Performance of image correlation techniques for landslide displacement monitoring

Abstract

The objective of this work is to present the applicability of image correlation techniques (applied to very-high resolution terrestrial optical photographs and to very dense Terrestrial Laser Scanning (TLS) point clouds) to monitor the displacement of continuously active landslides. The method has been developed to characterize the kinematics of very active landslides with cumulated displacement of several decimeters per year. The data are processed with a cross-correlation algorithm applied on the full resolution images (photographs and DEMs produced from the TLS data) in the acquisition geometry. Then, the calculated 2D displacement field is ortho-rectified with a back projection technique. The method allows to characterize the heterogeneous displacement field of the landslide in time and space, and to produce displacement maps. The performance of the technique is assessed using as reference differential GPS surveys of a series of benchmarks. The sources of error affecting the results are discussed. Because the proposedmethodology can be routinely and automatically applied, it offers promising perspectives for operational applications like, for instance, in early warning systems. © Springer-Verlag Berlin Heidelberg 2013.