Using the HySpex hyperspectral imaging system for mapping the alteration zones in the Yudai, Kalatage district, NW China

The HySpex hyperspectral data used in this study have a wide spectral response range, narrow bandwidth, and high spatial resolution, and they can be effectively applied to the extraction of mineral alteration information. We explore how to extract effective information from remote sensing images through remote sensing image classification technology and explore its utilization in geological science. This study aims to verify the reliability and accuracy of alteration information extracted by using a super-low altitude detection platform equipped with powered delta wings mounted with HySpex hyperspectral sensors in the Yudai area of the Kalatage district. Field data were collected and analyzed using Analytical Spectral Devices (ASD), and the results were compared with those obtained from the United States Geological Survey (USGS) spectral library. The analysis of the geological background and HySpex hyperspectral data was enhanced by Minimum Noise Fraction (MNF) transformation coupled with the Pixel Purity Index (PPI) to extract endmembers of altered minerals, including chlorite and jarosite, from different spectra (SWIR and VNIR) and spectral wavelengths. Additionally, two classification methods, the Spectral Angle Mapper (SAM) and Support Vector Machine (SVM), were applied to the data for effective mineral mapping. The best-performing method, i.e., SVM, was validated using ground-truth information obtained during field observations. The results from the classification methods revealed accuracies of 59.57% for SAM and 69.25% for SVM. The HySpex hyperspectral data obtained using a super-low altitude detection platform proved highly effective for detecting altered rock information. Thus, this approach has great potential for the rapid identification of geological and mineralogical features, especially in complex terrains.