SWAT Model Based On Copula And Posteriori Correction And Its Application In Snowmelt Runoff Simulation

In the middle and high latitudes,snowmelt is a typical natural phenomenon due to seasonal changes in temperature.Snowmelt runoff is an important part of snowmelt seasonal runoff and the main supply source of many rivers.Under the background of global warming,th e global snow gradually melts,and the change of snowmelt water is more likely to have a significant impact on the local ecological environment and production and life.Therefore,the simulation of snowmelt runoff has important practical significance for the sustainable utilization and ecological security of water resources in the watershed.This study applied distributed SWAT(Soil and Water Assessment Tool)model,combined with remote sensing technology,acquired land use data,DEM data and Soil data of t he Naoli river basin in Sanjiang plain of Heilongjiang Province based on GIS,established topographic and spatial data,and automatically divided the Naoli river basin into hydrological response units and sub-basins.Combined with the relevant physical variables input into the model,the daily runoff process and variation characteristics in the snowmelt period(March ? May)in the Naoli river basin were simulated.SWAT model has shortcomings in mechanism description,function improvement and application extension.In order to improve the accuracy of model simulation and reveal the spring hydrological law of the Naoli river basin,this study proposes two methods of applying Copula function to deal with the snowmelt runoff of SWAT model:(1)There is a certain error between the simulated runoff and the actual observed value,so it is necessary to use the error correction technology to reduce the simulation error in the simulated runoff process.This paper proposes a method for error correction of Copula functio n.Copula function is used to reveal the correlation between the simulated value and its error,describe the joint probability distribution of the runoff simulation results and their errors,and modify the simulated runoff value during the verification period to make the simulated value more approximate to the real value.Results show that the rate on a regular basis and validation period Nash efficiency coefficient were greater than 0.7,the building of the SWAT model is suitable for the snowmelt runoff si mulation in the study area,and after corrected by Copula function of Nash efficiency coefficient to 0.8 above,and runoff value of revised moving average method of smoothing data,the accuracy will be improved further.Therefore,the correction of Copula function for SWAT simulation of snowmelt runoff has better operability and accuracy.(2)Model parameter determination is an important aspect of hydrological simulation.In order to carry out accurate simulation,the sensitivity of parameters needs to be identified and optimized.The traditional parameter calibration method can determine the required parameter set,but cannot update these parameters over time.In order to solve the above problems,this method proposes a method of time-varying parameters,called adaptive updating,to improve the fitting degree between observed values and simulated values.The joint probability distribution of Copula function was combined with random sampling,new runoff data was added,and the posterior distribution of the two most sensitive parameters was calculated by screening with NS>0.6,so as to update the values of the two most sensitive parameters.The results show that the simulation results reflect the actual runoff in the snowmelt period and simulate the process of snowmelt runoff.Adaptive updating method after the best simulation results,Nash efficiency coefficient were greater than 0.8,and it also improved the application of the model parameters,a random sample of 800 set s of parameter value about 80% of the accord with the condition of Nash efficiency coefficient is greater than 0.6,and update the parameter value can reflect the change of climate and ground conditions.