Landslide recognition and mapping in a mixed forest environment from airborne LiDAR data

Görüm T.

ENGINEERING GEOLOGY, cilt.258, 2019 (SCI İndekslerine Giren Dergi) identifier identifier


A precise, accurate and complete landslide inventory is indispensable for the establishment of reliable landslide susceptibility and hazard maps. In the preparation of landslide inventories, dense vegetation cover is the major obstacle that confounds the topographic signature of landslides. Today, the growing usage of light detection and ranging (LiDAR) technology in the field of geoscience illuminates the mystery of the landscapes cloaked in dense vegetation by providing a new visual acuity to researchers. Turkey, similar to many mid-latitude mountainous countries suffering from landslides, is still continuing landslide inventory mapping with conventional methods such as aerial photo interpretations (API), although 68% of landslide events occurred under dense forest cover in the north of the country. Despite the country-wide medium-scaled landslide catalogs, the number and the abundance of landslides covered by forests remain largely unknown regardless of the increasing availability of high-resolution remote-sensing data. From these motivating insights and drawbacks, the study's focus is to evaluate the capability of mapping landslides by visually analyzing airborne LiDAR DTM derivatives and compare the results with the 1:25,000 scaled API-based inventory to understand the potential contribution of LiDAR technology in Europe's deadliest country (Turkey) in terms of landslides. The landslide mapping results for a test area located in the densely forested Ulus Basin, Western Black Sea region, reveal that the extent and the number of the mapped landslides (n = 902) from the airborne LiDAR data are much higher than those of the available API-based landslide inventory, which includes 67 landslides. Comparative analysis on topographic signatures of landslides is also underpinned on a distinct discrepancy of slope failure diagnostic indicators captured by the two available DEMs derived from LiDAR and 1:25,000 scaled topographic maps, emphasizing the success rate of LiDAR-derived DEM in recognition and mapping landslides with higher precision and accuracy. Together with all analysis results, the landslide recognizability degree assessment based on the forest cover percentage and slope height differences of landslides highlights that the MAR data enable landslides to be defined up to 100 m(2), while this value is 20 times lower than that of estimated values from API-based landslides. In addition to underlining the contribution of LiDAR technology to the recognition of landslides hidden under dense vegetation, these findings stress the importance of LiDAR data on complete and accurate landslide inventories for production of reliable susceptibility and hazard maps and further a better understanding of the landslide processes and reducing related losses.