基于SWAT模型的北洛河典型子流域降雨径流模拟研究

发布时间：2018-08-06 18:06

【摘要】：以黄土高原北洛河流域为研究区域,验证了PERSIANN降水产品在北洛河交口河水文站以上流域的精度,评价了SWAT模型在北洛河两个子流域的适用性,分析了不同DEM数据源、重采样方法、参数率定方法、研究时间尺度及DEM分辨率等因素对葫芦河子流域降雨径流模拟结果的影响。主要得到以下结论:(1)总体上,降水产品低估了降水量的值,不能很好地估计极端降水量。对于年平均降水量和汛期极端降水量,直接提取的方法相关系数较线性插值法的值大。降水产品在夏秋季低估降水,在冬春季高估了降水。对于季平均降水量,夏季的精度最高,春季次之。两种提取方法均存在误差,直接提取的方法精度较双线性插值法的精度高。对于PERSIANN降水产品,多年平均降水量随着高程的增大而减小。(2)SWAT模型在北洛河两个子流域的模拟结果满足精度要求。基于ASTER GDEM30m数据,针对葫芦河子流域和北洛河吴旗水文站以上子流域两个研究区域,分别建立月尺度的SWAT模型进行降雨径流模拟研究,模拟结果可满足精度要求,总体满足水量平衡。葫芦河子流域的模拟结果较吴旗水文站子流域的模拟结果精度高。(3)SWAT模型对于降雨径流的模拟结果差异有不同来源。基于ASTER GDEM和STRM DEM两种不同来源的DEM数据,使用最邻近插值、双线性插值、三次卷积插值三种重采样方法,建立月和日两种不同时间尺度的SWAT模型,分别用SUFI-2和PSO两种方法率定参数,引入多目标优选模型评价模拟结果。结果表明,30m ASTER GDEM数据提取流域地形信息精度最高;重采样后的DEM数据,最邻近插值法得到的DEM数据精度最高。30m ASTER GDEM数据建立的模型,模拟结果优于90m STRM DEM数据建立的模型;重采样的DEM数据建立的模型,最邻近插值法模型模拟的结果最优;最邻近插值法DEM建立的模型,使用SUFI-2方法的率定结果精度最高;月尺度的模型模拟结果比日尺度的模拟结果更优。(4)将ASTER GDEM 30m数据重采样为9种不同分辨率的DEM数据,分别建立月尺度的葫芦河子流域SWAT模型,对比不同分辨率DEM数据提取的流域水文地形因子差异并模拟径流。DEM分辨率对坡度、面积提取与流域划分影响相对较大,对高程提取影响相对较小。当DEM分辨率高于150m时,模型模拟的结果比较满意且合理,30m DEM数据建立的模型,需经过多次迭代求得最优参数组合;DEM分辨率为120-150m模型的模拟结果较优。分辨率低于600m DEM建立的模型,划分子流域的数量减少,模拟结果存在一定偏差。
[Abstract]:Taking the North Luohe River Basin on the Loess Plateau as the research area, the accuracy of PERSIANN precipitation products in the basin above the Beiluohe River hydrologic station is verified, the applicability of the SWAT model in the two subbasins of the North Luohe River is evaluated, and the different DEM data sources are analyzed. The effects of time scale and DEM resolution on the simulation results of rainfall runoff in Hulu River Basin were studied by resampling method and parameter rate determination method. The main conclusions are as follows: (1) in general, precipitation products underestimate the value of precipitation and can not estimate extreme precipitation well. For the average annual precipitation and extreme precipitation in flood season, the correlation coefficient of the direct extraction method is higher than that of linear interpolation method. Precipitation products underestimate precipitation in summer and autumn and overestimate precipitation in winter and spring. For seasonal average precipitation, the accuracy of summer is the highest, followed by spring. There are errors in both methods, and the precision of the direct extraction method is higher than that of the bilinear interpolation method. For the PERSIANN precipitation products, the annual average precipitation decreases with the increase of elevation. (2) the simulation results of the SWAT model in the two subbasins of the North Luohe River meet the precision requirements. Based on ASTER GDEM30m data, a monthly scale SWAT model was established to simulate rainfall runoff in two study areas of the Hulu River Basin and the North Luohe Wuqi Hydrologic Station. The simulation results can meet the precision requirements. Overall water balance is satisfied. The simulation results of Huluhezi watershed are more accurate than those of Wuqi hydrologic station. (3) the SWAT model has different sources for the simulation results of rainfall runoff. Based on the DEM data from two different sources of ASTER GDEM and STRM DEM, using three resampling methods such as nearest neighbor interpolation, bilinear interpolation and cubic convolution interpolation, the SWAT model of two different time scales of month and day is established. Two methods, SUFI-2 and PSO, are used to determine the parameters, and the multi-objective optimal selection model is introduced to evaluate the simulation results. The results show that the precision of extracting watershed topographic information from 30m ASTER GDEM data is the highest, and the model of DEM data obtained by nearest neighbor interpolation method is the best, which is better than the model built by 90m STRM DEM data after resampling DEM data. The model established by resampling DEM data has the best result of simulation by nearest neighbor interpolation method, and the model built by nearest neighbor interpolation method DEM has the highest precision of rate determination by using SUFI-2 method. The simulation results of the monthly scale model are better than that of the daily scale model. (4) the ASTER GDEM 30m data are resampled to 9 different resolution DEM data, and the SWAT model of the Hulu River Basin is established respectively on the monthly scale. Comparing the difference of hydro-topographic factors extracted from DEM data with different resolutions and simulating the effect of runoff. Dem resolution on slope, area extraction and watershed division, the effect on elevation extraction is relatively small. When the resolution of DEM is higher than 150m, the simulation results of the model are satisfactory and the model established by 30m DEM data is reasonable. The simulation results of the optimal parameter combination Dem model with the resolution of 120-150m need to be obtained through several iterations. When the resolution of the model is less than 600m DEM, the number of subbasins is reduced, and the simulation results are deviated to a certain extent.
【学位授予单位】：西北农林科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P333.1

【相似文献】