Extraction Of Mineralization Alteration Information In The Metallogenic Belt Of The Middle And Lower Reaches Of The Yangtze River Based On Sentinel-2 Data

The Nanling–Xuancheng ore district is a newly discovered region of mineralization within the Middle–Lower Yangtze River metallogenic Belt in China.In recent years,the Xuancheng ore concentration area has made some breakthroughs in the discovery and exploration of deposits.More than ten skarn-type deposits(or mineralization)have been discovered,which shows that the Nanling—Xuancheng ore district has good prospecting potential.Due to the influence of high vegetation coverage and Quaternary sediment coverage,it is difficult to apply remote sensing technology to extract mineralized alteration information in this area.Sentinel-2 satellite data have high spectral resolution in the visible-near-infrared band,and it provide three red-edge bands for effective monitoring of vegetation.Moreover,its high spatial resolution and short return period provide a new data source for remote sensing mineralization alteration information extraction.Based on the characteristics of Sentinel-2 satellite data,this paper takes Nanling-Xuancheng area as the study area,and studies the methods of vegetation information extraction based on red-edge index,band selection and interference information removal based on spectral characteristics based on the advantages of spatial resolution,red-edge characteristics and wave spectrum characteristics of Sentinel-2satellite data,in order to improve the extraction of mineralized alteration information by remote sensing in vegetation covered area This study is aimed at improving the accuracy of remote sensing mineralization alteration information extraction in vegetation-covered areas,and providing support for mineral resources exploration in the middle and lower reaches of Yangtze River mineralization zone.The main research results and innovations are as follows:(1)The three red-edge bands added to the Sentinel-2 satellite data provide a more effective means for vegetation information extraction in the mining area.In this study,the vegetation information of different coverage in the study area was classified using the combination of VNIR bands or red-edge vegetation indices based on Sentinel-2 satellite data.By comparing the classification accuracy of different combinations,it is found that red-edge band 704,red-edge band 740 and red-edge band 783 all have a certain improvement effect on vegetation classification accuracy.The vegetation classification accuracy of using three red-edge vegetation indices NDVI₇₀₄,NDVI₇₄₀ and NDVI₇₈₃simultaneously is better than that of using any of the red-edge indices alone,and can be more effective for vegetation disturbance information removal.(2)By analyzing the correspondence between the bands of Sentinel-2 satellite data and the spectral features of typical alteration minerals in the study area,B2,B6,B8a and B11were selected as the bands for extracting information on iron-stained anomalous alteration.B6,B8a,B11 and B12 were selected as Al-OH anomalous alteration information extraction bands;B2,B8a,B11 and B12 were selected as Mg-OH anomalous alteration information extraction bands.The results show that the Sentinel-2 data are more relevant in identifying iron-staining anomalous alteration information because of the more detailed division in the visible-near-infrared band;in identifying hydroxyl-containing alteration anomalies,the Sentinel-2 data can achieve further refinement in extracting hydroxyl-type alteration anomalies.(3)Based on Sentinel-2 satellite data,the interference information was removed by masking method and hybrid image element decomposition method,and the alteration information of iron-staining anomaly,Al-OH anomaly and Mg-OH anomaly in the study area was extracted by principal component analysis,and the results of mineralized alteration information extraction were analyzed by superposition with known mineralized(chemical)points and linear structures in the area and statistical analysis.The extraction results based on Sentinel-2 satellite data and Landsat-8 data are compared and analyzed,as well as the extraction results based on mask method and hybrid image element decomposition method for interference information removal.The results show that the extraction results of alteration information from Sentinel-2 satellite data and Landsat-8are in general agreement in terms of spatial distribution,and the results of Sentinel-2 data are in better agreement with known mineral sites,which verifies the feasibility and effectiveness of Sentinel-2 data for remote sensing mineral search in vegetation-covered areas.The area of the eclipsed area extracted by the hybrid image element decomposition method is larger than that by the mask removal method,and the correlation ratio with the ore point is also higher,indicating that the hybrid image element decomposition method can make more use of the remaining information after the interfering end element removal and spectral reconstruction than the mask removal method.