Abstract:

This study selected Yingzuishan goldmine and its surrounding area in the east of Altun Mountain in Northwestern China as research area, and tried to find out a method that could automatically extract out the information of rock mass and concealed rock mass from Landsat-7 ETM+data in this area. The multi-level separation and extraction technique of remote sensing information was used, and we also statistically analyzed in-situ spectral data of ground objects and multivariate data from rock and mineral samples, and finally acquired several kinds of band combination types related to different rock mass alterations. On the basis of multi-band remote sensing data, we utilized the arithmetical operations, ratios and principal component analyses (PCA) etc, and obtained some variable images after information enhance treatment. Combining with the research on characteristic of the known rock mass, we found out the best variable image that included rock mass and concealed rock mass information at most, and extracted out the information from this image in the end. Field validation and comparative analysis between the results and the archived geological rock mass data both proved that this method has the advantage of high goodness of fit and satisfactory effect. Furthermore, the reasons for the error of the extracted information were also discussed.

Key words: rock mass, concealed rock mass, remote sensing, information extraction, Yingzuishan goldmine, Gansu Province