Discrimination of the Weathering Degree of Speleothems Based on Hyperspectral Characteristics

Accurate and non-destructive identification of the weathering degree of speleothems in tourist caves holds critical implications for establishing scientific conservation criteria. To address the limitations inherent in conventional rock weathering classification systems, this study employs hyperspectral data acquired from speleothems exhibiting varying weathering intensities. A random forest algorithm was utilized to identify diagnostically significant spectral features, which were subsequently leveraged to construct a decision tree model. The model's performance was rigorously evaluated using metrics including the F1-score and the Area Under the Curve (AUC). The results demonstrate that: (1) lichen colonization does not markedly alter the overall spectral profile of speleothems but leads to a concentration of the wavelength at the absorption trough minimum; (2) among eight spectral features analyzed, the combination of \(\:{Q}_{SAI}\),\(\:{\lambda\:}_{1}\), and \(\:{\lambda\:}_{m}\) exhibits the highest discriminative capacity; (3) the decision tree model developed based on this feature combination effectively classifies speleothems according to their weathering degree. These findings provide a theoretical foundation for advancing the weathering classification of speleothems.