| The rapid development of economy will lead to the demand for mineral resources.At present,the scarcity of mineral resources in China and even the world requires the country to invest a lot of human and material resources to effectively explore mineral resources.In particular,it is located in the plateau area,and the journey is very difficult to reach.In order to not only save manpower and material resources,but also ensure the accuracy of exploration,the research of using remote sensing technology to predict minerals has obvious economic benefits.With the development of remote sensing technology,hyperspectral can not only record the spatial information of ground objects but also the spectral information of ground objects by virtue of its wide spectral range,so as to realize the advantage of integrating maps,and stand out among other remote sensing technology detection methods.Feldspar is the main rock forming minerals surface rock,occupy about60% of all the earth’s surface,analysis of feldspar in hyperspectral remote sensing range of the diagnostic spectrum with the contents of the related relationship,combined with the Hyperion applications of remote sensing image data,can be more accurate on the surface of various kinds of rock naming,thus to provide search for mineral resources directly or indirectly.The spectral data used in this study came from the spectral database established by the United States geological survey(USGS),namely the spectral data of 23 rocks in the USGS database and the content data of feldspar contained in them.Based on the above data,the original spectral reflectance data,to envelope spectrum reflectance data,differential treatment of spectral reflectance data and after wavelet decomposition of the layers of decomposed high frequency of reflectance spectra data of the feldspar content in feldspar and rock correlation analysis,summarized in the visible light-near infrared(VNIR,400-1000 nm),short wave infrared(SWIR,900-2500 nm)spectral data within the scope of the diagnostic spectrum.Based on this,through multiple stepwise regression analysis and partial least square regression modeling method,a semi-quantitative analysis method based on the correlation between spectral reflectance of rocks and feldspar content was proposed,and a remote sensing information identification method based on different feldspar content was constructed and applied to the classification of mineral types of satellite data.The main achievements of this paper are as follows:(1)Wavelet analysis is introduced to extract the fine components of rock spectrum dataFeldspar is the main rock-making mineral in the earth’s crust.However,the content of feldspar plays an inestimable role in the naming of surface rocks and the division of components.It is of great significance to explore the changes of characteristic spectra of feldspar under different contents for studying rock naming and mineralization mechanism.With the change of feldspar content,the spectral reflectance of rock also changes.Therefore,it becomes a new technical method to study the relationship between the electromagnetic radiation characteristics of rock and feldspar content.Based on the spectral response mechanism,it can be concluded that: In the measured spectral resolution of feldspar,the small absorption captured in the visible range is caused by Fe2+and Fe3+;in the range of 1002nm~2000nm,the absorption of 1400 nm and 1900 nm is the absorption valley caused by the vibration of OH-,in the range of 2000nm-2500 nm,it is the characteristic absorption band of igneous rocks.At the same time,on the basis of the original spectral analysis,envelope removal,differential spectral analysis and wavelet decomposition were introduced to obtain more obvious spectral absorption at 907 nm,1346nm,1437 nm,2204nm,2305 nm and 2325 nm than the original data.(2)The correlation between the spectral characteristics of rocks and the content of feldspar is summarizedIn the study of the correlation between the spectral characteristics of rocks and feldspar content,the characteristic spectra after the deenveloped processing is more obvious than that after the original spectral data processing.Among them,the characteristic spectrum after wavelet decomposition of the original data and after wavelet decomposition of the unenveloped data has the best processing effect.(3)Through the correlation between the characteristic spectrum and feldspar content,the optimal band is selected to construct the inversion model of feldspar content,and the data on the satellite are appliedBased on the correlation between spectral reflectance of rock and feldspar content,and combining with the band parameter information of remote sensing image data,the characteristic absorption bands of feldspar are proposed,which are mainly concentrated with b15,b29,b38,b81,b115,b157,b181,b215,b223,etc.At these bands,the absorption bands of rock have the greatest correlation with feldspar content.The inversion model of characteristic spectrum and content of feldspar was established.The results show that the multivariate stepwise regression analysis model and partial least square regression model after the first order wavelet decomposition have the best model accuracy,and the inversion model is the best when applied to the on-board data. |
PDF Full Text Request