Analysis of hyperspectral and lidar data: Remote optical mineralogy and fracture identification

Abstract

Karst systems are widely recognized as highly complex and often extremely productive reservoirs of water as well as petroleum. They are also often associated with mineralization. The availability of a large (several tens of square kilometers), well-preserved paleokarst outcrop is rare; therefore, maximizing the information that we can extract from examples like the Franklin Mountains is critical to the study of karst-related fluid flow. The mapping process is confounded by the need to map very large areas to find relatively small and somewhat unpredictable zones of extreme deformation. Moreover, the brecciated regions interpreted to be of karst origin are often composed of the same lithology as the surrounding rock and thus make traditional remote sensing data such as multispectral satellite imagery or photographic data inadequate to delineate such systems. The Franklin Mountains in El Paso, Texas, expose lower Paleozoic carbonates deposited over a giant carbonate platform referred to as the Great Ordovician Bank. The limestone-dominated bank was subsequently modified by surface karst and several large, vertically extensive caves that occupy up to 70,000 m2 of outcrop each. The breccia bodies are preferentially dolomitized within the limestone host rock. The size of these features is ideal for testing dolomite-calcite identification with high-elevation hyperspectral imagery at 20-m × 20-m pixel size. Terrestrial-based lidar (light detection and ranging) data were also utilized to identify collapse brecciation highlighted by hyperspectral image analysis. Results of this study delineate the distribution of dolomite and calcite in natural, passive light, well outside the visible spectrum, and combine active (lidar) and passive remote-sensing technologies to conduct remote mineralogical mapping linked to diagenetic alteration of carbonates. Through the combination of hyperspectral image processing and shape/texture analysis of terrestrial lidar data, a quantitative, multiscale facies map was generated in three-dimensional, geographically rectified space.