Research On The Inversion Model Of Farmland Soil Heavy-metal Based On Different Scales Spectrum

At present, the method for monitoring heavy metal pollution of farmland soil is mainly measured by physical or chemical analysis method based on the ground points sampling,this discrete、transient and regional monitoring methods could not dynamic inverse the overall situation of experimental area of heavy metal pollution by comprehensively due to lack of time and space continuity. With the increasing of heavy metal pollution of arable soil in our country, the heavy metal pollution monitoring needed to further develop from macro-level to micro-level, from static to dynamic and from the instantaneous to continuous.At the moment, the application of quantitative remote sensing technology is to meet the demand of the development.This paper chooses Qingshuitang industrial park of Zhuzhou city as the test area,collecting hyperspectral data and measuring the contents of heavy metal of farmland soil in the field. On the basis of the data acquisition, this study adopts the method by combining the spectral analysis with the method of correlation analysis to extract the spectral sensitivity parameters of soil spectrum of Cd, Cu, As, Pb and Zn in the soil, and then the hyper-spectral inversion model of the soil heavy metal is established by the least squares fitting. At the same time, using the measured hyper-spectral data to simulate multi-spectral data according to the spectral response function of Landsat ETM+ sensor. A study of the quantitative relationship between simulated multi-spectral data and soil heavy metals,aimed to explore the feasibility of using multi-spectral data to predict cultivated soil heavy metal pollution.Research results mainly includes: 1) The heavy metal concentrations such as Cd, Cu, As,Pb and Zn were over medium pollution in Qingshuitang industrial park, the Severity of pollution is: Cd>Pb>Zn>Cu>As, and there is a strong occurrence relationship between Cd and Pb and Zn and Cu; 2)It is concluded, through spectrum analysis, that the ranges of spectral characteristic wavelength mainly are 420 ~ 520 nm, 650 ~ 700 nm and 920 ~ 950 nm; 3) the Cu content inversion model based on the measured hyperspectral spectrum and the simulationspectrum of ETM+ sensors has a high consistency, shows that copper has sensitive spectral characteristics, this will lay the foundation for the quantitative inversion of other elements which do not have spectral characteristics(such as Cd, Zn, etc.). 4) The inversion model of measured hyperspectral data is better than the inversion model of ETM + multi-spectral data including the correlation coefficient and precision, shows that the proper spectral resolution can improve the accuracy and reliability of the quantitative analysis of soil heavy metal;5)Research has shown that multi-spectral has the potential to assess the soil heavy metal,especially for the element of Cu, As, the soil index can effectively improve the As’ sensitivity to the multi-spectral,its relative error was 11.57% lower.Research application shows that it can obtain the quantitative information of soil heavy metal based on hyperspectral data and simulate multi-spectral data, this provides new ideas and new method to achieve a wide range of monitoring soil heavy metal pollution. It is also of certain practical significance and of value for reference for similar remote sensing inversion of soil heavy metals.