Akademik Veri Yönetim Sistemi
Arabian Journal of Geosciences, cilt.14, sa.20, 2021 (SCI-Expanded, Scopus)
In this study, the landslide susceptibility maps (LSMs) of the Canik, İlkadım, and Atakum districts of Samsun province of Turkey were produced. The weight of evidence (WoE) and modified information value (MIV) models were used to produce the LSMs in this study. Elevation, slope, aspect, plan and profile curvature, distance to drainage networks and fault lines, topographic wetness index, and geological characteristics were used as parameters in the susceptibility analysis. In the study area, there are 131 landslide polygons in total, 54 of which are active and 77 are passive. The training and validation data sets were produced using two different approaches. In the polygon-based method, 70% of all landslide polygons in the study area (92 landslide polygons) were used for training and the remaining 30% (39 landslide polygons) were used for validation. In the point-based method, points were produced at 10-m intervals inside each landslide polygon. Then, 70% of the points remaining within each landslide polygon were chosen randomly for training, and the remaining 30% were chosen for validation. When these data were combined, a training data set comprising 211,649 points and a validation data set comprising 90,542 points were produced. The LSMs produced by the WoE and MIV models using two different training data sets were divided into five classes. Accordingly, it was determined that the LSMs produced by the WoE model were 90.6%, which was compatible with the validation data sets in the polygon-based method and 93.2% in the point-based method. Similarly, it was observed that the compatibility of the LSMs produced by the MIV model with the validation data sets was 63.7% in the polygon-based method and 79.9% in the point-based method. Results obtained showed that the point-based method is more effective in landslide susceptibility mapping in the study area.