多源融合降水产品SUPER在汉江流域的时空精度评估  

Spatiotemporal Accuracy Assessment of Multi-source Fusion Precipitation Product SUPER in the HanJiang River Basin

作　　者：刘松岩 魏玲娜[1,2] 董建志 葛慧 齐丹[5] LIU Songyan;WEI Lingna;DONG Jianzhi;GE Hui;QI Dan(Keylaboratory of Hydrometeorological Disasters Mechanism and Warning of Ministry of Water Resources,Nanjing University of Information Science&Technology,Nanjing 210044,China;School of Hydrology and Water Resources,Nanjing University of Information Science&Technology,Nanjing 210044,China;School of Earth System Science,Tianjin University,Tianjin 300072,China;Institute of Surface-Earth System Science,Tianjin University,Tianjin 300072,China;National Meteorological Center(China Meteorological Administration),Beijing 100081,China)

机构地区：[1]南京信息工程大学水利部水文气象灾害机理与预警重点实验室,南京210044 [2]南京信息工程大学水文与水资源工程学院,南京210044 [3]天津大学地球系统科学学院,天津300072 [4]天津大学表层地球系统科学研究院,天津300072 [5]国家气象中心(中央气象台),北京100081

出　　处：《地球信息科学学报》2024年第5期1335-1349,共15页Journal of Geo-information Science

基　　金：国家自然科学基金项目(52179021、51909121)。

摘　　要：当前的多源降水数据融合技术多基于地面雨量站观测进行校正与融合,但地面观测本身存在一定的不确定性,尤其在地面雨量站点匮乏的地区,基于数理不确定性理论的多源降水融合技术可以在不依赖于地面观测数据的基础上确定各降水数据集的误差及其互相关性,逐格点确定最优晴雨分类时间序列,有效提高了融合产品的可靠性。为了评估由该理论框架发展而来的新一代降水数据优化融合产品SUPER(Statistical Uncertainty Analysis-based Precipitation mERging Framework),本研究在汉江流域上游汉中流域以及中游郭滩流域,利用地面高密度雨量站点的观测数据,选取再分析降水数据ERA5(ECMWF Reanalysis v5)、卫星降水产品IMERG(Integrated Multi-satellitE Retrievals for GPM),从不同时空尺度,对比SUPER的精度及性能。结果表明:①相比ERA5与IMERG,SUPER产品在日、月两种尺度上均具有更优表现,与地面站点实测降水一致性更高、误差更小、误报率最低、探测成功率最高;②SUPER产品在地势平缓地区比在地形复杂地区性能更佳,地形对SUPER产品精度的影响在降水事件识别能力上更为显著,在地形复杂地区降水产品的精度会随地势升高而降低;③SUPER的融合算法及数据集可以有效的降低降水数据的随机误差与晴雨分类误差,但对系统偏差处理相对较为简单,仍有较大的改进空间。本研究综合的分析了SUPER产品在研究区的性能,为后续产品的实际应用与改进奠定了研究基础。At present,the multi-source precipitation data fusion technology is mostly based on the correction and fusion of ground rainfall station observations.However,there are still uncertainties in ground observations,especially in areas with a scarcity of ground rainfall stations.The multi-source precipitation fusion technique based on the theory of mathematical uncertainty can determine the errors and inter-correlations of each precipitation dataset without relying on ground observation data and establish optimal clear-rain classification time series for each grid,effectively enhancing the reliability of the fusion products.This study aims to evaluate the performance of the new generation precipitation data optimization fusion product,Statistical Uncertainty analysis-based Precipitation mERging framework(SUPER),developed based on this theory.The assessment is conducted in the upper Hanzhong basin and the middle Guotan basin of the Han River basin.Ground-based high-density rainfall station data along with ECMWF Reanalysis v5(ERA5)reanalysis precipitation data and Integrated Multi-satellitE Retrievals for GPM(IMERG)satellite precipitation products are used to evaluate the accuracy and performance of SUPER at various spatiotemporal scales.The results show that:(1)Compared with ERA5 and IMERG,SUPER product performs better at daily and monthly scales.SUPER has a higher consistency with ground measured precipitation,smaller errors,the lowest false alarm rate,and the highest detection success rate;(2)SUPER precipitation product performs better in regions with gentle terrain than in areas with complex topography.The accuracy of SUPER is higher in the Guotan Basin than in the Han River Basin.SUPER has a higher CSI index in the Guotan Basin.In areas with complex topography,the accuracy of precipitation products decreases with the increase of elevation,for example,in the Han River Basin,the accuracy of SUPER product decreases with increasing elevation from the south to the north;(3)SUPER's fusion algorithm and datasets can effec

关 键 词：降水产品 SUPER IMERG ERA5 时空尺度 精度评估 汉江流域 

分 类 号：P332[天文地球—水文科学]