Landslide susceptibility analysis using decision tree method, Phidim, Eastern Nepal

Abstract

The decision tree is one of the new methods used for the determination of landslide susceptibility in the study area. The Phidim area is selected for the application of this method. The total surface area is 168.07 sq. km, and is located at the eastern part of Nepal. There are total of 10 different data bases used for this study which are; geological formation, elevation, slope, curvature, aspect, stream power index, topographic wetness index, distance from drainage, lineaments, and slope length, and are considered as landslide conditioning factors. Geographical information system (GIS) is used as basic tools and ARC/View is used for the processing data analysis and final map preparation. For the decision tree analysis the PASW 18 (statistical tool) is used to generate values of each factor. According to the results of decision tree, two geological formations; stream power index and slope are found as the most effective parameters on the landslide occurrence in the study area. Using the predicted values, the landslide susceptibility map of the study area is produced. To assess the performance of the produced susceptibility map, the area under curve (AUC) is drawn. The AUC value of the produced landslide susceptibility map has been obtained as 95.9%. According to the results of the AUC evaluation, the produced map has showed a good performance. As to wrap up, the produced map is able to be used for medium scaled and regional planning purposes. DOI: http://dx.doi.org/10.3126/bdg.v15i0.7419 Bulletin of the Department of Geology, Vol. 15, 2012, pp. 69-76

Multi-Criteria Analysis Applied to Landslide Susceptibility Mapping

“…When compared to the results achieved in similar studies, the AUC value obtained for Porto Alegre can be considered high (e.g. Suh et al 2011;Akgün 2012;Poudyal 2012;Guillard and Zêzere, 2012). It is believed that the high AUC value obtained is mainly due to the support of local experts in all steps of the study, accompanied by a careful analysis of the set of criteria.…”

Landslide susceptibility mapping in eastern hills of Rara Lake, western Nepal

“…Paper Accepted: 06 February 2016 There are ample of studies (Dangol 2002, Ghimire 2010, Kayastha et al 2012, 2013, Poudyal 2012) on landslide hazard assessment in Nepal Himalaya. Most of the researches were focused on central and eastern Nepal.…”

“…In portions having high and very high susceptible areas, the role of drainage is most important in determining the susceptibility (Poudyal 2012). Nearer the stream higher the weight values and farther the stream, the weight values decreases.…”

Landslide Susceptibility Assessment of a Part of the Western Ghats (India) Employing the AHP and F-AHP Models and Comparison with Existing Susceptibility Maps

“…Worldwide, researchers employed methods such as the AHP [14,15], F-AHP [15], fuzzy logic [16], frequency ratio [17], analytical network process [18], FROC [19], and Mora-Vahrson-Mora (MVM) [20,21] for the mapping of landslide-susceptible zones. Researchers also applied artificial intelligence (AI)/machine learning (ML) models such as support vector machine (SVM) [22][23][24], Naïve Bayes (NB) [25][26][27], decision tree [28][29][30], K-nearest neighbor [31,32], random forest (RF) [33,34], adaptive neuro-fuzzy inference system (ANFIS) [35], convolutional neural network (CNN) [36][37][38], artificial neural network [39,40], logistic regression [41][42][43], support vector regression [44,45], recurrent neural network [36,37], Adaptive Boosting [46], extreme gradient boosting [35], Random Subspace (RSS) [47], Reduced Error Pruning Tree (REPTree) [48], etc., for landslide susceptibility modeling. The mixed ensemble models of ML techniques such as multi-layer perceptron [49], ANFIS [50], genetic algorithm [51], RSS…”