寒旱区流域受降雨径流和融雪径流联合补给，坡面产流和融雪过程可能对流域水文产生重要影响。本文以分布式水文模型SWAT为平台，选取模型多个参数为关键因子，借助EFAST方法探索融雪期（3月-5月）、非融雪期（6月-次年2月）的径流峰值以及全时段径流均值的参数敏感性，这些参数涉及降雨径流、积雪消融、蒸散发、下渗、地下水补给和壤中流等多种水文过程。同时，结合流域特征及参数物理意义，深入分析参数敏感的原因，并揭示参数背后的水文过程对开都河流域产汇流的重要影响。研究发现，坡面产流、下渗以及积雪消融等水文过程对开都河流域水循环具有重要影响。此外，对于寒旱区流域重要的融雪过程参数，其总敏感性显著而一阶敏感性不显著，表明通过EFAST方法得到的水文模型参数总敏感性更为合理。结果揭示了流域水文敏感因子及关键过程，为探索水循环机理，水文科学预测、管理流域水资源奠定了基础。

In cold and arid watersheds, streamflow is mainly contributed by snowmelt runoff in the ablation period and rainfall runoff during summer. Surface runoff and snowmelt may have important influence on the water cycle. In this study, the extended Fourier amplitude sensitivity test (EFAST) method was employed for SWAT model parameter sensitivity analysis. The test involved hydrological parameters that describe multiple processes including rainfall runoff, snowmelt, evaporation, percolation, baseflow, and lateral flow. Peak flows in non-melting and snow-melting periods as well as average annual flow were all considered in the sensitivity analysis. In addition, the reasons for the sensitivity of the parameters and the associated hydrological processes that affect the water cycle in the watershed were analyzed. Results showed that the surface runoff, percolation, and snowmelt have significant influence. Besides, the total-order sensitivity of the parameters related to snowmelt process was significant whereas the first-order sensitivity was not significant, indicating that the total-order sensitivity obtained using EFAST is more reasonable. The findings will help hydrological models improve their capability for simulating/predicting water resources during different seasons for snowmelt-precipitation-driven watersheds.

中央高校基本科研业务费专项资金（2014XS69）