DEM栅格分辨率和子流域划分对杏子河流域水文模拟的影响

Effects of DEM resolution and watershed subdivision on hydrological simulation in the Xingzihe watershed

以黄土丘陵沟壑区杏子河流域为研究区,采用数字滤波法分割了杏子河招安水文站1958—1974年的基流量,评价了SWAT模型在该流域水文模拟的适用性,并分析了DEM栅格分辨率和子流域划分对SWAT水文模拟的影响。结果表明,SWAT模型适用于该流域年河川径流、地表径流、基流及产沙量的模拟。当DEM栅格分辨率在20—150 m之间时,SWAT能有效地模拟年河川径流、地表径流、基流及产沙量,各水文要素模拟结果的R2和NSE分别在0.93和0.51以上,RSR在0.43以下;而当栅格分辨率大于150 m时,各水文要素的模拟效果存在差异。子流域划分对流域产流模拟影响较小,而对产沙模拟影响较大。当子流域提取阈值在12—100 km2之间时,不同的子流域划分对产沙量几乎没有影响,若超出该阈值范围,模型会低估产沙量。因此,可针对不同的水文要素选择合适的DEM和子流域提取阈值,以提高模拟精度和运行效率。

The Soil and Water Assessment Tool（SWAT） is a watershed-based,semi-distributed hydrologic model that has become one of the most powerful tools for simulating hydrological processes at different spatial scales.The SWAT model utilizes geographic information systems and digital elevation model（DEM） to delineate watersheds and extract the stream network.The simulation accuracy depends on how well the model inputs describe the watershed.High resolution DEM generates more accurate topographic estimations,but such information is costly to obtain,especially in the hilly and gully region of China′s Loess Plateau.In addition,the size of the extracted subwatersheds strongly depends on the pre-selected threshold value of the watershed subdivision.The size and number of subwatersheds can also affect the watershed modeling process and subsequent results;however,there are no accepted guidelines for selecting the threshold value,making the extraction of subwatersheds a subjective process.Therefore,the objective of this study was to investigate the effect of DEM resolution and watershed subdivision on the simulation of hydrological processes at the outlet of the Xingzihe watershed in the hilly and gully region of the Loess Plateau.In our study,the digital filter method was used to separate base flow from the measured streamflow data at the Zhao′an hydrological station from 1958 to 1974.The applicability of the SWAT model in the Xingzihe watershed was assessed on an annual basis.The impacts of DEM resolution and watershed subdivision on the simulation of annual streamflow,surface runoff,base flow,and sediment yield were evaluated by running the model with ten DEMs（20,30,60,90,150,300,500,1000,1500 and 2000 m） and seven threshold values（10,12,15,30,50,100 and 200 km2） for the watershed subdivision.The results showed that the SWAT model accurately simulated annual streamflow,surface runoff,base flow,and sediment yield in the Xingzihe watershed.DEMs with fine resolution（i.e.,20 to 150 m） were required to simultaneously simulate annual streamflow,surface runoff,base flow and sediment yield with a high degree of accuracy（R2〉0.93,NSE〉0.51,RSR〈0.43）.For DEM resolutions greater than 150 m,the accuracy of the simulation varied among the hydrologic elements.The threshold values of the watershed subdivision had little effect on streamflow,surface runoff and base flow.However,the threshold values had a large effect on sediment yield.Watershed subdivision threshold values of 12 to 100 km2 were required to adequately predict sediment yield.Threshold values less than 12 km2 or greater than 100 km2 led to the underestimation of sediment yield.These findings indicated that choices regarding the DEM resolution and threshold value of the watershed subdivision should be based on the hydrological element of interest.The SWAT model was limited in its ability to simultaneously simulate all of the hydrologic elements in the Xingzihe watershed at short time scales（i.e.,daily and monthly）;therefore,this paper did not discuss the effects of DEM resolution and watershed subdivision threshold values under different time scales.Additional research is needed to ascertain if the results will change when running the model at different time scales.