SWAT模型参数自动校准方法对比及适用性研究——以泾河中上游地区为例  

Comparison and Applicability of Automatic Calibration Methods of the Parameters in SWAT Model——A Case Study of the Middle and Upper Reaches of the Jinghe River Watershed

作　　者：张丽[1] 柳烨[1] 蔡朵朵 樊琨[1] 马孝义[1] ZHANG Li LIU Ye CAI Duo-duo FAN Kun MA Xiao-yi(College of Water Resources and Architectural Engineering, Northwest Agricuhural and Forestry University, Yangling 712100, Shaanxi Province, China)

机构地区：[1]西北农林科技大学水利与建筑工程学院,陕西杨凌712100

出　　处：《中国农村水利水电》2016年第11期76-81,共6页China Rural Water and Hydropower

基　　金：国家科技支撑计划(2012BAD08B01);"十二五""863"计划课题项目(2011AA100509);国家自然科学基金项目(51279167)

摘　　要：应用SWAT 2012模拟泾河中上游流域的月径流量,并结合SWAT-CUP中的Parasol、SUFI-2、GLUE和PSO等4种参数优化算法对模型参数校准,并进行对比及适用性分析。结果表明:4种算法均能满足模拟精度要求,月径流模拟纳什系数(NS)和决定系数(R2)均在0.6以上,且GLUE和SUFI-2算法参数优化后模型模拟精度更高。GLUE算法优化后模型在校准期NS和R2分别为0.84和0.76,验证期为0.85和0.78;SUFI-2算法优化后模型在校准期NS和R2分别为0.81和0.67,验证期为0.86和0.70。如果条件允许,GLUE算法更适合较复杂的流域;SUFI-2算法可以缩短优化时间,但精度略低,因此更适合径流变化情况相对简单的流域。Parasol算法和PSO算法模拟历时长,精度较低,不适用于本流域参数优化,但PSO算法对于双峰径流模拟效果较好,较适用于有春汛出现的地区。A hydrological model of the middle and upper reaches of the Jinghe River watershed is established by SWAT 2012 to simulate monthly runoff. Four algorithms, such as Parasol, SUFI-2, GLUE and PSO in SWAT-CUP are selected to optimize model parameters, and comparison and applicative studies are conducted to determine the optimum algorithms. Result indicate that both Nash efficiency coefficient （NS） and the determination coefficient （R2） are all above 0.6, and SUFI-2 and GLUE algorithms are more suitable for watershed parameter optimization with higher simulation accuracy. NS and R2 after GLUE algorithm in calibration and validation periods are 0.84 and 0.76, and 0.85 and 0.78 respectively. For SUFI-2 algorithm, NS and R2 values in calibration and validation periods are 0.81 and 0.86, and 0.86 and 0.70, respectively. GLUE algorithm, if conditions permit, is more proper for the watershed with relatively complex situations. SUFI-2 algo- rithm can shorten optimization time, but it is unsuitable for the runoff situation of the watershed with complex conditions due to low accuracy. PSO and Parasol algorithms are unsuitable for model parameters optimization in the Jinghe River watershed because of low accuracy and time -consuming simulation. However, PSO algorithm is suitable for the area appearing spring flood because it has a good simulation for two-peak flow situations.

关 键 词：优化算法 自动校准 SWAT模型 

分 类 号：TV121[水利工程—水文学及水资源] P334.92[天文地球—水文科学]