Lithium (Li) resources are widely used in many strategic emerging fields; recently, several large-scale to super-large-scale pegmatite-type lithium deposits have been discovered in Dahongliutan, NW China. However, the natural environmental conditions in the Dahongliutan area are extremely harsh; hence, manpower in field exploration is difficult to achieve. Efficient and rapid methods for identifying Li-rich pegmatites, based on hyperspectral remote sensing technology, have great potential for promoting the discovery of lithium resources. Ground spectral research is the cornerstone of regional hyperspectral imaging (HSI) for geological mapping. Direct observation and analysis by the naked eye are part of a process that is mainly dependent upon abundant experience and knowledge from experts. Machine learning (ML) technology has the advantages of automatic feature extraction and relationship characterization. Therefore, identifying the spectral features of Li-rich pegmatite via ML can accurately and efficiently distinguish the spectral characteristics of Li-rich pegmatites and Li-poor pegmatites, enabling further excavation to identify the strongest predictors of Li-pegmatite and laying a foundation for the accurate extraction of Li-rich pegmatites in the West Kunlun region using HSI. The spectral characteristics of pegmatite in the visible near-infrared and shortwave infrared (VNIR–SWIR) spectra were observed and analyzed. Li-rich pegmatite was identified based on the diagnostic spectral waveform characteristic parameters of the local wavelength range. The results demonstrated that the pegmatite ML recognition model was based on spectral characteristic parameters of the local wavelength range, with good model explicability, and the area under the curve (AUC) calculated for the model is 0.843. A recognition model based on full-range spectrum data achieved a higher precision, and the AUC value was up to 0.977. The evaluation of the Gini coefficient presented the strongest predictors, which were used to map the spatial distribution lithology, based on GF-5, in Akesayi and the 509 mines, producing encouraging lithological mapping results (Kappa > 0.9, OA > 94%).

This study investigates the application of hyperspectral image space–spectral fusion technology in lithologic classification, using data from China’s GF-5 and Europe’s Sentinel-2A. The research focuses on the southern region of Tuanjie Peak in the Western Kunlun Range, comparing five space–spectral fusion methods: GSA, SFIM, CNMF, HySure, and NonRegSRNet. To comprehensively evaluate the effectiveness and applicability of these fusion methods, the study conducts a comprehensive assessment from three aspects: evaluation of fusion effects, lithologic classification experiments, and field validation. In the evaluation of fusion effects, the study uses an index analysis and comparison of spectral curves before and after fusion, concluding that the GSA fusion method performs the best. For lithologic classification, the Random Forest (RF) classification method is used, training with both area and point samples. The classification results from area sample training show significantly higher overall accuracy compared to point samples, aligning well with 1:50,000 scale geological maps. In field validation, the study employs on-site verification combined with microscopic identification and comparison of images with actual spectral fusion, finding that the classification results for the five lithologies are essentially consistent with field validation results. The “GSA+RF” method combination established in this paper, based on data from GF-5 and Sentinel-2A satellites, can provide technical support for lithological classification in similar high-altitude regions.

Mineral Prospectivity Mapping (MPM) is shifting toward intelligent deep mineralization searches in the era of big data and the increasing difficulties of surface deposit detection. Comparative analysis of two forms of mineralization prediction based on the Apriori algorithm was performed in the Meiling South mining area in the eastern Hami region of Xinjiang, China. In comparison 1, we use the Apriori algorithm to mine ore-forming information and determine the ore-forming voxel positions based on spatial distance and angle analysis. Then, we compare the ore-forming voxel positions determined by Apriori with the ore-forming voxel positions predicted by the mathematical model based on the conceptual model of mineralization, and these mathematical models include Gaussian Naive Bayesian (GNB) and Support Vector Machine (SVM). In comparison 2, the optimal prediction model is SVM, which is trained using the elements of mineralization prediction determined by the conceptual model of mineralization. Then, two sets of new elements of mineralization prediction are extracted from the original elements of mineralization prediction using the Apriori and Chi-square methods and then input into the SVM model for training. After we obtain the mineralization prediction results, we compare them with the original mineralization prediction results. The preceding comparison produced the following results. (1) Using the Apriori algorithm, the distribution characteristics of the high and low-grade ore bodies and the association rules between ore-bearing information were determined. (2) The prediction results of the GNB and SVM models displayed corresponding trends on the high and low-grade ore-bearing voxels identified by Apriori, which matched the rules mined by Apriori. (3) In comparison to the mineralization prediction elements screened by Chi-square and the original mineralization prediction elements based on the conceptual model of mineralization, the elements of mineralization prediction chosen based on Apriori have the best prediction effect in SVM when tested in new drill holes. Based on the mineralization prediction elements screened by Apriori, the number of accurate ore-bearing voxels (prediction probability greater than 0.5) predicted by the SVM model is 6, 5, and 1 in drill holes V1, V2, and V3, respectively. The collective results demonstrated that Apriori is explicit, intuitive, and interpretable for mineralization prediction and has a certain reference value for refining the determination of mineralization prediction elements and discovering mineralization mechanisms and laws.

Lithium is a rare metal widely used in the renewable energy industry. The Altyn region in Xinjiang, China, contains abundant granitic pegmatite-type lithium resources; however, the deeply incised and complex terrain limits the accuracy of conventional two-dimensional remote sensing approaches for dike identification and segmentation. To address this limitation, a remote sensing segmentation method incorporating terrain information was proposed. A digital elevation model (DEM) derived from LiDAR data, together with its associated topographic factors, was integrated into the Spatial–Spectral Mamba framework to enable the joint utilization of spectral and terrain features. Rather than performing explicit three-dimensional geometric modeling, the proposed approach enhances a two-dimensional segmentation framework by introducing elevation-derived information, allowing the model to capture terrain-related spatial variations of pegmatite dikes. This design enables improved representation of both the planar distribution and terrain-influenced morphological characteristics of dikes under deeply incised conditions. The Xichanggou lithium deposit in the Altyn region is a large-scale, economically valuable pegmatite-type lithium deposit, and was therefore selected as the study area for pegmatite dike segmentation. The results demonstrated that, compared with conventional two-dimensional approaches and representative machine learning methods, the proposed method achieved higher segmentation accuracy in complex terrain. Improvements were also observed in the continuity and spatial consistency of the extracted dike patterns. Field verification indicated that the major pegmatite dikes delineated by the model were highly consistent with their actual surface exposures. Sampling analyses further confirmed the validity and reliability of the identification results. Overall, the terrain-integrated remote sensing segmentation approach exhibited good applicability and robustness under deeply incised and complex geomorphological conditions.