Hyperspectral Remote Sensing Inversion Study Of Farmland Soil Heavy Metal Content

Remote sensing technology can diagnose and monitor soil heavy metal pollution,which has an important practical significance on food security,public health and the ecological environment.However,high spectral responses mechanism of vegetation's biochemical features in the natural environment is complex.Due to the interference signal,it is difficult to identify slight changes of vegetation's spectral response signal polluted by heavy metal.The reliability of employing the original spectrum signal to assess soil heavy metal pollution level is not high.Therefore,in order to realize the separation and strengthen of remote sensing weak signal,stripping background noise is essential.Furthermore,the establishment of soil heavy metal pollution assessment model is the key to the recognition and diagnosis of farmland pollution of heavy metals stress.It is also an scientific problem to be resolved when the application of remote sensing technology becomes more fine.A region of Yueyang county is the experimental study area.Wavelet analysis and singularity detection mechanism are used to identify and extract spectral weak signal of vegetation polluted by heavy metal stress.On this basis,linked with vegetation biochemical response stressed by the pollution of heavy metals and mechanism of remote sensing information,the comprehensive evaluation of farmland soil's heavy metal pollution stress levels can be acquired.The main content and research results in this paper are as follows:(1)Combined with characteristics of physical and chemical properties of soil samples in the experimental zone and correlation analysis of soil heavy metals,different degrees of soil heavy metal accumulation is showed;(2)Weak remote sensing signal of soil polluted by heavy metal stress is identified and diagnosed by wavelet analysis.Then,Based on spectrum and spectral decomposition signal singularity,two kinds of spectrum indexes related with the diagnosis of soil heavy metal pollution are constructed.They are wavelet detail coefficients characterizing the distortion information and the index of vegetation reflecting the singularity spectrum;(3)The comprehensive diagnosis index with a good detection ability can be filtered by the statistical correlation analysis.Then,mapping these spectrum parameters onto the fuzzy neural network model as soil heavy metal pollution level is performed.In addition,a spectral analysis model of soil heavy metals in the natural farmland ecosystem is proposed in order to find out the tiny change using remote sensed data in large area.The research represented by natural farmland ecosystem in a given area of Yueyang.Based on the analysis of research on the farmland pollution levels of heavy metals,the effects of soil heavy metal pollution on the surrounding environment is showed,which provides basic data supporting for the ecological environment management.The research can promote the application of quantitative theory in farmland environment pollution monitoring.Meanwhile,it is also able to lay a foundation for the rapid,quantitative monitoring and evaluation of soil heavy metal pollution.