THE APPLICATION OF HYPERSPECTRAL IMAGING TO GEOLOGICAL STUDIES

Hyperspectral data of the earth’s surfaces with a high degree of accuracy is often very difficult to obtain. Therefore, hyperspectral imagery is largely underutilized in geological studies. Through this research project, the application of hyperspectral imaging for the identification of igneous rocks were explored. In order to acquire hyperspectral data, geological samples collected in the field were studied. Three different types of materials were used to generate hyperspectral images, including: hand samples, thin sections, and rock powders. The goal was to establish which material would provide the highest degree of accuracy when depicting the spectral reflectance patterns present in the samples. Various imaging techniques were applied when generating the hyperspectral images. These techniques were used to overcome certain obstacles, such as oversaturation and orthorectification. In order to gauge accuracy, the spectral reflectance patterns generated in the lab were compared to a pre-existing United States Geological Survey (USGS ) spectral library. This was done by using an original python script to compare similarities in spectral reflectance curves in both data sets. The results show, that due to the imagers inability to magnify the materials present in the image, thin sections are not good candidates for hyperspectral imaging. The results also found that materials that formed intrusively generate the best results in a hyperspectral image. Also, because of the coarse grained nature of intrusive igneous rocks, each individual mineral would present a clear and established spectral reflectance pattern.