Research On Remote Sensing Detection Methods For Mineral Resources In The East Tianshan Region Of Xinjiang

Rocks are composed of a wide range of minerals,and the absorption characteristics of their reflectance spectra are closely related to their mineral abundance.With the rapid development of remote sensing technology,the range of geological applications of multispectral and hyperspectral remote sensing data has increased from a few spectral bands to a hundred spectral bands.Geological exploration methods are diverse,and remote sensing technology is characterized by remote exploration,timeliness and low cost compared with traditional field exploration.Reflectance spectra in the visible-near-infrared bands are often used to study the physical and chemical properties of minerals.Geologists use diagnostic spectral absorption band features such as wavelength position,depth,width,area,and other absorption indices in the spectral absorption bands,which are sensitive to small differences in mineral spectra,to quantify mineral distribution and abundance.In this study,multiple statistical regression model,partial least squares model,and multilayer perceptron model are used to correlate rock spectra and mineral abundance,and on this basis,quantitative inversion of mineral abundance is carried out.The East Tien Shan of Xinjiang is an important mineral-rich area in China.In this study,we used multispectral image data based on remote sensing technology to qualitatively analyze the mineral species and mineral mapping in the East Tien Shan region of Xinjiang,and used hyperspectral image data to quantitatively invert the mineral abundance in the East Tien Shan region of Xinjiang.Based on Landsat 9 multispectral data,the band ratio method,the minimum noise separation method,and the hybrid tuned matched filter method are used to enhance the features of different mineral species in the images.Meanwhile,based on GF-5 AHSI hyperspectral data,the minimum wavelength mapper,spectral angle matching method for mineral identification,and the sparse mixing Hapke radiative transfer model are used to quantitatively invert the abundance of minerals in the East Tianshan region of Xinjiang.Through the comparative analysis of multi-source satellite data,the spatial distribution characteristics of minerals are studied,and the abundance is investigated using the quantitative inversion method.The main research results achieved in this study are as follows:(1)In the correlation analysis between rock spectra and mineral abundance,the absorption depth of spectral absorption features has the highest correlation with mineral abundance,and the multivariate statistical quadratic curve regression model is the most effective in the quantitative inversion of the abundance of dolomite based on absorption depth.In the quantitative inversion of minerals based on multidimensional spectral absorption features(absorption depth,absorption width,absorption area),the partial least squares regression model is better than the multilayer perceptron model,and the stability of the multidimensional absorption feature model is better than that of a single absorption feature.(2)The optimal band combinations(B4/B2,B6/B7,B10)of multispectral data Landsat 9 RGB are determined based on the band ratio method to enhance the spatial distribution characteristics of iron oxide minerals and silicate minerals.In addition,the minimum noise separation method enhances the distribution characteristics of chalcopyrite and clay minerals.Using the minimum wavelength mapper method for hyperspectral data processing,the spatial distribution of silicate minerals and iron oxides is obtained based on the spectral features of wavelength position and absorption depth,which lays the foundation for quantitative inversion.(3)Mineral abundance inversion using the sparse mixing Hapke radiative transfer model,the highest abundance of subimage element is chlorite as 0.78386%,and the high value is distributed in the northern part of Aqi Mountain;followed by acidifier chalcopyrite as 0.77631%,and the high value is distributed in the area of shallow metamorphic sandstone in the Great Fracture;and the highest abundance of iron-bearing oxides hematite and acicular ferrite are 0.76495% and 0.77059%;Montmorillonite,acicular iron ore and hematite abundance of high value distribution effect is the best with the geological map lithological distribution of spatial pattern is basically consistent.In this paper,taking the East Tien Shan region of Xinjiang as the study area,we carry out the extraction and quantitative inversion of mineral information by remote sensing technology,and put forward the detection method combining the Landsat 9 mineral distribution and the abundance inversion of the GF-5 AHSI sparse mixing Hapke model,which has the advantages of rapidity,economy and high efficiency,and provides the corresponding technical support and reference for the further prospecting of the mineral resources distribution in this region and thus for the mineral searching service.The method is fast,economical and efficient.