COMPARISON OF MANUAL AND SICCM SEMI-AUTOMATED MODES TO CONSTRUCT LANDSLIDE INVENTORIES ALONG THE BUFFALO NATIONAL RIVER OF ARKANSAS

The recognition of potential landslides is crucial in areas with a high frequency of mass wasting events due to physical and financial impacts on linear infrastructure. Different types of landslide hazard maps employ different approaches, each with their own advantages or disadvantages of each type of landslide. The study seeks to construct a spatial inventory of various types of landslides, regardless of age or frequency. Recently published 1:24,000 scale geologic maps along the Buffalo National River are being used as the project base maps because the area is underlain by three to nine shale units of varying thickness and plasticity. This research seeks to compile a regional inventory of pre-existing landslides, and to evaluate new sensors such as LiDAR in comparison to more traditional methods focusing on disturbed slope morphology. We are also exploring automated techniques for identifying landslides in hazard-prone areas utilizing topographic maps, aerial imagery, and LiDAR data. Traditional manual mapping uses anomalous topographic expression to identify potential landslide features, often in combination with hillshaded LiDAR images, drainage patterns, and geologic structure. The methods used in this study are intended to test the feasibility of recognizing and categorizing landslides features in a heavily vegetated area along the River using anomalous topographic signatures. The results of manual methods are then combined and compared to results derived from a workflow that uses a scarp identification and delineation tool (SI) and a landslide deposit delineation algorithm called the Contour Connection Method (CCM), which is SICCM. The SICCM can quickly and consistently approximate landslide scarps and deposits on LiDAR-DEM hillshades. These provide preliminary maps of landslide extents to assist experienced geologists in digitizing landslide inventories using protocols. The traditional manual mapping method with the semi-automatic framework has the potential to improve the efficiency of compiling landslide inventories by streamlining the mapping of landslide scarps and delineating deposits of deep-seated and translational landslides before the field reconnaissance begins. The manual topographic contour method with SICCM results should not be used as an alternative to a detailed landslide inventory, but they can provide an economic means to construct preliminary hazard maps of larger areas. Human interpretation is recommended during scarp identification.