Study On Optimization Of Groundwater Monitoring Network Based On Identification Of Pollution Source Leak Rate

Groundwater is an important part of water resources.With the aggravation of human activities on the environment,groundwater pollution problem has attracted more and more attention.Reasonable layout of groundwater pollution monitoring well is an effective means to understand the groundwater pollution problem.In this paper,a simulation-optimization method is used to carry out a monitoring network optimization study with a chromium-contaminated site as an example,and the monitoring network is optimized based on the prediction of the pollution halo area by the numerical simulation of solute transport.However,because the chromium slag is buried underground,the leakage rate of pollutants discharged to the aquifer is difficult to determine quantitatively.Therefore,before optimizing the monitoring network,it is first necessary to reverse and identify the pollution source leakage rate with the existing pollutant concentration observation data.Firstly,we establish a numerical simulation model based on the geological and hydrogeological conditions of the case area;then determine the possible value range of the pollution source leakage rate through site investigation and literature review,adopt Latin Hypercube Sampling(LHS)method to sample within the feasible region of the leakage rate,substitute the sampling results into the simulation model,and run the simulation model to obtain the input-output training set of the substitute model;apply the training set to establish a substitute model of the simulation model for the deep residual dense convolutional network(DRDCN)and the multi-residual deep convolutional network(MRDCN)proposed in this article,bring the 4000 sets of possible leakage rates identified by the inversion into the substitue model,and perform uncertainty analysis on the distribution range of the pollution halo in the study area,then obtain the mean and variance distribution of the pollutant concentration,the simulation-optimization method is used to lay out the points of the monitoring wells,and the groundwater monitoring network optimization plan of this article is proposed.Through the study of this paper,the following conclusions are obtained:(1)The substitute model established by the MRDCN network proposed in this paper has high accuracy,and the additional residual connection structure can significantly improve the prediction accuracy of the substitute model,and can establish a high degree of nonlinearity between the pollution source leakage rate and the spatial distribution of pollutant concentration.(2)The MRDCN-ILUES inversion framework proposed in this paper can effectively identify the leakage rate of pollution sources.The error between the inversion identification concentration and the observation concentration at the three monitoring wells is about 1%,indicating that the inversion framework can be a pollution source for other actual sites.The inversion identification of leakage rate provides a feasible method.(3)The monitoring network optimization scheme proposed in this paper based on the uncertainty of the spatial distribution of pollutants concentration can monitor the center,edge of the pollution halo and the points sensitive to the pollution source,which can make it possible to provide more effective information for pollution remediation and treatment under a limited number of monitoring wells.