Abstract:Saishiteng belongs to the tectonic belt in the northern margin of Qaidam Basin, and its lithology is complex and changeable. The lithologic boundary of the study area is generally divided by predecessors. Landsat- 8 OLI, ASTER and Sentinel- 2A image data are selected to identify the main lithology by information enhancement method. Methods: The optimal band index was adopted to determine the band combination of each image, which highlighted the boundary of different rocks. Landsat- 8 + ASTER (LA) data and Sentinel- 2A + ASTER (SA) data were formed by combining ASTER short- wave infrared band with Landsat- 8 OLI and Sentinel- 2A visible- near infrared band. The standard spectral information of resampled rocks was analyzed, and the calculation formulas of different rock bands were drawn up. Based on multifractal theory, the threshold range of different rock types was selected to obtain the distribution of main lithology. According to the resampled standard spectral curve of biotite, 2262 nm band and 2336 nm band of SA data were selected for directional principal component analysis, and the second principal component was segmented by Crosta threshold method, and the anomaly grade of biotite was divided, which was associated with lithology distribution and identified the main lithology distribution. Results: The remote sensing interpretation results were improved by combining the measured spectral analysis of rocks, the identification under thin section microscope and the field geological investigation. The results of lithologic extraction showed that the newly discovered U- shaped belt on the east side of Xiaosaishiteng was gabbro—tonalite—monzogranite. The boundaries of the third and fourth rock groups of Dakendaban Group, the first group of Tanjianshan Group, granite, monzogranite, biotite granite, rhyolite, porphyritic quartz diorite, quartz diorite, tonalite, gabbro diorite and gabbro were re- delineated. Conclusions: Based on the spectral resolution advantage of Sentinel- 2A and Landsat- 8 data in visible- near infrared band, the optimal band combination is used to enhance the image color. The results show that the pseudo- color composite image obtained by this method is rich in color and clear in boundary, and has more advantages in boundary division than ASTER data with the same resolution. The short- wave infrared band of ASTER data is processed with the visible- near infrared band of Landsat- 8 data and Sentinel- 2A data, and the SA data and LA data obtained can retain the spectral diagnostic characteristics of rocks and minerals to the greatest extent. Based on the different characteristics of spectral diagnosis of rocks and minerals, the band operation method is used to highlight the different lithologic information in collaborative data, which can provide guidance for the next field geological work. According to the spectral characteristics of biotite minerals in short- wave infrared band, Based on directional principal component analysis and field investigation, it can be seen that the anomaly grade of biotite is related to the rock type in the area. When the anomaly grade of biotite is concentrated in a specific range, it can indicate the range of Dakendaban Group (Pt1D) and provide reference for lithologic mapping in adjacent areas.