A SEMI-AUTOMATED KARST FEATURE PREDICTION MODEL FOR THE WHITE RIVER NATIONAL FOREST, COLORADO BASED ON BARE-EARTH LIDAR IMAGERY

Investigation into surface karst formation is significant to hazard prediction, hydrogeologic drainage, and land management. The White River National Forest in central Colorado contains over 377,000 acres of mapped karst-forming bedrock including limestone, dolomite, and gypsum. The objective of this project is to develop a semi-automated model to map and delineate surface karst features within the White River NF from publicly available, 3-ft resolution bare-earth LiDAR imagery using ArcGIS Pro 2.7.3. A semi-automated approach of mapping karst features provides a dataset that minimizes error from noise while maintaining accurate depression location and catchment boundaries. Several semi-automated models with different size parameters were used to determine the ideal minimum area and depth thresholds to filter out erroneous “noise” inherently found in DEMs while retaining reliable prediction and accurate measurement of surface karst features. These models were compared and verified in-field during the summer of 2021 with two other GeoCorps participants and various Forest Service employees and volunteers. Some limitations of this project include a relatively short time frame for manual editing and field verification, numerous karst-forming rock types with variations in feature morphology, and imperfect bedrock mapping of karst-forming rock. The polygon layer resulting from this project can be used to avoid karst hazards during wildfire mitigation, protect sensitive bat habitats from land disturbance, and eventually tie together complex surface and subsurface relationships between hydrologic insurgences, springs, and cave systems in the region.