Comparative Study Of SWAT Hydrological Parameters Optimization Method Based On Sensitivity Analysis

SWAT (Soil and Water Assessment Tool) model with powerful function of watercycle simulation has been applied widely throughout the world. It can dividerainwater catchments automatically and form water source area boundary according tospatial data of terrain built by GIS platform; simulate hydrological process combiningwith various physical distributed input files in study area. This distributedhydrological model not only has small dependence on the observed data in study area,but also have powerful operability, especially can complete hydrological processesanalysis under environmental change. Futher more, the efficiency and quality of watercatchment area partitioning can be guaranteed.Changchun as the typical urban of water shortage in northern of China, the watersupply security work for Xinlicheng reservoir which is closest to Changchun isparticularly important. This article built monthly time-scale runoff SWAT model, andexamined the applicability and reliability of the model, provided scientific basis forwater conservation of Xinlicheng reservoir.This article chosed the Xinlicheng reservoir as the research area, researched thestructure of the model and running mechanism, and analyzed three stages which isrelated to runoff simulation in the model, hydrologic cycle land runoff stage,convergence stage and river confluence calculation stage. Based on spatial data anddaily hydrometeorological data of the study area, GIS was applied to process thespatial data such as DEM, land use/plant cover map and soil type map, and Matlab,pcpSWAT, dew02software were applied to create the soil database andmeteorological database of the study area. And on the basis, run the SWAT model,extracted water system of river basin and determined the flow boundary of the basin.Through the limitation on the area threshold was6600ha, this study area was dividedinto16sub-basins. By loading the land use/cover plant data, soil data andmeteorological data, set the land use and soil type threshold value were5%and10% respectively, the study area was subdivided into93HRUs (hydrological responseunits). Successfully established the SWAT basic data model based Xinlichengreservoir catchment area.The SWAT model was used to simulate the runoff of the study area by monthlytime step. Two different methods were used respectively to conduct sensitivityanalysis and model parameters calibration. The2004-2008observed runoff data ofYitong hydrological station were used to correct model parameters and the2009-2010observed runoff data were used to verify this model. In the end, the modelapplicability was evaluated respectively according to the results of two kinds ofmethod. Results showed that using the SWAT automatic correction module inchecking and verification stage, the monthly runoff coefficients of determination were0.86and0.99respectively; The model efficiency coefficients Ens were0.80and0.97respectively; The relative error Re were20.0%and19.3%respectively; Using SUFI-2algorithm in checking and verification stage, the monthly runoff coefficients ofdetermination were0.92and0.98respectively; The model efficiency coefficients Enswere0.91and0.96respectively; The relative error Re were6.0%and10.3%respectively. The indicators of the two methods for evaluation of model applicabilitymeet the requirement of research, suggested the SWAT model can be applied in thisbasin. Also compared the two methods, no matter from the output quality, the degreeof operating convenience and the model operation efficiency SUFI-2algorithm wasmuch better than automatic correction module.