Application of LiDAR and discriminant analysis to determine landscape characteristics for different types of slope failures in heavily vegetated, steep terrain: Horseshoe Run watershed, West Virginia

作者：Konsoer Kory M^*; Kite J Steven

来源：Geomorphology, 2014, 224: 192-202.

DOI：10.1016/j.geomorph.2014.06.030

• Landslide inventory maps
• LiDAR
• Discriminant analysis
• Natural hazards
• Slope failure assessment

A landslide inventory map was created for Horseshoe Run watershed, West Virginia, using high-resolution topographic data obtained from airborne LiDAR. A total of 152 landslides were remotely mapped and classified as planar slides, rotational slumps, debris flows, debris fans, debris slides, and active slopes based on morphologic characteristics interpreted from LiDAR-derived shaded relief maps, with field verification at several locations. Seven landscape variables were calculated for the watershed from the high-resolution (0.5-m) topographic data: elevation, slope angle, slope aspect, distance from roads, distance from streams, plan curvature, and profile curvature. Statistical analyses were conducted to determine the mean and standard deviations for each variable and corresponding slope failure type. Discriminant analyses were conducted on the different types of failed slopes, with a maximum 86% accuracy level, to determine which landscape characteristics contribute to each type of landslide. Results from the discriminant analyses indicate that statistically significant differences exist between the failure classes mapped within the watershed and that slope angle, slope aspect, and landscape curvature are important variables influencing the type of process. The results from this study have important implications for predicting where certain types of slope failures may occur based on landscape characteristics. Published by Elsevier B.V.

• 出版日期2014-11-1