Inversion Of Heavy Metal Pb Content And Study Of Optimum Spectral Sampling Interval Of Farmland Soil In A Certain Area

Soil is the natural resources for human survival.With the development of modern economy,the destruction of soil is becoming more and more serious.The heavy metal pollution has become a major concern of human beings.The traditional heavy metal monitoring method is accurate,but it needs to carry on the destructive sampling and the cost is huge,the rapid development of remote sensing technology provides a new method for heavy metal monitoring.In this paper,using the study area of image multi spectral and measured hyper spectral of farmland soil as the original data,in order to establish the prediction model for transformed spectrum value and the heavy metal Pb content,the model of the soil spectrum is established by the first derivative reflectance spectra,the logarithm and the reciprocal,respectively,compared to the accuracy of the model.The measured spectra were resampled with Gauss function simulated the spectral response function,established the prediction model of Pb by spectral data of different spectral sampling interval,used the model decision coefficient,root mean square error and ratio of performance to deviation as the comprehensive evaluation of the accuracy of the model evaluation index,get the best spectral sampling interval of Pb content prediction in this study.In this paper,through the study draws the following conclusions: 1)The prediction model based on multi spectral images has the ability to roughly predict the content of heavy metal Pb;2)Through correlation analysis and stepwise regression analysis to obtain the characteristic wavelength position of Pb content inversion is 1262 nm,2358nm,842nm;3)Compared the precision of prediction model by reciprocal?logarithmic and the first derivative transform,equation shows that the first derivative model has the highest accuracy,the first derivative transform is an effective way of spectral transform monitoring of heavy metal content of Pb;4)The accuracy of Pb content prediction model is affected by spectral sampling interval,it increases with the increase of spectral sampling interval,and then decreases.When the sampling interval is 42 nm,the model has the highest accuracy,and the ratio of performance to deviation is 2.15.The results show that the multi spectral remote sensing image has a rough prediction of cultivated land soil content of heavy metals Pb,can meet the requirements of the low accuracy of a wide range of monitoring,through discussion on the measured spectra of best spectral sampling interval,for wave monitoring of heavy metals Pb content of satellite sensor section provides reference for the establishment,and supplies a new idea for monitoring of other heavy metals.