Landslide susceptibility mapping by geographical information system-based multivariate statistical and deterministic models: in an artificial reservoir area at Northern Turkey


AKGÜN A. , ERKAN O.

ARABIAN JOURNAL OF GEOSCIENCES, cilt.9, sa.2, 2016 (SCI İndekslerine Giren Dergi) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 9 Konu: 2
  • Basım Tarihi: 2016
  • Doi Numarası: 10.1007/s12517-015-2142-7
  • Dergi Adı: ARABIAN JOURNAL OF GEOSCIENCES

Özet

In Turkey, landslide phenomenon is one of the most important natural hazards. Due to landslide occurrence, several landforms and man-made structures are adversely affected and may cause many injuries and loss of life. In this context, landslide susceptibility assessment is an important task to determine susceptible areas to landslide occurrence. Especially, several dam reservoir areas in Turkey are threatened by landslide phenomena. For this reason, in this study, a dam reservoir area, located in the northern part of Turkey, was selected and investigated in the point of view of landslide susceptibility assessment. A landslide susceptibility assessment for Kurtun Dam reservoir area (Gumushane, North Turkey) was carried out by geographical information system (GIS)-based statistical and deterministic models. For this purpose, logistic regression (LR) and stability index mapping (SINMAP) methodologies were applied. In this context, eight contributing factors such as altitude, lithology, slope gradient, slope aspect, distance to drainage, distance to lineament, stream power index (SPI) and topographical wetness index (TWI) were considered. After assessment of these parameters by LR and SINMAP methods in a GIS environment, two landslide susceptibility maps were obtained. Then, the produced maps were analyzed for validation purpose. For this purpose, area under curvature (AUC) approach was used. At the end of this process, the AUC values of 0.73 and 0.65 were found for LR and SINMAP models, respectively. For the performance of the SINMAP model, statistical results produced by the model were also considered. In this context, landslide density of the stability index (SI) classes were taken into account, and it was determined that 89.5 % of the landslides fall into lower and upper threshold classes which almost correspond to moderate and high susceptibility classes. These two validation values indicate that the accuracy of landslide susceptibility maps is acceptable, and the maps are feasible for further natural hazard management affairs in the area.