A new algorithm for landslide geometric and deformation analysis supported by digital elevation models

Abstract

Geometric analysis of a landslide boundary, in particular, automatic determination of the length and width of landslide and classification is a challenge. In this regard, developing an integrated automatic algorithm to determine and measure length, width, area, failure flow direction, mass displacement material, and to classify a landslide, all at one time seem to be a useful method for updating landslide inventory with reliable outcomes and efficient time for disaster management. This study presents a new automatic mapping and modelling algorithm for landslide geometric analysis include calculating landslide displacement and failure flow direction. We utilized LiDAR high resolution digital elevation model (DEM) (5 m), ASTER DEM (30 m), and Unmanned Aerial Vehicle (UAV) associated with ground truth observations to support the geometric deformation measurements. This study aims to refurbish generating landslide inventory dataset of 2015 by implementing the proposed algorithm in a quicker time than existing and traditional methods. The proposed algorithm is scripted in MATLAB based on the DEMs of before and after a landslide. The proposed new automatic method contributes measure, determine, and calculate (a) length, width, area, (b) the flow direction of the material movement, (c) the volume of the material displacement after the onset of failure, and (d) type of a landslide, in an acceptable accuracy performance. I considered two study areas (1) Alborz Mountain of Iran and (2) Madaling of Guizhou Province in China. The proposed algorithm was validated by (a) the ground truth observations, (b) the existing inventory dataset and (c) implementing the same data in ArcGIS 10.4 to compute the relative measurement errors. The relative error for area, length, width, and volume is 0.16%, 1.67%, 0.30%, 5.50%, respectively.