Landslide causative factors evaluation using GIS in the tectonically active Glafkos River area, northwestern Peloponnese, Greece

Landslide events are a common geohazard and cause significant damage to urban facilities. Understanding landslides' effects involves determining the relationship between landslides and their causative factors. In many cases, this is not well defined. The present study aims to identify the relationship of causative factors with landslide activity. The tectonically active Glafkos River area in northwestern Peloponnese, in Greece, is the case study. This area had seriously suffered from landslides in the past. Lithology, distance from tectonic discontinuities, slope angle, and distance from drainage network are considered causative factors and categorized into different classes. The mapped landslide events of the study area at random divided into two subsets: a calibration and a validation set. The causative factors related to the calibration set by using Geographical Information Systems and statistical spatial analysis. The landslide density and the density ratio are calculated for each causative factor. The outcomes of the statistical analysis are validated using the validation set. The size of the landslide area has the highest value in limestones, and it is increased within 0 to 50 m, approximately, from tectonic discontinuities, while decreasing in distances beyond 50 m from them. The maximum landslide density values are observed on steep slopes (>30(degrees)), while it reduces as the distance from streams increase. The causative factors which have strong relationship with landslide occurrences are flysch and chert formations, distances of 0100 m from tectonic discontinuities, steep slopesand 0-200 m distance from streams. The closer the distance to tectonic elements and streams, the higher the relationship with landslide activity. The density ratio values of the above-mentioned causative factors have with correlated each other. The results prove that chert formations and the area closer to streams are more prone to landslides. According to the validation procedures, the accuracy of the proposed method in detecting landslide phenomena ranged from 76.5 % to 79.5 %. The proposed statistical spatial analysis proves to be efficient in examining similar landslide prone areas, and thus to proper spatial planning and landslide protection measures.