Accuracy Analysis Comparison of Supervised Classification Methods for Anomaly Detection on Levees Using SAR Imagery

Abstract

This paper analyzes the use of a synthetic aperture radar (SAR) imagery to support levee condition assessment by detecting potential slide areas in an efficient and cost-effective manner. Levees are prone to a failure in the form of internal erosion within the earthen structure and landslides (also called slough or slump slides). If not repaired, slough slides may lead to levee failures. In this paper, we compare the accuracy of the supervised classification methods minimum distance (MD) using Euclidean and Mahalanobis distance, support vector machine (SVM), and maximum likelihood (ML), using SAR technology to detect slough slides on earthen levees. In this work, the effectiveness of the algorithms was demonstrated using quad-polarimetric L-band SAR imagery from the NASA Jet Propulsion Laboratory’s (JPL’s) uninhabited aerial vehicle synthetic aperture radar (UAVSAR). The study area is a section of the lower Mississippi River valley in the Southern USA, where earthen flood control levees are maintained by the US Army Corps of Engineers.