Citation

Abdollahizad S, Balafar MA, Feizizadeh B, Babazadeh Sangar A, Samadzamini K. Appl. Geomat. 2023; 15(1): 109-125.

Copyright

(Copyright © 2023, Holtzbrinck Springer-Nature)

DOI

10.1007/s12518-023-00488-w

PMID

unavailable

Abstract

Mapping of landslide susceptibility is an important tool to prevent and control landslide disasters for a variety of applications, such as land use management plans. The main objective of this study was to propose an application of artificial intelligence systems, then evaluate and compare their efficiency for developing accurate landslide susceptibility mapping (LSM). The present study aims to explore and compare the frequency ratio (FR) method with three machine learning (ML) techniques, namely, random forests (RF), support vector machines (SVM), and multiple layer neural networks (MLP), for landslide susceptibility assessment in East Azerbaijan, Iran. To achieve this goal, 20 landslide-occurrence-related influencing factors were considered. A sum of 766 locations with landslide inventory was recognized in the context of the study, and the relief-F method was utilized in order to measure the conditioning factors' prediction capacity in landslide models. In the forthcoming phase, three ML models (SVM, RF, and MLP) were trained by the training dataset. Lastly, the receiver operating characteristic (ROC) and statistical procedures were employed to validate and contrast the predictive capability of the FR model with the obtained three models. The findings of the study in terms of the relief-F method for the importance ranking of conditioning factors in the context area uncovered those eleven factors, such as slope, aspect, normalized difference vegetation index (NDVI), and elevation, have the highest impact on the occurrence of the landslide. The results show that the MLP model had the utmost rate of landslide spatial prediction capability (87.06%), after which the SVM model (80.0%), the RF model (76.67%), and the FR model (61.25%) demonstrated the second, third, and fourth rates. Besides, the study revealed that benefiting the optimal machine with the proper selection of the techniques could facilitate landslide susceptibility modeling.

Language: en

Keywords

Landslide susceptibility; Multiple layer neural network; Random forest; Support vector machine