SWAT模型融雪模块的改进与应用研究

Research on Improvement and Application of Snowmelt Module in SWAT

融雪是水文过程的重要影响因素,为了对天山北坡气候极端高寒地区进行长期洪水预报预警,该研究以SWAT(土壤和水评估工具)模型为基本框架对其融雪模块进行修改,将基于物理过程的分布式融雪径流模型耦合入SWAT模型,并基于高程带对改进前后的模型进行了对比验证。通过对新疆天山北坡军塘湖河流域进行模拟、参数率定与验证后发现,原SWAT模型在参数率定期间(2000年1月1日至2004年12月31日)的Nash-Stucliffe效率系数(NSE)略低于基于物理过程的模型SWAT-JTH,但在验证期(2006年1月1日至2010年12月31日),SWAT原模型NSE高于改进后的SWAT-JTH模型。两种方法模拟效果均可以满足实际融雪模拟精度的需求。研究发现,温度因子方法(SWAT融雪模块)虽较基于物理过程的模型结构更简单,但在合理率定参数后却可以得到较好的效果。虽然现今观测和实验手段还不能满足物理过程模型精度要求,以及物理模型本身的错误叠加现象对目前的研究带来了诸多不确定性,但由于物理模型对于极端条件下的模拟及物理过程的模拟具有很大优势,未来随数据精度的提高,将继续深入该物理模型的改进研究。该研究将对高寒山区分布式融雪径流预报预警的发展起到一定的促进作用。

Snowmelt process is an important part in hydrological analyses. Many hydrology models have interduced snowmelt model in order to simulate the snowmelt processes in the catchment ranging from simple temperature-index method to sophisticated process-based model. In this paper, we make a comparison of the two approaches to identify the performances of temperature-index and process-based snowmelt approaches which coupled with the SWAT model in Juntanghu watershed of northern slope of Tianshan Mountains, Xinjiang, China. The result indicates insignificant differences between the temperature-index methods and physically energy budget based snowmelt approaches. The simulation performance from the original SWAT model in calibration period （2000-01-01-2004-12-31） was worse than that from the developed SWAT-JTH model; in validation period （2006-01-01-2010-12-31）, the original SWAT model provided a better performance than the SWAT-JTH model. The simulation results are acceptable for snowmelt runoff simulation for both models. Through comparison, the structure of temperature-index method is simple and can achieve good simulation performance if the model is well calibrated. Although it is hard to meet the data requirement of the physically-based model in current observation and experiment approaches and some uncertainties associated with physical processes, there are some advantages for the physically-based modet which can simulate the extreme weather conditions. In this study, the physically-based distributed snowmelt runoff model was developed and has a promising future.