Accurate identification and classification of lithology are the basis of mineral exploration. Xinjiang region is rich in minerals with great exploration potential. Due to its vast territory and sparse population， conducting field surveys of lithology is challenging， thus using remote sensing images to extract lithological information has become an important method. Furthermore， the region has minimal surface vegetation coverage， making it easier to identify rock types using remote sensing images effectively. Additionally， the increasing availability of open-source remote sensing data provides a rich data source for extracting lithological information. However， there are still many issues that need further research on how to comprehensively utilize these open-source remote sensing data for lithological identification. This study focuses on the western part of the Junggar Basin in Xinjiang， and Landsat 8 OLI， Sentinel-2 MSI， and ASTER data of different temporal phase are obtained. Firstly， various image band ratios， principal components， and other variables are extracted and selected for color composition to highlight different rock types. Subsequently， based on geological map， training samples are obtained from the known Darbut ophiolite belt. Then， SVM， maximum likelihood， and random forest methods are used to classify the Landsat 8 OLI and Sentinel-2 MSI images. Results of both methods indicate the presence of lithological anomaly belt in the northern part of the Darbut ophiolite belt in the western Junggar Basin， with remote sensing image characteristics similar to those of the Darbut ophiolite belt. The anomaly belt extends in a northeast-southwest direction for over 20 km， with a maximum width of approximately 1 km. However， there exist different opinions on the lithological characteristics of this anomaly belt， and further verification through field survey is needed to confirm the final results.