Evaluation of the influence of source and spatial resolution of DEMs on derivative products used in landslide mapping. Issue 6 (1st November 2016)

• Record Type: Journal Article
• Authors: Mahalingam, Rubini; Olsen, Michael J.
• Is Part Of: Geomatics, natural hazards & risk. Volume 7:Issue 6(2016)
• Page Start: 1835
• Page End: 1855
• Publication Date: 2016-11-01
• Subjects: Landslides -- LiDAR -- spatial resolution -- DEM -- Oregon; Geomatics -- Periodicals; Geomatics; Periodicals; 526.905
• DOI: 10.1080/19475705.2015.1115431
• Languages: English
• ISSNs: 1947-5705

Abstract

ABSTRACT: Landslides are a major geohazard, which result in significant human, infrastructure, and economic losses. Landslide susceptibility mapping can help communities plan and prepare for these damaging events. Digital elevation models (DEMs) are one of the most important data-sets used in landslide hazard assessment. Despite their frequent use, limited research has been completed to date on how the DEM source and spatial resolution can influence the accuracy of the produced landslide susceptibility maps. The aim of this paper is to analyse the influence of spatial resolutions and source of DEMs on landslide susceptibility mapping. For this purpose, Advanced Spaceborne Thermal Emission and Reflection (ASTER), National Elevation Dataset (NED), and Light Detection and Ranging (LiDAR) DEMs were obtained for two study sections of approximately 140 km 2 in north-west Oregon. Each DEM was resampled to 10, 30, and 50 m and slope and aspect grids were derived for each resolution. A set of nine spatial databases was constructed using geoinformation science (GIS) for each of the spatial resolution and source. Additional factors such as distance to river and fault maps were included. An analytical hierarchical process (AHP), fuzzy logic model, and likelihood ratio-AHP representing qualitative, quantitative, and hybrid landslide mapping techniques were used for generating landslide susceptibility maps. The results from each of the techniques were verified with the Cohen's kappa index, confusion matrix, and a validation index based on agreement with detailed landslide inventory maps. The spatial resolution of 10 m, derived from the LiDAR data-set showed higher predictive accuracy in all the three techniques used for producing landslide susceptibility maps. At a resolution of 10 m, the output maps based on NED and ASTER had higher misclassification compared to the LiDAR-based outputs. Further, the 30-m LiDAR output showed improved results over the 10-m NED and 10-m ASTER output, indicating that finer resolution does not necessarily result in higher predictive accuracy in landslide mapping. The source of the data-sets is an important consideration and can have significant influence on the accuracy of a landslide susceptibility analysis.