Lithium-Bearing Pegmatite Identification, Based on Spectral Analysis and Machine Learning: A Case Study of the Dahongliutan Area, NW China

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