Spatial Prediction And Location Distribution Of Site Pollution Based On Three-dimensional Interpolation Simulation

At present,there is a large number of polluted sites in our country,and the pollution situation is complicated.It is very expensive to conduct site surveys and contaminated soil remediation work.Site survey sampling is an important means to obtain the spatial distribution of site pollution.The spatial distribution information of site pollution was obtained by interpolation simulation.Affected by the highly biased data of the site and the interpolation limitations of conventional interpolation methods,the application of 3D interpolation simulation can better reflect the spatial distribution of pollutants.important meaning.At the same time,scientific and reasonable sampling density and reasonable sampling method have a great influence on the prediction results of site pollution.Therefore,it is necessary to propose a set of methods suitable for site survey and sampling from the aspects of three-dimensional interpolation method,sampling density and supplementary sampling method..In this study,relying on the real high bias data characteristics of heavy metal Sb in the study area,a suitable three-dimensional interpolation method was obtained by comparing the interpolation accuracy of three three-dimensional interpolation methods,and compared with the two-dimensional interpolation method.Lattice sampling density analysis,based on three-dimensional interpolation results superimposed to extract the intersection area of local high variability area and pollution probability uncertainty area as pollution comprehensive uncertainty area,to guide supplementary sampling,through reasonable and effective point distribution method and appropriate interpolation method The prediction results can identify a more accurate pollution range under the condition of less number of points.The main conclusions of this study are as follows:(1)There are real high-biased data in the study area,and the Sb pollution data does not conform to the normal distribution.After the data is converted to BOX-COX data,the skewness,kurtosis and coefficient of variation are significantly reduced,and the K-S test P The values are all greater than 0.05.After the transformation,the stratified data conforms to the normal distribution.It can be judged that the BOX-COX data transformation can effectively reduce the data bias and make the data meet the requirements of the normal distribution of geostatistical interpolation.Therefore,it is necessary to analyze and evaluate the structure characteristics of the site data before performing interpolation analysis on the site data,and perform corresponding data conversion as needed.(2)Perform three-dimensional interpolation analysis on the Sb data in the study area,respectively apply ordinary kriging interpolation(OK),inverse distance weighting(IDW),and radial basis function(RBF)for interpolation simulation,and compare with two-dimensional interpolation.Interpolation is compared,and the cross-validation accuracy of the interpolation results of each interpolation method is verified.The results show that whether it is two-dimensional or three-dimensional interpolation,the crossvalidation accuracy of ordinary kriging interpolation(OK)is compared with inverse distance weight interpolation(IDW)and radial.The basis function method(RBF)is higher,and the three-dimensional inverse distance weighting method with the lowest three-dimensional interpolation accuracy is higher than the two-dimensional ordinary kriging interpolation method with the highest two-dimensional interpolation accuracy.At the same time,by verifying and comparing the sample data of 102 validation sets to compare the accuracy of the pollution point judgment results of the two-dimensional and three-dimensional interpolation results,it is found that whether it is two-dimensional or three-dimensional,the ordinary kriging interpolation method is the most able to identify Accurate interpolation method for pollution points,and the three-dimensional interpolation method is more accurate than the two-dimensional interpolation method for judging results of pollution points.Considering the cross-validation accuracy and the accuracy of the pollution point judgment results,the three-dimensional ordinary kriging interpolation method is the most suitable interpolation method for the site in the study area.(3)Using grid sampling to obtain sampling data sets under different grid densities,by analyzing the data structure characteristics such as coefficient of variation,skewness,and kurtosis of each grid data set,and verifying and comparing through threedimensional ordinary kriging interpolation The cross-validation accuracy and pollution range prediction accuracy of each grid dataset,comprehensive analysis shows that 40 m ×40 m is the minimum grid sampling spacing to be achieved for grid sampling;When it is equal to 1m,the collected sample data has better representativeness.(4)Under the grid sampling interval of 40 m × 40 m,the pollution uncertainty area of the polluted site is obtained by three-dimensional interpolation to guide the corresponding supplementary sampling,and a more accurate pollution prediction range can be obtained with fewer sampling points.In this study,the influence of abnormally high values was reduced by introducing three-dimensional indicator kriging interpolation,and the pollution comprehensive uncertainty area was extracted through the threedimensional indicator kriging interpolation results.Feasibility and advantages of sampling methods.In this study,a suitable interpolation method was obtained based on the comparison of the accuracy of the three-dimensional interpolation method,and a set of systematic and feasible supplementary sampling point distribution method was proposed,which is expected to obtain a better prediction effect with fewer points,and has a good effect on the investigation and restoration of polluted sites.Important theoretical and practical reference significance.