Abstract Views :193 |
PDF Views:126
Authors
Affiliations
1 School of Resources and Geoscience, China University of Mining and Technology, Xuzhou-221116, CN
Source
International Journal of Earth Sciences and Engineering, Vol 9, No 6 (2016), Pagination: 2383-2389
Abstract
The aim of this study is to prepare a reliable susceptibility mapping using artificial neural networks (ANN) approach based on geographic information system (GIS) for the Daguan County, China. At first, a landslide inventory map was constructed by previous studies, aerial photographs and field surveys, and a total of 194 landslides were detected and mapped. Then, the landslide inventory was randomly split into two parts: 70% for training the models and the remaining 30% for validation purpose, respectively. In this study, 15 landslide conditioning factors were considered to assess landslide susceptibility. These factors are altitude, distance to faults, distance to rivers, distance to roads, curvature, lithology, NDVI, plan curvature, profile curvature, Rainfall, slope angle, slope aspect, SPI, STI, and TWI. Subsequently, landslide-susceptible areas were mapped using the ANN model based on landslide conditioning factors. Finally, the accuracy of the landslide susceptibility maps produced from the ANN model was verified by using receiver operating characteristic (ROC) curve in MATLAB software. The areas under the curve (AUC) plot estimation results showed that the susceptibility map using ANN model has the training accuracy of 90.32%, and the prediction accuracy was 88.45%. Thus, the model can be used to reduce landslides and to select site.
Keywords
Landslide, Susceptibility, Artificial Neural Networks (ANN), MATLAB.
Full Text