Application of geographical information system (GIS) using artificial neural networks (ANN) for landslide study in Langat Basin, Selangor

Abstract

The landslide was recognized as the most common geologic hazard around the world. The assessment of the relationship landslide conditioning factors is a critical step in managing landslide hazards and risks. Several models have been made to develop the landslide model in recent years. The Artificial Neural Networks (ANN) model was used in this study to develop a landslide model and to identify the most important landslide conditioning factors. Eight conditioning factors, including elevation, slope, aspect, curvature, lithology, soil series, Topographic Wetness Index (TWI), and rainfall, were selected and analyzed using the Geographical Information System (GIS) approach. The multilayer perceptron module and one hidden layer method extracted weighted conditioning factors. The landslide model was validated using the area under the curve (AUC) method. This model validation showed a success rate for training and testing is 0.876, respectively. This study found curvature is the most crucial factor affecting landslide occurrence in the Langat Basin with a 0.213 weight index, followed by rainfall (0.143) and elevation (0.141). Finally, the landslide model can be used as an indicator to identify the most important landslide conditioning factors and assess the relationship between these factors and landslide occurrences.