2257
Landslide Susceptibility Mapping Using Bayesian Conditional Probability Model at
An Linh Commune, Tuy An District, Phu Yen Province, Vietnam
Élaboration de la carte de risques de glissement de terrain sur la commune de An Linh, district de
Tuy An, province de Phu Yen, à l’aide d’un modèle Bayesien de probabilité conditionnelle
Son N.T.
Tuy Hoa Industrial College, Phu Yen Province, Vietnam
Ha P.T.S.
Facutly of Geology and Petroleum Engineering, Ho Chi Minh City University of Technology, Ho Chi Minh City, Vietnam
Son L.M.
flatGEO Consulting Co., Ltd.
ABSTRACT: Landslide phenomenon at An Linh commune, Tuy An district, Phu Yen province is more and more severe year after
year. The prediction of a highly landslide susceptibility zone will support the government and local residents to save their properties.
In this paper the authors carried out the landslide susceptibility mapping at An Linh commnune using the deterministic approach with
knowledge-driven methods and the stochastic approach with data-driven methods. The deterministic approach is conducted using
traditional survey, field investigation, geophysical survey… to evaluate not only geotechnical properties of rock and soil materials but
also the mechanism of landslide phenomenon in the studied area. The stochastic approach with probability methods is chosen to
evaluate the weights of influence factors such as the elevation, aspect, slope angle, distance to roads and distance to drainages,
geology and topographical wetness index of the materials. Landslide susceptibility zonation is developed using the model of Bayesian
conditional probability with different criteria for weighting factor calculation integrated with GIS analytical tools. Landslide
susceptibility maps from different models are validated with the current landslide inventory to select the most reasonable model for
the study area.
RÉSUMÉ : Les glissements de terrain sont devenus de plus en plus fréquents sur la commune de An linh, district de Tuy An, province
de Phu Yen, au cours des dernìeres années. Les auteurs ont réalisé une carte de risques de glissement de terrain sur la commune An
Linh selon une approche déterministe fondée sur une connaissance experte et à l’aide d’une approche stochastique basée sur les
données. L’approche déterministe est conduite à partir des méthodes de reconnaissance, d’observation du terrain, de prospection
géophysique… afin de quantifier non seulement les propriétés géotechniques du sol et de la roche mais aussi le mécanisme de
glissement de terrain dans la région concernée. L’approche stochastique fondée sur les modèles statistiques est choisie pour évaluer le
poids de facteurs d’influence comme l’altitude, l’aspect, la pente, la distance aux cours d’eau, la distance aux voies routières, et au
drains, les paramètres géologiques, et l’indice de l’humidité des sols. Le zonage des risques de glissement de terrain a été développé
en utilisant un modèle de probabilité conditionnelle de Bayes avec différents critères de calcul du poids des variables, en intégrant les
outils analytiques de SIG. Les cartes de risques élaborées selon ces modèles ont été validées à l’aide de l’ inventaire des glissements
actuels afin de choisir le modèle le plus approprie pour ce site.
KEYWORDS: Landslide susceptibility; Bayesian conditional probability; Stochastic model.
1 INTRODUCTION
The study area locates at An Linh commune, Tuy An district,
Phu Yen province in the south central coast of Vietnam (see
Fig. 1).
Figure 1. Location of the study area
During the past few years, landslides often occur in the study
area after heavy and long rains. This phenomenon damages
roads and construction works and results in serious cracks in the
residential areas that in turn affect to human lives. Therefore the
landslide prediction is an urgent task for scientists to support the
government in planning and save human lives of local residents.
1.1
Geographic characteristics
An Linh area which has hill and mountain landscape locates on
the eastern side of Van Hoa highland whose elevation of
12 ~ 500m. In that area, hills are rounded and have steep slopes,
decorated by narrow valleys which follow along north-south
direction. The ground surface which inclines to the north is a
bare area with basaltic boulders or cobbles. This topographic
landscape is favorite for erosion progress and stimulate
landsliding and cracking process.
The study area belongs to the monsoon tropical region with
oceanic climate. The annual precipitation is approximate
2000mm. Especially in the rainy season the precipitation can
reach 70 ~ 80% the annual precipitation. Furthermore, the
precipitation concentrates in a short time – only one or two
months at the end of a year.
