Extraction Of Mineralized Alteration Information In Qinghai Gouli Area Based On Hyperspectral Remote Sensing

Mineralization alteration is an effective metallogenic indicator of information,which has important geological significance for mineral resources exploration and evaluation,prospecting prospect delineation,and exploration target area.Hyperspectral remote sensing gradually rose and developed in the 1980 s.It uses electromagnetic waves to detect and image surface objects,which can not only obtain spatial images but also provide rich continuous spectral information.Hyperspectral imaging technology can accurately detect and analyze the composition of surface rocks and minerals.It is an important means to detect mineralization and alteration information quickly,in a large area,and at a low cost.It has gradually become hightech in the field of green exploration of mineral resources.This study is based on a ZY1-02 D hyperspectral remote sensing image,taking the Gouli area on the southern edge of Qaidam Basin in Qinghai as the research area.Based on the full analysis of existing geological data,the hyperspectral remote sensing image is preprocessed,and the quality of image processing results is evaluated by Pearson correlation coefficient combined with field measured spectrum.By analyzing the metallogenic mechanism and wall rock alteration characteristics of typical deposits in the study area,determining the types of typically altered minerals in the area,obtaining the spectral sample data of altered minerals based on the spectral angle matching method,extracting the typical mineralization and alteration information of the study area by using the matched tuned filter model(MTMF)and extreme random tree model(ERT),and according to the geological and metallogenic conditions,the distribution of known deposits(points)and field exploration,The information extraction results of altered minerals are comprehensively compared,analyzed and evaluated.The main work and achievements are as follows:(1)The water vapor band removal,band repair,radiometric calibration,atmospheric correction,and other preprocessing work are carried out on the ZY1-02 D space-borne hyperspectral data to eliminate the influence of external environment,atmosphere,and other factors,and the real surface reflectance is obtained;Combined with vegetation index(NDVI)and water body index(NDWI),the vegetation and water information that interfere with the extraction effect of image alteration information are eliminated by mask method,so as to provide a reliable image data source for alteration information extraction.(2)According to the metallogenic mechanism and wall rock alteration characteristics of typical deposits in the study area,limonite,chlorite,calcite,kaolin,and sericite are determined as the main altered mineral types in the area,and their spectral characteristics and formation mechanism are deeply analyzed.Based on the standard spectra of typical minerals in the USGS spectrum library and the spectral angle matching method,the spectral sample sets of altered minerals are constructed respectively.Combined with MTMF algorithm and ERT algorithm,the hyperspectral extraction model of mineralization alteration information is constructed to complete the extraction of typical mineralization alteration information in the Gouli area of Qinghai Province.(3)The results show that the mineralization alteration information extracted by the two models is highly similar to the spatial distribution trend of fault structures in the area.At the same time,it overlaps with the known metallogenic area,which is in line with the geological and metallogenic characteristics of the study area.The extraction effect of ERT model is better than that of MTMF model.