Share Email Print
cover

Proceedings Paper

Identification of landslides in clay terrains using Airborne Thematic Mapper (ATM) multispectral imagery
Author(s): Malcolm Whitworth; David Giles; William Murphy
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

The slopes of the Cotswolds Escarpment in the United Kingdom are mantled by extensive landslide deposits, including both relict and active features. These landslides pose a significant threat to engineering projects and have been the focus of research into the use of airborne remote sensing data sets for landslide mapping. Due to the availability of extensive ground investigation data, a test site was chosen on the slopes of the Cotswolds Escarpment above the village of Broadway, Worcestershire, United Kingdom. Daedalus Airborne Thematic Mapper (ATM) imagery was subsequently acquired by the UK Natural Environment Research Council (NERC) to provide high-resolution multispectral imagery of the Broadway site. This paper assesses the textural enhancement of ATM imagery as an image processing technique for landslide mapping at the Broadway site. Results of three kernel based textural measures, variance, mean euclidean distance (MEUC) and grey level co-occurrence matrix (GLCM) entropy are presented. Problems encountered during textural analysis, associated with the presence of dense woodland within the project area, are discussed and a solution using Principal Component Analysis (PCA) is described. Landslide features in clay dominated terrains can be identified through textural enhancement of airborne multispectral imagery. The kernel based textural measures tested in the current study were all able to enhance areas of slope instability within ATM imagery. Additionally, results from supervised classification of the combined texture-principal component dataset show that texture based image classification can accurately classify landslide regions and that by including a Principal Component image, woodland and landslide classes can be differentiated successfully during the classification process.

Paper Details

Date Published: 23 January 2002
PDF: 9 pages
Proc. SPIE 4545, Remote Sensing for Environmental Monitoring, GIS Applications, and Geology, (23 January 2002); doi: 10.1117/12.453675
Show Author Affiliations
Malcolm Whitworth, Univ. of Portsmouth (United Kingdom)
David Giles, Univ. of Portsmouth (United Kingdom)
William Murphy, Univ. of Leeds (United Kingdom)


Published in SPIE Proceedings Vol. 4545:
Remote Sensing for Environmental Monitoring, GIS Applications, and Geology
Manfred Ehlers, Editor(s)

© SPIE. Terms of Use
Back to Top