Ground-Based Hyperspectral Imaging and Data Analysis for Near-Vertical Outcrop Surfaces

Advances in remote sensing technology have expanded outcrop data collection and analysis capabilities creating new opportunities to carry out geological studies in greater detail. Until recently, the hyperspectral data used in geology were acquired mostly from airborne platforms along with some from spaceborne platforms. Due to nadir-viewing geometry, however, these data fail to provide an adequate viewing angle for near-vertical outcrops which often provide excellent opportunity for detailed observations. Thus, there has been an increasing interest in hyperspectral data collected from ground-based platforms for studying near-vertical rock exposures. The purpose of this research is to use multi-modal data with an emphasis on ground-based hyperspectral imagery for spectral analysis of near-vertical outcrops complementing the conventional field methods. Ground-based hyperspectral data were collected using a field-deployable imaging system which employs two SpecIm hyperspectral sensors operating in visible-near infrared (VNIR) and shortwave infrared (SWIR) ranges. A workflow including data collection, sensor artifact removal, image preprocessing, and spectral data analysis was implemented. Three case studies from different geological settings were carried out to evaluate the potential of ground-based hyperspectral imaging in geological studies. In Lisbon Valley, Utah, multi-scale remote detection of fluid-related diagenetic mineralogical changes in Wingate Sandstone was performed. Spectral analysis at different scales not only suggested multiple pathways of fluid flow but also provided insight about relative direction of the flow. In southwestern Missouri, a roadcut exposing Reeds Springs Formation was studied to map different lithologies and alteration products. The spatial distribution pattern of limestone and tripolite suggested leaching of carbonates and subsequent tripolitization was a result of subaerial exposure and caused by percolating meteoric water. In Goldstrike District, Utah, full-spectrum image analysis bringing data from VNIR and SWIR sensors was performed for remote lithological mapping of the Claron Formation rocks. The results demonstrate the potential of ground-based hyperspectral imaging as a non-contact method for detecting and identifying lithological units and for analyzing chemical-mineralogical properties of near-vertical outcrops with centimeter-scale spatial resolution.