Hyperspectral imaging for mineral composition and weathering recognition of rock slope on the reservoir bank

The mineral composition and weathering degree of the hydro-fluctuation belt are affected by water–rock interaction. However, traditional methods for mineral identification and weathering assessment have lag and risk. Moreover, there is still a lack of uniform quantitative criteria for weathering assessment of rock slopes on the reservoir bank. Here, a hyperspectral imaging technique with superior performance in characterizing chemical reactions of object surfaces is applied in the weathering assessment of the hydro-fluctuation belt. Then, the hyperspectral image is combined with the Spectral Angle Mapping algorithm to achieve the classification of minerals. Meanwhile, a weathering evaluation system based on the Random Forest classification model is constructed in accordance with the proposed weathering index, which has a mapping relationship with the weathering degree. The mineral characteristics of the hydro-fluctuation belt are present with the enhancement of water–rock interaction, the relative content decreases in carbonate minerals and increases in clay minerals. The relative content of minerals is also affected by water–rock interaction time, rock mass deterioration, and water erosion. The weathering index is positively correlated with the weathering degree, which is affected by the changes in mineral and rock deterioration. The recognition of minerals and weathering degree by hyperspectral image provides a fast, non-contact, and quantitative method that improves the efficiency of mineral detection and weathering evaluation on the reservoir bank.