【作者】 贺维；

【导师】 张建军；

【作者基本信息】 北京林业大学 ， 地图学与地理信息系统， 2007， 硕士

【摘要】 黄土高原地区生态环境脆弱，水土流失非常严重，而这些问题都和水有着直接联系。因此，选择建立适合西部黄土区应用的水文模型，对于了解该地区水文状况、预测水文过程、水资源的合理利用，以及解决相关生态、经济问题具有重要意义。本文依托国家973研究项目“西部典型区域森林植被对农业生态环境的调控机理”中“流域生态水文过程的尺度效应与转换方法”课题，以山西吉县蔡家川流域为研究对象，应用分布式水文模型SWAT对该区降雨—径流进行模拟研究，重点探讨流域下垫面变化和降雨径流的响应关系。论文应用集于Arcview中的SWAT分布式水文模型，通过处理野外观测数据、在参阅相关文献的基础上，确定了研究地区应用SWAT模型所需的初始参数，利用GIS进行模型运行时空间数据库、属性数据库的建立和管理，并生成运行参数。在此基础上利用SWAT模型模拟了山西吉县蔡家川流域3年部分月份的逐日径流量，并对模拟结果和实测结果进行了比较，探讨了不同土地利用条件和不同雨量条件下的径流变化。通过参数敏感性分析得出：影响径流过程的主要参数为径流曲线数、土壤有效持水率、土壤蒸发补偿系数。通过模型有效性分析得出，校正期相关系数为0.843，效率系数为0.795，验证期相关系数为0.802，效率系数为0.765。通过精度分析得出：模型在雨季的模拟结果较好，尤其是月降雨在15mm以上时，月模拟相关系数都在0.8以上，而在旱季模拟精度较低，在日径流深小于0.01mm时，会出现模拟结果为零的情况。通过建立情景模拟，对不同降水、不同森林植被覆盖率和土地利用类型变化下水文响应进行分析，得出当流域从全林流域变化到无林流域时，产水量从37.49万m~3增加到93.51万m~3，增加了149.4％，尤其在雨季，森林植被可以明显的调节产流量；农地产流量是荒草地产流量的1.56倍。研究得出，SWAT模型适用于黄土区，本文的研究结果，对于指导该地区植被建设，建立适用于黄土区的分布式水文模型有一定的实践指导意义。更多还原

【Abstract】 Soil erosion is very serious on the Loess Plateau, and these issues are directly in connection with water. Therefore, it is great significant to opt for the suitable hydrological model for understanding the regional hydrology mechanism, predictiing and reasonable using of water resources. It is also meaningful for resolving the relevant ecological and economic problems. Under the support of National Basic Research Program (973), this paper applied the distributed hydrological model SWAT on the Caijiachuan watershed of Jixian County, Shanxi Province, simulated the rainfall and runoff, and anylase the relationship between watershed landuse change and hydrological process.Distributed hydrological model SWAT embedded in Arcview software is used for simulating daily runoff from March to November in Caijiachuan watershed. Through a lot of data processing and literature review, the initial parameters were determined. With the GIS software, the spatial and attribute datasets for running SWAT were built, the relationship between watershed landuse change and hydrological process were anylased. By model parameter sensitivity analysis, the main parameters influencing runoff are SCS Runoff Curve Number, available water capacity of the soil layer SOL_AWC, soil evaporation compensation factor ESCO. The model was calibrated by adjusting hydrological parameters. In calibration period, relative coefficient is 0.843, Nash-Suttcliffe coefficient is 0.795, in validation period relative coefficient is 0.802, Nash-Suttcliffe coefficient is 0.765.According to the analysis results, the hydrological process was well simulated in the rain seasons, compared to the dry seasons, especially above 15mm precipitation, with the relative coefficient above 0.8. The simulation values will be zero while daily runoff is smaller than 0.01mm. In conclusion, the SWAT model is applicable on the Loess Plateau. It shows the spatial distribution of precipitation in three years by subbasins division. Through the creation of simulation scenarios for different precipitation, the response different forest vegetation cover and landuse type changes in the hydrological scenario were analysed. It concludes that the runoff increase with vegetation cover reducing. Water yield increased to 149.4% from 374,900 m~3 when the forest cover changes from entire valley to non-forested watershed. In the rainy season, the forest vegetation can significantly influence runoff. It concludes water production in farmland is 1.56 times as much as that in pasture. The results can be reference to landuse management and establishment of the Loess Distributed hydrological model.更多还原