地理位置处在我国高寒山区的尼洋河流域地面气象观测站点极少，是典型的缺资料地区，卫星降水数据产品是降水数据的重要补充。由于卫星降水数据在获取上的间接性及本身的不确定性，数据精度问题一直是阻碍其有效应用的主要因素。 基于集合卡尔曼滤波（EnKF）同化算法，选取TRMM、CHIRPS、Cmorph_Vl.0、PERSIANNCDR及Gldas_Noah 5种卫星降水产品，在对其与林芝站网格区域实测降水进行精度分析的基础上，进一步进行多源降水数据同化与融合。结果表明：同化后5种降水数据产品与实测降水量的相关系数CC均在0.98以上，BIAS 均在10%以下，EM在同化后均小于0.2 mm/d，RMSE均小于0. 6 mm/d, EnKF的同化效果显著。 将同化后的5种降水与原卫星降水之间的误差序列推广至全流域，从而在得全流域的5种同化降水用于融合，融合后的降水数据综合了5种阵水数据产品在精度指标的各自优势，其精度和可靠性更高。利用克里金插值法对融合后的降水进行尼洋河降水时空分布特征分析，结果显示年降水量空间上由中部向四周逐渐递减，时间上呈现逐年增加的趋势。 通过对卫星降水数据进行同化及融合．在提升降水数据产品精度的同时提供了满足水文模拟及水资源管理需求的时间序列数据，具有重要的应用价值。

There are few meteorologicol stations located in the plateau and cold mountain orea of the Niyang River Basin,China. It is a typical area that has a data shortage. Therefore, satellite precipitation data is an important supplement. Due to the indirectness of the acquisition of satellite precipitation data and its uncertainty,data accuracy has always been a major barrier to its effective application. Based on the Ensemble Kalman Filter (EnKF) assimilation algorithm, five satellite precipitation products including TRMM,CHIRPS,Cmorph_Vl.O,PERSLANN－CDR and Gldas_Noah were selected.The multi-source precipitation data assimilation and fusion were carried out because of the accuracy analysis of the measured precipitation in the grid area of Linzhi Station. The results showed that the correlation coefficients of the five satellite precipitation data products with original satellite precipitation increased up to 0. 98 after assimilation, the BIAS was below 10%, ME was less than 0.2 mm/day,and the RMSE was less than 0.6 mm/day,respectively,The effect of EnKF’s assimilation was significant. The error sequence between the assimilated five kinds of precipitation and the original satellite precipitation was extended to the whole basin so that the five assimilated precipitations in the whole hasin were used for fusion. The combined precipitation data integrated the advantages of the five precipitation data products in the accuracy index, while its accuracy and reliability were much higher. Kriging interpolation method was used to analyze the tempornl and spatial distribution characteristics of precipitation after integration in Niyang River Basin. The results showed that the spatial distribution of annual precipitation decreased gradually from the midland to the surrounding, and showed an increasing trend year by year. Through assimilation and integration of satellite precipitation data, it may provide a time series of precipitation for the requirements of hydrological simulation and water resources management, which has important application value.

国家自然科学基金（91747103；51679145；51779146）；国家重点研发计划（2016YFA0601602)