Yapay sinir ağı ile heyelan duyarlılığının haritalandırılması: Artvin örneği

Aydin O., Ircan M. R., Duman N.

I. Ulusal Samsun Coğrafya Sempozyumu, Samsun, Türkiye, 2 - 04 Nisan 2026, cilt.1, ss.46, (Özet Bildiri)

  • Yayın Türü: Bildiri / Özet Bildiri
  • Cilt numarası: 1
  • Basıldığı Şehir: Samsun
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.46
  • Ankara Üniversitesi Adresli: Evet

Özet

Landslides are important natural hazards that cause loss of life and property in mountainous areas. Because slope instability is determined by the

nonlinear interactions of topography, hydrology and geology, accurate mapping of susceptibility is difficult. In areas such as Artvin, where slope

is high, the drainage network is dense and lithology is heterogeneous, traditional approaches often remain limited. Therefore, artificial

intelligence based methods that can evaluate different conditioning factors together offer a practical alternative. The aim of this study is to reveal

the capacity of a feed forward multilayer perceptron artificial neural network model, developed based on elevation, slope, aspect, distance to

river, Topographic Wetness Index (TWI), precipitation, land use/land cover, soil type and geology and lithology variables, to map landslide

susceptibility in the Artvin study area. The discriminative ability of the model on an independent test set was evaluated with the ROC–AUC value

and was calculated as 0.8054. The trained model was applied to the entire study area to produce a continuous susceptibility index in the range 0 to

1 for each pixel, and the output was mapped by converting it into five susceptibility classes. The results show that the probability based

susceptibility surface provides a strong and interpretable output in terms of risk ranking and decision support.