Automatic identification of continuous or non-continuous evolution of landslides and quantification of deformations

Nowadays, the 3D modelling of surfaces is widely used to provide point clouds in geosciences. To study the evolution of landslides, many point clouds are available but post-processing can be very computationally demanding, especially to analyze their dynamics. This paper proposes automatic tools to identify areas with continuous motion and those with non-continuous motion using a new continuous motion indicator. Displacements of continuously moving areas are computed by two “displacements” methods (the image correlation method and the Gefolki differential optical flow method), which are compared. The evolution of non-continuous motion areas is quantified by a distance metric. The developed method is applied on two sites with different deformation mechanisms and evolution speeds to test the ability to show the 3D spatial evolution of landslides. The comparison with instrumental data shows a good concordance between the calculated data and the reference data. The Gefolki differential optical flow method is considered as a relevant option for the image correlation method with the new continuous motion indicator.