Distributed simulation for hydrological process in Plain River network region using SWAT model

Abstract: Distributed hydrological modeling plays an important role in water resource management and regional non-point source pollution assessment. The Soil and Water Assessment Tool (SWAT) is a popular modeling tool for understanding regional hydrological processes. However, the general approach based on the SWAT model was only applicable to the mountain and hilly dominated area. There is no effective way to modeling the hydrologic process in plain river network regions, which is characterized by large flat areas, consisted of many lakes and artificially hydrological polders, and intersected stream networks, etc. The existing methods cannot effectively extract the channels in flat and pit areas, parallel channels or discontinuous rivers and the definition error of the catchment areas. To overcome these problems, we developed a novel method for modeling the distributed spatial discretization of the plain river network area based on the SWAT model. There are three key techniques are discussed: making the rings and crossed rivers to dendritic stream networks by cutting the river ways shortly, restoring the distribution of water between reaches by transferring water from one reach to another one on the basis of flow rate of each reach and simulating the exchange of water inside and outside of the polders according to the scheduled rules of the polder areas by adding a 'virtual reservoirs' within the SWAT model. In this paper, the typical plain river network region located in western Taihu watershed was chosen as the study area, and a large number of basic geographic data such as topography, soil, climate and land use were collected and parameterized. The modeling procedures were used to simulate the monthly runoff of the area of western Taihu Lake from the year of 2008 to 2010, and the applicability of the method to the plain river network region was also verified. The simulated results matched mostly well to the observed data of Rongdengqiao, Hujiawei, and Yixing hydrological stations. The calculated Nash-Sutcliffe efficiency coefficient and correlation coefficient of three hydrological stations were 0.84, 0.80, 0.67 and 0.94, 0.95, 0.93, respectively. It indicated that our developed framework for the SWAT model was practical and capable of representing the hydrological processes in the plain river network regions.