 25km×25km) with dense ground observation networks for precipitation will be constructed within the semi-arid Laohahe Basin located in northeast China and the southern humid Mishui Basin, respectively. We intend to install some new in-situ instruments of tipping-bucket rain gauges within these two tested-grids to benchmark the satellite rainfall products. The overarching goal of this proposal is to investigate the accuracy and error of current high resolution real-time satellite-borne precipitation estimates (i.e., TMPA-RT, CMORPH, PERSIANN) and assess their hydrologic application. This will be evaluated using the observations from the previously described two instrumented basins. Further, we will investigate and identify the key factors that affect the accuracy of three mainstream satellite precipitation estimates from the perspective of precipitation retrievals. Then,the satellite precipitation estimates will be integrated into an improved distributed hybrid hydrologic model for hydrologic simulation at daily scale and flood prediction at 3-hourly scale over our study basins. Additionally, a kind of satellite rainfall error model (SREM2D) will be used to characterize the multidimensional error structure of satellite-driven flood prediction at different spatial resolutions. Finally, we will investigate how the introduce of several new sensors and crucial algorithm upgrades affect the potential of satellite precipitation in hydrologic simulation and prediction.The research results of this project can provide theory and technology references for the application of forthcoming Chinese Precipitation Radar Satellite in the flood prediction.

降水是水文模拟和预报的关键。作为全球降水观测计划（GPM）的前身，新一代多卫星遥感降水反演技术的出现使得低成本快速获取数据质量更好、时空分辨率更高、覆盖范围更广的实时连续降水资料成为可能，多卫星遥感降水与分布式水文模型的集成为大尺度流域（特别是无资料或少资料地区）的水文模拟和洪水预报提供了新的契机。本项目拟在中国南北两个典型流域各建立一个加密实验格网，并结合站点观测数据对三种国际上主流的多卫星遥感降水进行地面验证，从反演机理上探讨影响遥感降水精度的关键因素；将遥感降水与分布式垂向混合产流模型进行集成，定量评估三种主流遥感降水在典型流域的水文模拟和预报能力；解析不同空间分辨率遥感降水在洪水预报中的多维误差结构，揭示新传感器引入和关键算法更新对遥感降水的水文模拟和预报能力的影响。本项目研究成果可以为我国即将发射的降雨雷达星在洪涝灾害预报中的应用提供重要的技术借鉴和理论参考。

降水是水文模拟和预报的关键。新一代多卫星遥感降水反演技术的出现使得低成本快速获取数据质量更好、时空分辨率更高、覆盖范围更广的实时连续降水资料成为可能，多卫星遥感降水与分布式水文模型的集成为大尺度流域（特别是无资料或少资料地区）的水文模拟和洪水预报提供了新的契机。. 本项目开展的主要研究及重要结论包括：① 建立了中国第一个全球降水计划GPM地面验证基地。② 发展了新一代多卫星遥感降水联合反演的新方法，揭示了卫星降水实时反演系统产生高雨强低估、低雨强高估的成因，发现了全球降水反演中日、月尺度系统误差与小时尺度随机误差之间的相互转移特性，探明了系统误差的雨强均化效应与随机误差的尺度变化特征对流域洪旱模拟与预报的影响。③ 实现了新一代多卫星遥感降水与分布式水文模型TOPX的耦合，在两个典型流域进行大尺度水文过程模拟，提出了针对多卫星联合反演的贝叶斯滑动平均集合模拟方法，揭示了主流遥感降水的水文误差传递特性，完善了中高纬度流域地面观测与卫星降水的融合反演机理。④ 实现了GPM卫星降水的传感器层面误差源追踪，开展了星载雷达对地形雨监测的适用性分析，对不同卫星传感器的不同算法进行了误差溯源分析与订正。研究成果可为我国第一颗全球水循环观测卫星（WCOM）的产品研发及其水文应用提供算法基础和理论支撑，在国家重大需求上具有重要的战略价值。. 在本项目支持下，项目组共发表与本项目内容密切相关的SCI论文18篇，其中ESI高被引论文2篇，一区2篇、二区8篇。研究成果获2014年度教育部自然科学一等奖1项（项目负责人排名第2）。此外，授权与本项目有关的发明专利2项。本项目共培养博士研究生3人、硕士研究生14人（毕业7人）。2015年，本项目核心代表作发表在大气遥感领域旗舰杂志《美国气象协会公报》（BAMS，IF=11.808），论文入选地学领域ESI高被引，被PNAS等杂志作为亮点引用，该研究得到全球降水计划首席科学家George J. Huffman 的充分肯定，核心算法已被应用于全球降水计划GPM系统。

内容获取失败，请点击重试