基于 SWAT 模型的北山水库流域地表径流模拟

为评价北山水库流域地表水资源,选用SWAT软件构建了该流域的分布式水文模型。基于2016-2018年实测水文数据以及高程、土地利用类型、土壤类型等数据,完成了对北山水库流域的SWAT模型构建,利用SWATCUP软件对参数进行率定及敏感性分析,选取决定系数(R~2)、纳什效率系数(NSE)、平均相对误差绝对值(MARE)和均方根误差(RMSE)作为模型评价指标。结果表明:北山水库月蓄水量模拟值与实测值吻合良好,模型率定期和验证期的R~2均达到0.89,NSE分别达到0.88与0.85,MARE分别为5.04%与4.23%,RMSE分别为1.15×10~6 m~3与0.90×10~6 m~3,由此表明该模型能近似反映研究区的径流变化特征,展示了SWAT模型在该流域的适用性。此外,基于验证的水文模型对主要参数进行敏感性分析,表明影响该区径流模拟最为显著的两个参数是浅层地下水回归流阈值和径流曲线数。

Surface runoff simulation based on SWAT model in Beishan Reservoir watershed

SWAT soft war e was used to emplo y the distributed hy dr olog ical model t o evaluate the surface water resources in the Beishan Reser voir w atershed. Based on measur ed hydro lo gical data fr om 2016 to 2018 prov ided by the lo ca l wat ershed manag e2 ment institutio n, and the elevat ion, land use, and soil type data, the SWAT and SWAT2CUP so ftwar e w ere used to calibr ate and analyze the sensitiv ity of the par ameters in the Beishan Reserv oir w atershed. The coefficient of determination ( R2 ) , Nash2Sut2 cliffe efficiency ( NSE) , mean absolute relative erro r ( M ARE) and ro ot2mean2square erro r ( RM SE) w as selected as the model ev aluatio n index es. The results indicated that the simulated value of Beishan Reser vo irc s mo nthly water sto rag e show ed go od agr eement w ith the measured value. The mo del evaluation indexes R2 , NSE, M ARE and RMSE, w ere 01 89, 01 88, 51 04% and 11 15 @ 106m3 in calibration period, and 01 89, 01 85, 41 23% and 01 90 @ 106 m3 during validation period, r espectiv ely. This sug ges2 ted that the calibrated model can appr ox imately r eflect t he char act eristics of runo ff in the study ar ea, indicating the displayed best applicability of SWAT model in the w atershed. Mor eover, the sensitivity analysis of the main parameters based o n the vali2 dated hy dr olog ical mo del indicated that GWQMN ( threshold of shallow g r oundwater r egr ession flow) and CN2 ( number of SCS runo ff cur ves) w ere the most impo rtant par ameters show ed sig nificant effect o n the model r esults.