ASSESSING PAST CAVE COLLAPSE PROBABILITY USING GIS

The genesis of karst areas can result from a combination of dissolution and mechanical processes. Collapse of cave passages is recognized as one mechanical process that alters the surface topography in karst systems. For example, karst windows provide an indication of collapse; however, extended collapse along longer stretches can be difficult to identify. To assess the likelihood of past episodes of cave collapse, an analysis incorporating surface slope and karst features was conducted using GIS for an area bounded within Carter Caves State Resort Park, Kentucky. Two weighted overlay operations were performed, each utilizing three geospatial layers. Two slope layers were calculated, one using a ten-meter (10-m) DEM (14 m horizontal resolution and vertical accuracy of 0.363 ± m), and the other using LiDAR data (0.68 m or better horizontal resolution and vertical accuracy of 15.0 centimeters). The creation of two slope layers, to be used in separate weighted overlay operations, allowed for resolution comparisons and assessments of effectiveness for each resolution. Slope layers were overlaid on a distance to streams layer and a distance to cave features layer to classify the probability of an area previously experiencing cave collapse. Two areas with the highest probability were identified. The first area occurred along the upper valley wall that represents the transitions between the overlying siliciclastic rocks and the underlying limestones. The locations coincide with the highest level of caves identified in the area. The second area occurred at the transition to the next lower cave level where steep walls along the stream are observed. The higher probability levels provide a distinction between the surface representation of the cave levels, which would be consistent with areas of collapse. Distinctions between cave levels was only identifiable when using the weighted overlay which utilized the slope layer calculated with LiDAR data. The lower resolution weighted overlay (utilizing slope calculated by 10-m DEM) provided more generalized results. Based on this, LiDAR data provided greater definition of assigned probability levels, identifying areas that the 10-m DEM did not define.