Influence of Seismic Displacement Models on Landslide Prediction at the Egkremnoi Coastline During the 2015 Lefkada Earthquake

Abstract

Strong earthquakes can trigger thousands of landslides in mountainous areas, and accurate prediction of landslide occurrence can reduce the risk of infrastructure and communities to earthquake-induced regional landslides. However, current methods for regional landslide prediction are still in their infancy and prediction of landslide occurrence remains a challenge. In this study, a pseudo-three-dimensional (pseudo-3D) procedure was implemented to a 1.35 km² area along the Egkremnoi coastline of the Lefkada island in Greece to assess the triggering of landslides and their geometry. To better understand the influence of permanent seismic displacement models on landslide prediction, seven models were employed in the prediction procedure. The adopted displacement models include four Newmark-type rigid block models and three flexible models. The results show that in this specific case and for the input parameters used in the analyses, the Bray and Macedo model achieves the best performance in landslide prediction in terms of the percentage of correctly predicted landslides and the centroid distance between mapped and predicted landslides, which are used to quantify the location accuracy of predicted landslides. The Rathje and Antonakos model predicts more landslides that overlap with mapped landslides, but also overpredicts more overall.