The ability for timely evaluation of sudden increases in ground moisture levels would be a valuable tool in reliable assessment of rainfall triggered landslide risk. Surface soil moisture estimated based on satellite images would be vital in such evaluations. In this study, three alternative stochastic classification models, logistic regression, decision tree and bagged tree have been developed to identify locations of high landslide risk based on site attributes of geology, soil type, slope, land cover and the corresponding satellite based soil moisture estimates. As opposed to the commonly used validation set approach, in this work, cross validation was employed to improve the prediction accuracy of the models. It was seen that all three classification models provided reasonably accurate predictions. It is expected that the findings of this research would lay the groundwork for the future formulation of a timely, reliable and effective method for landslide hazard prediction.
CITATION STYLE
Muñoz, E., Poveda, G., Ochoa, A., & Caballero, H. (2017). Multifractal Analysis of Spatial and Temporal Distributions of Landslides in Colombia. In Advancing Culture of Living with Landslides (pp. 1073–1079). Springer International Publishing. https://doi.org/10.1007/978-3-319-53498-5_122
Mendeley helps you to discover research relevant for your work.