A comparative analysis of landslide mapping techniques using earth observation (EO) data
Publication Date
July 31, 2025
Creator
Abstract
Landslides pose significant threats to human life, infrastructure, and the environment. Rapid accurate assessment and identification of these landslides within days to weeks of their occurrence are crucial for timely disaster management and effective emergency response strategies. This study evaluated the performance of freely accessible EO-based tools - HazMapper and Google Earth Engine (GEE), alongside manual delineation technique in the detection and rapid mapping of landslides in Glengyle. Recent landslides which typically leave visible scars on the landscape were primarily considered using two approaches applied by these EO-based tools: pixel-based analysis and SLIP algorithm for change detection. The integration of these methods provides a comprehensive assessment of landslide areas as they consider both changes in vegetation cover and topography. By comparing the pre- and post-event composites of the study area, the landslides were detected based on the NDVI and landslide tracker binary raster image generated by HazMapper and GEE, while the manual technique employed the physical delineation of the landslides using satellite imagery. The performance metrics of these methods identified GEE as the best-performing method with an F1 score of 0.83 over HazMapper and manual technique’s 0.81 and 0.79 F1 scores respectively. The kappa value of 0.62 for GEE suggests that the tool’s efficacy in rapid landslide mapping performed substantially better than random chance. Replication of the GEE in the secondary (Rest and Be Thankful) and tertiary (Dortyol) study areas further assessed the tools efficacy in mapping other landslide types. With an F1 score of 0.82 and 0.93 for the secondary and tertiary locations, GEE correctly identified a substantial number of landslides in these locations better than random chance. The results suggest that GEE is a robust tool for rapid landslide mapping in emergency situations.
Item Type
ethesis
Thesis Type
MRes
Supervisors
Subjects (LC)
Associated Schools / Departments
Faculty of Engineering
eprints ID
80395
UoN Repository URI
Except where otherwise noted, this item's license is described as
File(s)![Thumbnail Image]()
Name
Lilian Akudo Akanazu_MRes Thesis.pdf
Type
Full-text
Description
Examined
Size
3.33 MB
Format
Adobe PDF
Checksum (MD5)
29bab406208f5e59b93467ba242b459a