INVESTIGATING AUTOMATED MAPPING TECHNIQUES TO IDENTIFY LANDFORM ASSEMBLAGES IN WESTERN NORTH CAROLINA

Geomorphological terrain classification for landslide hazard assessment is traditionally performed by manually mapping the contacts and boundaries of landforms using a LiDAR DEM, orthophotography and contour lines. Our study intends to investigate the use of GIS geoprocessing tools in automating the mapping of landform assemblages and the efficacy of capturing relationships with larger, hazardous landslides. Andy Cove in Pisgah National Forest was selected as the field area because it contains four landslides that triggered within similar landform classes due to the 1916 storm of record and are among the largest failures recorded in the NCGS database. The automated mapping approach uses a high-resolution LiDAR DEM (0.5m cell size) as input for morphometric analysis. Landform classes were determined by parametric sensitivity analysis of topographic position index (TPI), slope angle, curvature, and elevation classes. Initial results of the automated mapping approach were further refined using comparative iteration with manual mapping criteria. Conflating slope angle and TPI rasters yielded the most favorable geomorphological representation and terrain classification. Ranges of morphometric parameters that are unique to a specific landform class were determined by analyzing the overlap between the manual and automated mapping results. The automated approach identified six major landform classes within the watershed and the four landslides triggered at similar positions in the landscape. Ultimately, these automated methods will be implemented across western North Carolina for a region-wide landslide hazard assessment focused on capturing occurrences of similar geologic conditions, modes/mechanism of slope failure and runout to the four large landslides in the study area.