Zoning Inversion Of Soil Heavy Metal Content Based On Vegetation Index

Environmental pollution is a major problem faced by mankind today.In addition to air pollution and water pollution,the problem of heavy metal pollution in the soil has also received much attention.How to effectively monitor soil heavy metal pollution is the focus of people's exploration.Soil composition analysis and material content determination,hyperspectral remote sensing,multi-spectral remote sensing and other techniques have been applied to soil heavy metal pollution monitoring.Inversion of soil heavy metal content using measured sample soil spectral data,the inversion accuracy is greatly affected by the number and distribution of sampling points;the method of inversion of soil heavy metal content by hyperspectral remote sensing image,although the spectral resolution is higher,Data acquisition is difficult and the processing method is complicated.The multi-spectral remote sensing image acquisition is relatively easy,the data redundancy is small,and the data processing is convenient.When the heavy metal content of the soil is inverted in large areas and large scales,the purpose of monitoring the heavy metal content of the soil can be achieved at a macroscopic level.In the inversion of soil heavy metal content using image spectral data,if the entire study area is uniformly modeled,regardless of vegetation cover or land type division in the study area,the overall accuracy may be low.Therefore,when multi-spectral image data is used to invert the soil heavy metal content,the method of constructing the inversion model by partition can be used to monitor the heavy metal content of the soil in the whole region and improve the inversion precision.This paper takes the western part of Tianfu New District in Chengdu,Sichuan Province as an example.According to the vegetation coverage in the area,the area is divided into three categories: vegetation area,bare area and non-inversion area.Using the vegetation index and single-band reflectivity extracted from WorldView-3 multi-spectral image data,three regression methods,principal component regression,multiple stepwise regression and partial least squares regression,were used to establish a regression model with the measured heavy metal content of soil samples.In the modeling,two-thirds of the measured soil heavy metal content data were used to establish the inversion model,and the remaining one-third of the data was used to verify the accuracy of the inversion model.In the vegetation area,three inversion methods were used to establish inversion models of soil heavy metal As,Co,Cr,Mo and Pb;in the bare area,three regressions were used.Methods The inversion model of heavy metals such as As,Co,Mn,Ni and V was established.Finally,through the two indexes of fitness and precision,six optimal inversion models were selected from 30 models established in the vegetation area and the bare area.The heavy metal contents of the soils of As,Co,Cr and Mo in the vegetation area were obtained.The best model for inversion and the best model for inversion of heavy metal content in As and Co soils in bare areas.The main results of this research are as follows:(1)In the correlation analysis between soil heavy metal content,vegetation index and single-band reflectivity,the correlation between soil heavy metal content and vegetation index in vegetation area is stronger than single-band reflectivity,while soil heavy metal content and single-band reflectivity in bare area.The correlation is stronger than the vegetation index,so the method of model inversion of the study area partition can improve the overall accuracy to some extent.(2)In this study area,the contents of As,Co,Cr and Mo in soil heavy metals have strong correlation with DVI,GDVI,GNDVI,GRVI,NDVI in vegetation index,and WorldView-3 single-band reflectivity.The Coastal Blue,Blue,Green,and Yellow bands have a good correlation and have a low correlation with the short-wave infrared band.(3)A total of 30 regression equations were established in the vegetation area and the bare area using principal component regression,multiple stepwise regression,and partial least squares regression.The model was evaluated by model evaluation indicators and one-third of the measured soil heavy metal content data that were not involved in the modeling.In the vegetation area,the inversion model of heavy metal As and Mo is established by multivariate stepwise regression method,which is better than the model established by principal component regression and partial least squares regression;the heavy metal Co and Cr established by partial least squares regression method The content inversion model is better than the model established by the multiple stepwise regression and principal component regression methods.In the bare area,the inversion model of heavy metals As and Co established by multivariate stepwise regression is better than the model established by principal component regression and partial least squares regression.