Close-Range Photogrammetry Application to Monitor Levee Erosion in Case of Wave Overtopping

Abstract

Coastal and riverine levees are prone to wave and steady-flow overtopping, which can result in breaching and consequential flooding of the hinterland. Full-scale field experiments on a levee are a sound approach to understanding of the underlying physical processes without suffering from scale effects, provided that accurate and valuable data can be collected. This paper outlines how close-range photogrammetry was applied to monitor a levee’s morphological evolution continuously and intermittently during wave overtopping. High-resolution spatial topographical models were obtained from which erosion rates could be quantified. Two camera setups—a system of multiple stationary synchronized cameras and a single mobile camera—were used for photo acquisition. The first configuration enabled reconstruction of the target zone as a dynamic scene and reduced photogrammetric analysis time through simultaneous processing of several frames [four-dimensional (4D) processing]. The second configuration applied the structure-from-motion (SfM) technique using multiple overlapping photographs to obtain three-dimensional (3D) elevation models at specific moments. Both approaches resulted in high-resolution digital elevation models (DEMs) with low mean absolute errors (MAEs) on the order of a few centimeters. Furthermore, photogrammetry proved to be highly flexible and could be applied to different types of levees with various cover layers, hydraulic loads, and meteorological conditions. Monitoring erosion due to wave overtopping of levees was achieved here using close-range photogrammetry in two series of field experiments with distinct camera setups. The experiments were conducted at 1∶1 scale at the Living Lab Hedwige Prosperpolder to study the erosion processes of the levee’s clay’s protection layer (1-m-thick clay) on the landward-side slope subjected to loads by overtopping waves. The experiments were conducted on a levee cover in (1) its natural state; and (2) after treating clay with lime for reinforcement. In the first series of experiments, a system of 10 synchronized stationary cameras was used to capture photos. To focus on the erodibility of the natural clay cover, the 0.2-m-thick grass cover was removed over an area of 14 m2 at three locations on the slope prior to testing. In the second series, a single camera was moved around the scene at specific time intervals. An object featuring a building was also embedded in each new slope revetment to evaluate the effects on erosion of flow around obstacles. Both generated accurate erosion maps, with a slightly higher accuracy and ease of post-treatment for the multiple-camera setup.