为能够更好地指导高寒山区融雪径流模拟研究与洪水预报方面的工作，本文以天山西部山区的喀什河流域为研究区，根据研究区内1990～2000年内的水文气象数据、CFSR再分析数据、DEM、土壤及植被等数据，借助构建的VIC模型进行径流模拟研究。根据模拟结果，可以看出：1）气象资料匮乏区可以将CFSR再分析数据作为数据源，但此数据与实际情况存在一定偏差，需进行相应修正；2）本研究以优选水文站降水数据作为CFSR再分析数据修正依据，以水文站所在CFSR再分析数据网格为标准，通过与其他CFSR网格降水数据建立相关关系进行CFSR异常数据剔除的方法是可行的；3）构建的VIC模型在本研究区具有一定的适用性，且模型参数取用较合理；4）率定期内，月尺度下模型效率系数和多年径流相对误差分别为0.80和4.7%，日尺度下分别为0.64和3.3%；而在验证期内，月尺度下模型效率系数和多年径流相对误差分别为0.84和14.5%，日尺度下分别为0.70和13.1%。

To guide the studies on snowmelt runoff simulation and flood forecast in alpine regions, in this paper, we took the Kashi River basin in Western Tianshan Mountains as the study area, and built a VIC model to conduct runoff simulation study based on the hydro-meteorological data, CFSR reanalysis data, DEM, soil and vegetation data of the study area during 1990 to 2000. It was found that 1) CFSR reanalysis data can be used as a data source in areas lacking meteorological data, but they should be calibrated as they deviate from the real situation. 2) It is feasible to remove CFSR abnormal data by using the precipitation data of an appropriate hydrological station as the basis on which CFSR reanalysis data are calibrated, taking the CFSR reanalysis data grid of the hydrological station as the standard, and establishing its correlations with other CFSR grids. 3) The established VIC model had certain applicability in the study area, and the parameters in this model were reasonable. 4) In the calibration period, the relative errors of model efficiency coefficient and multi-year runoff were 0.80 and 4.7% respectively on the monthly scale, and 0.64 and 3.3% respectively on the daily scale. In the verification period, the relative errors of model efficiency coefficient and multi-year runoff were 0.84 and 14.5% respectively on the monthly scale, and 0.70 and 13.1% respectively on the daily scale.

国家自然科学基金项目(51469034,51209181,51569031)；新疆自治区地方公派出国留学成组配套项目(XJDF201307)；新疆水文学及水资源重点学科基金项目( xjswszyzdxk20101202)