Simulation And Uncertainty Analysis Of Non-point Source Pollution In Guaugli River Watershed

In recent years,with the point source pollution problem is effectively controlled,non-point source pollution has became a major source of water environmental degradation.Compared with point source pollution,non-point source pollution has the characteristics of randomness,hysteresis,ambiguity and difficulty in research control.Using mechanism model to quantitative analysis and evaluation of non-point source pollution load is an important means of non-point source pollution control and research.However,since the non-point source pollution model is a generalization of the natural system,its model structure,input data and parameters often have many uncertainties in application.These will inevitably affect the accuracy of the simulation results.Among them,the model parameters are the main source of uncertainty of simulation results.Therefore,it is necessary to understand the influence of uncertainty of the parameters in non-point source pollution simulation on the model accuracy and the simulation results,objectively evaluate the accuracy of the model simulation results,which can provide guidance for managers to develop effective pollution prevention measures.This paper takes the Guangli River Basin as the study area.Collecting spatial data and attribute data from the study area to construct a SWAT model for non-point source pollution simulation.Analyzing the pollution characteristics of non-point source pollution ammonia nitrogen and total phosphorus in the basin.Identifying the key source areas of pollution,Puting forward pollution control measures and carrying out scenario simulation to predict the reduction effect of pollution.Finally,training the wavelet neural network as an alternative model,Using the Monte Carlo method to analyze the uncertainty of the simulation results of the SWAT model in the study area.The main research results are as follows:(1)Construction and calibration of SWAT model for watershed non-point source pollution simulation.Useing the methods of consulting materials and on-site research,This study collected data on water system distribution,administrative division,digital elevation,land use,soil distribution,agricultural production,pollution load,etc.in the Guangli River Basin.Establishing a spatial database and attribute database in the study area to construct a SWAT model for non-point source pollution.Using the data of runoff,ammonia nitrogen and total phosphorus in the Guangli River from 2015 to 2016 to calibrating and verificating the model.The results show that the R2 coefficient and ENs coefficient of runoff,ammonia nitrogen and total phosphorus in the calibration and verification period are all above 0.65,which meets the simulation requirements of the model.The SWAT model can be betterly applied to the non-point source pollution simulation in the Guangli River Basin.(2)Analysis of non-point source pollution characteristics and identification of key source areas in river basins.The annual average load of ammonia nitrogen in the Guangli River Basin is 13.88×104 kg,and the average annual output of total phosphorus is 18.25×103 kg,which means the ammonia nitrogen pollution is worse than total phosphorus.The changes of ammonia nitrogen and total phosphorus showed the characteristics of large yield during flood season and small yield during non-flood period.The trend from pollution to gradual increase from upstream to downstream.The key source areas of ammonia nitrogen and total phosphorus pollution are concentrated in the downstream and densely populated areas.(3)Scenario simulation and effect evaluation of watershed non-point source pollution management measures.Setting three kinds of pollution management measures:fertilizer reduction(15%reduction),grass planting channel and vegetation filter belt,and scenario simulation in SWAT model.The results showed that the average reduction rate of ammonia nitrogen pollution in the basin is 6.87%,29.45%and 39.91%,and the average reduction rate of total phosphorus pollution is 5.27%,22.81%and 36.50%.The most significant measure is the vegetation filter belt,followed by the grass planting channel and fertilizer reduction(4)The establishment of a non-point source pollution alternative model for watersheds.Monte Carlo method is used to analyze the uncertainty of SWAT model parameters.In order to solve the problem of large amount of computation when using Monte Carlo method for uncertainty analysis,wavelet neural network is trained as an alternative model for non-point source pollution simulation SWAT model.The results show that the wavelet neural network model can not only replace the SWAT model in the study area,but also can greatly reduce the calculation time of Monte Carlo simulation.(5)Uncertainty Analysis of Non-point Source Pollution Simulation in Watershed.The Chebyshev inequality and the coefficient of variation Cv are used to response the output uncertainty of the model.The results show that confidence intervals of ammonia nitrogen and total phosphorus output in the study area are 4.02×104～25.85×104,9.01 ×103～28.38×103,and the Cv values of ammonia nitrogen and total phosphorus output are 23.12%and 16.38%,which indicates that the output of ammonia nitrogen is more discrete and more uncertain.