Remote sensing technology is widely used in basic geological survey, mineral resources exploration, environmental assessment and geological disaster survey. It has developed from multispectral to hyperspectral. Landsat- 8 is the most representative and commonly used multispectral data at present. ASTER has high resolution and multi band characteristics. Resource- 1 02D (ZY1- 02D) satellite is a hyperspectral operational satellite launched in 2019. In order to better understand the role of multi- source remote sensing data in rock and mineral identification, relevant research was carried out in karamaili area of East Tianshan, Xinjiang. Methods：PCA transformation of Landsat- 8 data can obtain an image with rich color and huge amount of information. According to the color, texture, shape, water system and other characteristics of false color synthetic images (R: PC3, G: PC2, B: PC1) of different formation lithology after principal component analysis, all strata and most lithology in the study area can be identified with reference to geological data. For Landsat- 8 OLI image after atmospheric correction, the smacc method is used to extract the image endmember spectrum. For ASTER and ZY1- 02D ahsi images after atmospheric correction, fringe removal and MNF processing, PPI and n- dimensional visualization tools are used to extract the endmember spectrum, and spectral analysis is carried out with the pop of USGS standard mineral spectrum library as a reference to determine the mineral types represented by the extracted endmember pop. Finally, the endmember pop is regarded as a known pop, Sam method and image pop are used for comparative analysis to achieve the purpose of mineral classification and recognition. Through field verification, GIS software is used to modify the mineral identification and classification map of multi- source remote sensing data, so as to obtain an accurate comprehensive map of mineral identification and classification in the study area.Results：The results show that the PCA transformation results of landsat- 8 oli clearly identify different lithology and strata in the study area; Using landsat- 8 oli, aster and zy1- 02d hyperspectral data, different image endmember extraction methods are adopted respectively. On the basis of spectral analysis, the classification map of main minerals in the study area can be obtained by spectral angle mapping (SAM). Through field verification, GIS technology is applied for integration and analysis, and the accurate comprehensive map of mineral classification is obtained after correcting the relevant maps. Conclusions：The research shows that the integration of multi- source remote sensing data has good effect and great prospect in rock and mineral identification.