Paper No. 14
Presentation Time: 12:30 PM

EVALUATION OF DIGITAL TERRAIN AND OTHER DATA FOR IDENTIFYING SANDSTONE CAVE LOCATIONS


FISSEL, Lauren E., Department of Geosciences, University of Akron, MS-4101, Akron, OH 44325-4101, SASOWSKY, Ira D., Dept. of Geosciences, The University of Akron, Akron, OH 44325-4101, MCQUADE, Theresa L., 4181 Appian Way West, Apt. E, Gahanna, OH 43230, BUTLER, Kevin, ESRI, Inc, 380 New York Street, Redlands, CA 92373 and CURTIS, Robert L., Metro Parks, Serving Summit County, 975 Treaty Line Road, Akron, OH 44313, lef23@zips.uakron.edu

Most significant caves are developed in limestone, but in areas where this rock is absent, small caves developed in other lithologies can be important. In northeastern Ohio carbonate rocks are not common, and numerous caves are found in the Berea Sandstone (Mississippian) and the Sharon sandstone/conglomerate (Pennsylvanian). About 70 caves are known from a 4 county region, but it is expected that many others are unreported/undiscovered. The caves serve as habitat for a variety of regionally rare organisms, including endangered bat species. Because of this, the identification of as-yet unknown caves is critical. We developed numerous GIS representations/comparisons with the goal of using them to identify common geomorphologic/topographic indicators of sandstone caves. Three types of terrain data were compared (30 m DEM, LIDAR, and DLG) for two sandstone caves in Summit County, Ohio (Mary Campbell Cave and Icebox Cave). There is an association of longer caves with springs, low-order streams and small cliffs. The LIDAR data allow recognition of fine-scale topographic indicators of caves that are obscured or absent in the DEM or DLG data. This product, however, is much more bulky. File sizes for a 0.18 km-sq area vary from 8 kb to 3.5 Mb between DLG and LIDAR. A logistic regression approach was also employed. Values for ten ecological variables were extracted from GIS layers at each of the cave locations. Overall the model converged and based on the Hosmer and Lemeshow test, fit the data well (χ2=10.07, df=8, p=.260). Using the ten explanatory variables, the predictive power of the model is high with 87.4% of the cases correctly classified. Cave presence was positively associated with higher slope, eastwardly facing slopes, sandstone bedrock, and higher bedrock elevations. Cave presence was negatively associated with deeper drift thickness and higher elevations in the study area. An ongoing goal of the project is to evaluate algorithms that can be applied using synthetic vector hydrography and the terrain grid in order to automate identification of interest zones.