A Data Based Model to Predict Landslide Induced by Rainfall in Rio de Janeiro City

Abstract

Landslide prediction is complex and involves many factors, such as geotechnical, geological, topographical, and even meteorological. This work presents a methodology by using a Data Mining approach in order to predict landslide occurrences induced by rainfall in Rio de Janeiro city. Landslide and rain data records from 1998 to 2001 were obtained from field technical reports and 30 automatic rain gauges, respectively. It was also collected data regarding soil parameters, including urban areas, forest, vulnerability, among others, and totalizing 46 soil variables. All the information was inserted into a Geographic Information Systems. Clustering (Dendrogram and k-means) andStatistical (Principal Component Analysis and Correlation) techniques were used to regionalize the rain data and select the rain gauges to be input on Artificial Neural Networks , which were used to replace the missing rain values. The landslide volume variable also presented missing values and it was completed by the k-Nearest Neighbor method. After data preparation, some models were built to predict landslide and rainfall using Data Mining techniques. The obtained model's performance is also analyzed.

A data-based model for predicting wildfires in Chapada das Mesas National Park in the State of Maranhão

“…The main results from a classifier are the confusion matrix and correct classification rate (CCR) or accuracy, Souza and Ebecken (2012). Table 5 shows the confusion matrix after validation of the ANN model.…”

“…A model to predict landslides induced by heavy rains in Rio de Janeiro, which is a more complex phenomenon, was developed using a history of only 4 years of data (Souza and Ebecken 2012). Another model to locate mass movements induced by earthquakes in Sichuan (China), considered an interval smaller than 4 months to train the neural network (Souza 2014).…”

Displacement prediction of landslide based on generalized regression neural networks with K-fold cross-validation

“…Currently, a large number of methods have been widely used for predicting the displacement of landslide [8][9][10][11][12][13][14][15][16], such as statistical, artificial intelligent methods, linear or multiple regression. A linear combination model with optimal weight is applied to landslide displacement prediction [8].…”

Deformation prediction of landslide based on functional network

“…At present, to solve such cutting-edge scientific issue like the landslide forecast and prediction has become a hot topic in landslide research [9][10][11][12][13][14][15]. Melchiorre et al [9] demonstrate that the landslide susceptibility analysis is performed by means of ANNs and cluster analysis.…”

“…Li et al [11] present a methodology by an application of linear combination model with optimal weight in landslide displacement prediction. A data mining approach to predict landslide are adopted by de Souza and Ebecken [12]. Sezera et al [13] present that the neuro-fuzzy model using remote sensing data and Geographic Information System (GIS) for landslide susceptibility analysis.…”