多源遥感降水评估及其在水文模拟中的应用  

Comprehensive Comparisons of Multi-source Remote Sensing Precipitation Estimates for Hydrological Applications

作　　者：高真 黄本胜[1,2,3] 陈晓宏 邱静[1,2,3] GAO Zhen;HUANG Ben-sheng;CHEN Xiao-hong;QIU Jing(Guangdong Research Institute of Water Resources and Hydropower,Guangzhou 510610,China;National and Local Joint Engineering Laboratory of Estuary Hydropower Technology,Guangzhou 510635,China;Guangdong Water Security Collaboratine Innovation Center,Guangzhou 510635,China;Center for Water Resources and Environment Research,Sun Yat-sen University,Guangzhou 510275,China)

机构地区：[1]广东省水利水电科学研究院,广州510635 [2]河口水利技术国家地方联合工程实验室,广州510635 [3]广东省水安全科技协同创新中心,广州510635 [4]中山大学水资源与环境研究中心,广州510275

出　　处：《中国农村水利水电》2021年第4期27-32,共6页China Rural Water and Hydropower

基　　金：国家自然科学基金项目(U1911204,51861125203);中国博士后科学基金项目(2019M662828);广东省水利科技创新项目重点项目(2015-08);珠江三角洲水资源配置工程课题研究(WW2018230)。

摘　　要：多源遥感降水产品为无资料或缺资料地区的水文过程模拟提供了新的高质量的数据来源。以美国爱荷华州Iowa-Cedar中型流域为例,基于地面雨量站降水数据CPC-U定量评估了5种多源遥感降水产品(NLDAS2、Stage IV、TMPARP、TMPART和IMERG)的精度,并结合分布式水文模型DRIVE(Dominant river tracing-Routing Integrated with Variable Infiltration Capacity Environment)进行了水文效用评估。研究表明:①NLDAS2和Stage IV都能很好地捕捉降水事件,与CPC-U具有良好的一致性,其次是IMERG和TMPARP,日尺度上相关系数达到0.63~0.80,偏差为-0.21%~15.37%,探测率为0.61~0.74,而TMPART产品精度较低;②采用不同降水数据驱动分布式水文模型进行径流模拟,NLDAS2和CPC-U表现效果最好,纳什效率系数分别为0.82和0.8,其次是Stage IV和TMPARP,纳什效率系数分别为0.79和0.63,而IMERG和3B42RT表现较差,纳什效率系数分别为0.46和-1.09。对比结果表明,NLDAS2、Stage IV降水产品精度评估表现较好,适合在Iowa-Cedar流域进行水文模拟应用,其次是TMPARP和IMERG,但IMERG水文效用评估相对较差,说明多源遥感降水数据融合产品更适合于水文模拟应用。Multi-source remote sensing precipitation estimates with high spatial and temporal resolution have provided hydrologists a potential alternative source for hydrological simulations in ungauged or poorly gauged basins,especially for the medium-and small-size rivers lack of consistent and long-term records.This study evaluates the accuracy and performance of the five remote sensing precipitation products(NLDAS2,Stage IV,TMPARP,TMPART,and IMERG)against CPC-U and then estimates their hydrological performance with the distributed hydrologic model DRIVE(Dominant river tracing-Routing Integrated with Variable Infiltration Capacity Environment)during 2002-2013 over the Iowa-Cedar River basin,a mid-size basin in Iowa.Results indicate that:①NLDAS2 and Stage IV perform better than the other three datasets against CPC-U data at daily scale,and IMERG correlates slightly better than TMPARP;these four precipitation products can well capture the precipitation events with the CC(correlation coefficient)values 0.63~0.8,the BIAS(relative bias)values are-0.21%~15.37%,and the POD(probability of detection)values 0.61~0.74,while TMPART performs the worst.②Using the same calibrated parameter sets from the Iowa-Cedar River basin,NLDAS2,CPC-U,Stage IV,and TMPARP-driven distributed hydrologic model performs good with the NSCE(Nash-Sutcliffe coefficient of efficiency)values of 0.82,0.80,0.79,and 0.63,respectively;while IMERG and 3B42RT performs poorly overall with the NSCE value of 0.46 and-1.09.Comparison shows that NLDAS2 and Stage IV precipitation products demonstrate the good hydrologic utility in the Iowa-Cedar River basin,followed by TMPARP and IMERG,indicating that multi-source remote sensing-based precipitation estimates is more suitable for flood modeling.

关 键 词：遥感降水 NLDAS2 IMERG DRIVE模型 水文模拟 

分 类 号：TV121.1[水利工程—水文学及水资源]