Isolating Locations of Potential Episodes of Cave Collapse and Their Relationship to Cave Level Development Through Major River System Incisions

Abstract

Karst landscapes, in which dissolution of bedrock is the dominant geomorphic process, make up 10%-20% of Earth’s land surfaces and supply between 20%-25% of the global population with drinking water. Dissolution dominates the genesis of karst systems, creating flow pathways, conduits, and caves. Cave patterns from dissolution can be influenced by regional factors, such as water table base-level fluctuations correlating to major river system incisions. During periods of negligible regional incisions, multiple cave levels may form. Despite the significant role dissolution plays in karstic genesis, physical erosional processes can enhance the formation of these karst systems and should not be ignored. For example, the lowering of the water table within a cave can expose the cave to more vadose conditions – leading to a decrease in roof-supporting buoyancy and ultimately the catastrophic failure of conduit ceilings resulting in areas of cave collapse. Cave collapse is an important indicator of the past hydrogeological and geomorphological conditions of a karst system; however, the location and extent of cave collapse are not always easily identifiable. Identifying areas that have experienced cave collapse can help uncover key clues for dissecting regional geologic history in terms of delineating cave levels for estimating previous base-levels and for reconstructing the timing of river system incisions. Using a LiDAR derived DEM, this study improves on previously constructed models for the delineation of cave levels as well as explores a new methodology for isolating areas that have experienced cave collapse. For cave level delineation, a histogram generated from extracted cave entrance elevations is clustered into four distribution groups. Two separate methods of delineation are explored, one using visual breaks in the data and the other utilizing Jenks Natural Breaks. Initial results indicated more investigation would be necessary to determine which methodology was more accurate. For isolating areas of cave collapse, a weighted overlay was constructed utilizing three parameters – slope, distance from caves, and distance from streams. A sensitivity analysis was conducted to determine the most effective weighted distribution, resulting in a distribution of 70%, 20% and 10% respectively. After comparisons between the two methodologies of cave levels and areas with the highest probabilities of past collapse, it was determined that the natural breaks methodology was more accurate. This conclusion was drawn as visible correlations exist between areas of high probability of past collapse and the contacts between cave levels. This indicates that cave collapse may preferentially occur at