Runoff Simulation And Parameters Uncertainty Analysis Based On SWAT Model In Le'an River Xiangtun Catchment

With the economic development in our country,the urbanization area is also larger and larger,which greatly changes the underlying surface of the land and thus undermines the natural hydrological cycle,resulting in the continual increase of disasters in our country.However,due to the complexity of hydrological processes,the errors of hydrological data and the imperfectness of model structure,model simulations bring great uncertainty to the hydrological model parameter calibration and hydrological forecast,thus affecting the Accuracy of simulation results.Therefore,it is of great practical significance and scientific value to deeply and systematically carry out the research on the uncertainty analysis of hydrological simulation and model parameters in distributed watershed,and to prevent and control floods and droughts and control and distribute water resources.Based on this,this paper takes the SWAT model which is widely used as the research object,taking the Xiangtun River Basin as an example,studies the uncertainty of the parameters in the runoff simulation process for the SWAT hydrological model.The SWAT model database was constructed by collecting land use,soil and meteorological data.The daily runoff of the study area was simulated.The SUFI-2 method was used to analyze the uncertainty of the parameters.The iterative method was used to process the parameter interval.The variance of the prior interval in the posterior interval was compared and analyzed.The influence of the changed parameter interval on the simulation effect was analyzed.According to the "same effect"-2010 runoff to verify the use of interval probability peak method to do the model simulation,select the validation period 5-6 month Fengshui month as the object of observation,compared with the optimal parameter group and interval probability peak simulation value and the actual runoff relationship.The main conclusions are as follows:(1)Through the sensitivity analysis of the parameters and the physical characteristics of the study area,the sensitive parameters affecting the runoff simulation results in the study area were screened out.This paper selects the SCS runoff curve coefficient(r_CN2),Manning River(v_CH_N2),then the coefficient of groundwater evaporation coefficient(v_GW_REVAP),soil evaporation compensation coefficient(v_ESCO),soil available water content(r_SOL_AWC),the runoff coefficient of shallow groundwater(v_GWQMN),the main river hydraulic conductivity(v_CH_K2),shallow groundwater but the evaporation coefficient(v_REVAPMN),He Ji flow coefficient(v_ALPHA_BF)of nine parameters.(2)Through the simulation,it is found that the parameters v_CH_N2,v_CH_K2 and v_ALPHA_BF have a great influence on the simulation uncertainty.The distribution of thethree parameters is analyzed.The proposed interval after model simulation is defined as the prior interval and the Bootstrap edge probability distribution transformed interval The results show that R2 = 0.85,efficiency coefficient Ens = 0.83,R2 = 0.90 in the best parameter set of posterior interval model,efficiency The coefficient Ens = 0.85.It can be seen,the posterior parameters of the simulation rate set a better outcome.The comparison between the simulation results of the bestparameter of the posteriori interval between the probabilistic interval posteriori probability peak and the validation period showed that the correlation coefficient of the posteriori probability peakparameter validation in the validation period was R2 = 0.83,the Nash efficiency coefficient Ens = 0.80,the posteriori interval validation Period of R2 = 0.80,Nash efficiency coefficient Ens = 0.78,indicating that the effect of the peak by the probabilistic method is better.(3)Using 500 sets of parameters to simulate the runoff from 2008 to 2010 one by one,500 simulated runoff values can be obtained every day,and the interval distribution analysis of daily simulated runoff values is performed.The interval probability peak is used as the current day runoff simulation Value,calculate the Nash efficiency coefficient and the correlation coefficient,the comparison found that the interval probability peak simulation effect is better.Then,the measured monthly runoff from May to June in each year was compared,and the relationship between the optimal parameter set and interval probability peak simulated value and actual runoff was compared.The results show that the simulation results using the interval probability peak method are better than the optimal parameter group method,and the posterior range probability peak method is more effective than the prior interval probability peak simulation method.(4)Uncertainty analysis using SUFI-2 method.Compared with the runoff simulation of the a priori interval and the a-posterior interval,we found that the R factor of the posterior interval is smaller when the P factor is close,which means the posterior interval uncertainty is smaller.