Hydrology
Elham EbrahimZadeh; Ebrahim Rahimi; Vahid Gagheri
Abstract
Identifying landslide-prone areas provides a basis for slope-stabilization and mitigation programs. In Hablehroud watershed, artificial neural network and fuzzy logic (FL) as one of the methods of multicriteria-decision analysis based on ArcGIS were used in the scientific evaluation of landslide-prone ...
Read More
Identifying landslide-prone areas provides a basis for slope-stabilization and mitigation programs. In Hablehroud watershed, artificial neural network and fuzzy logic (FL) as one of the methods of multicriteria-decision analysis based on ArcGIS were used in the scientific evaluation of landslide-prone areas. For this purpose, MATLAB, IDRISI and ArcGIS software were used. After preparing landslide-susceptibility maps, the prone zones predicted by FL and multilayer perceptron artificial neural network (MLP-ANN) were compared with the Hablehroud landslide database (distribution map). The results indicate a good overlap between the prone zones predicted by the MLP-ANN and landslide field observations. Finally, the performance of different methods in generating landslide-susceptibility maps were compared to each other using the validation indicators of "quality-sum index (Qs)" and "receiver-operating-characteristic curve (ROC)" to specify the optimal and applicable method for the landslide risk management of the Hablehroud watershed. By analyzing the obtained zoning maps and considering the Qs and "area-under curve (AUC)" values of different FL operators and MLP-ANN for the landslide-susceptibility maps, it is observed that the Qs (1.6299) and AUC (0.806–very good) values of the MLP-ANN are higher than those calculated for the sensitivity maps by different FL operators.
