Abstract: Hyperspectral remote sensing imagery, with its abundant spectral information, has proven to be a vital tool in mineral exploration by effectively extracting and identifying mineral alteration features. The integration of hyperspectral data with machine learning met hods has led to numerous successful applications in mineral prospectivity mapping in rece nt years. However, conventional remote sensing-based exploration approaches predominantly focus on the hyperspectral data itself and lack effective integration with geological inform ation, which limits further improvement in exploration effectiveness. To bridge this gap, th is study develops a novel methodology that integrates hyperspectral imagery with critical g eological determinants—specifically pluton boundaries and fault systems, makeing 39 chann el hyperspectral geological information comprehensive data set and introduces an enhanced Graph Convolutional Network (GCN) model. Architectural improvements include the incor poration of residual connections and the systematic application of batch normalization acro ss both residual modules and convolutional layers, substantially stabilizing and accelerating the training process. Validation using ZY-1 hyperspectral data from the Dahongliutan area demonstrates that our refined GCN model achieves superior accuracy in identifying miner alized granitic pegmatites. Quantitative evaluations confirm substantial performance gains, with accuracy improvements of 7, 22, and 27 percentage points over the baseline GCN, C onvolutional Neural Network, and Support Vector Machine models, respectively. This work establishes an effective and automated framework for high-precision prediction of lithiumand beryllium-mineralized granitic pegmatites via hyperspectral remote sensing.