Please use this identifier to cite or link to this item:
Title: Extraction of multi-scale landslide morphological features based on local Gi* using airborne LiDAR-derived DEM
Authors: Shi, W 
Deng, S 
Xu, W
Keywords: Landslide
Local spatial pattern
Morphological feature extraction
Issue Date: 2018
Publisher: Elsevier B.V.
Source: Geomorphology, 2018, v. 303, p. 229-242 How to cite?
Journal: Geomorphology 
Abstract: For automatic landslide detection, landslide morphological features should be quantitatively expressed and extracted. High-resolution Digital Elevation Models (DEMs) derived from airborne Light Detection and Ranging (LiDAR) data allow fine-scale morphological features to be extracted, but noise in DEMs influences morphological feature extraction, and the multi-scale nature of landslide features should be considered. This paper proposes a method to extract landslide morphological features characterized by homogeneous spatial patterns. Both profile and tangential curvature are utilized to quantify land surface morphology, and a local Gi* statistic is calculated for each cell to identify significant patterns of clustering of similar morphometric values. The method was tested on both synthetic surfaces simulating natural terrain and airborne LiDAR data acquired over an area dominated by shallow debris slides and flows. The test results of the synthetic data indicate that the concave and convex morphologies of the simulated terrain features at different scales and distinctness could be recognized using the proposed method, even when random noise was added to the synthetic data. In the test area, cells with large local Gi* values were extracted at a specified significance level from the profile and the tangential curvature image generated from the LiDAR-derived 1-m DEM. The morphologies of landslide main scarps, source areas and trails were clearly indicated, and the morphological features were represented by clusters of extracted cells. A comparison with the morphological feature extraction method based on curvature thresholds proved the proposed method's robustness to DEM noise. When verified against a landslide inventory, the morphological features of almost all recent (< 5 years) landslides and approximately 35% of historical (> 10 years) landslides were extracted. This finding indicates that the proposed method can facilitate landslide detection, although the cell clusters extracted from curvature images should be filtered using a filtering strategy based on supplementary information provided by expert knowledge or other data sources.
EISSN: 0169-555X
DOI: 10.1016/j.geomorph.2017.12.005
Appears in Collections:Journal/Magazine Article

View full-text via PolyU eLinks SFX Query
Show full item record

Page view(s)

Last Week
Last month
Citations as of Oct 14, 2018

Google ScholarTM



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.