GEOMORPHIC QUANTIFICATION OF COLLUVIAL DEPOSITS IN THE INTERIOR LOW PLATEAUS USING LIDAR-DERIVED MAPS

Freely available regional lidar elevation data in Kentucky and other states make it possible to significantly improve the efficiency of surficial mapping using digital terrain modeling. We used lidar-derived digital elevation models (DEMs) and derivative maps (slope angle, curvature, and topographic roughness) to identify colluvium-covered slopes and compared those results to the extent of colluvium on published surficial geologic maps developed using more traditional fieldwork-intensive methods. In this case study, we focused on the DeMossville and Valley Station, Kentucky, 7.5-minute quadrangles. To determine the accuracy of our DEM-based approach, we (1) generated 1,000 random points for each quadrangle and extracted slope angle, curvature, and roughness values for each point and (2) assigned a true/false tag for each extracted point based on the extent of colluvium shown on existing surficial geologic maps of the quadrangles. The percentage of points that were correctly identified as colluvium demonstrates the success of each approach and enables comparison. The DeMossville colluvial model has an 89.1% accuracy and the Valley Station colluvial model has an 96.8% accuracy. Success rates varied with different moving window sizes and slope angle thresholds; for this study, the most successful for both quadrangles are a circular window (8-pixel radius) with a 10-degree slope threshold. Ultimately, we show that digital terrain modeling of colluvial slopes in areas of moderate topography efficiently produces reliable boundaries between colluvium and adjacent map units (typically ridge-top residuum and valley-bottom alluvium).