Este ítem es privado
Modelización de la susceptibilidad a los movimientos en masa en el entorno del tramo Bogotá – Villavicencio de la ruta 40 (Colombia) mediante técnicas geomáticas y de aprendizaje automático.
Archivos
NO SE HA AUTORIZADO la consulta de los documentos asociados
Fecha
2024-01-08
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Jaén: Universidad de Jaén
Resumen
[ES] El objetivo de este trabajo fue la modelización a la susceptibilidad de cinco tipologías de movimientos en
masa (avalanchas, derrubios, flujos, deslizamientos, reptaciones) en un área de 746 Km2 del entorno de la
carretera Bogotá – Villavicencio (Ruta 40), mediante métodos convencionales, estadísticos y de
aprendizaje automático. Para este fin se realizó un inventario de 2506 movimientos clasificados por
tipologías y se seleccionaron 4 factores de 10 analizados (índice de posición topográfico, altitud, litología y
pendiente) mediante análisis de correlación cruzado. Estos factores junto al inventario fueron utilizados
para elaborar modelos de susceptibilidad mediante 9 métodos, que se validaron mediante el área bajo la
curva – ROC y la matriz de confusión. Tanto el modelamiento como la validación se implementó mediante
los softwares SAGA, QGIS y R. A partir de estos análisis, los métodos evaluación multicriterio, regresión
múltiple, análisis discriminante, bosques aleatorios, máquinas de soporte y redes neuronales arrojaron
resultados confiables de los modelos de susceptibilidad analizados.
[EN] The aim of this work was to model the susceptibility of five types of landslides (debris avalanche, debris flow, flows, landslides, creep) in an area of 746 Km2 , around the Bogotá to Villavicencio highway. The modelling was performed using conventional, statistical and machine-learning methods. For this purpose, an inventory of 2506 movements classified into five typologies was carried out and four of ten factors (topographic position index, altitude, lithology, and slope) were selected by means of cross-correlation analysis. These input data were used for the modelling with 9 methods, whose results were validated by the area under the curve (Receiver Operating Characteristic) and the confusion matrix. Both the modelling and the validation were carried out using the SAGA, QGIS and R software. From these analyses, the multicriteria evaluation, multiple regression, discriminant analysis, random forests, support machines and neural network methods allow obtaining reliable results of the susceptibility models analyzed.
[EN] The aim of this work was to model the susceptibility of five types of landslides (debris avalanche, debris flow, flows, landslides, creep) in an area of 746 Km2 , around the Bogotá to Villavicencio highway. The modelling was performed using conventional, statistical and machine-learning methods. For this purpose, an inventory of 2506 movements classified into five typologies was carried out and four of ten factors (topographic position index, altitude, lithology, and slope) were selected by means of cross-correlation analysis. These input data were used for the modelling with 9 methods, whose results were validated by the area under the curve (Receiver Operating Characteristic) and the confusion matrix. Both the modelling and the validation were carried out using the SAGA, QGIS and R software. From these analyses, the multicriteria evaluation, multiple regression, discriminant analysis, random forests, support machines and neural network methods allow obtaining reliable results of the susceptibility models analyzed.