Production of Landslide Susceptibility Map using Bayesian Probability Model

Production of Landslide Susceptibility Map using Bayesian Probability Model

Halil Akıncı (Department of Geomatics Engineering, Artvin Çoruh University, Artvin, Turkey), Ayşe Yavuz Özalp (Department of Geomatics Engineering, Artvin Çoruh University, Artvin, Turkey), Mehmet Özalp (Department of Forestry Engineering, Artvin Çoruh University, Artvin, Turkey), Sebahat Temuçin Kılıçer (Department of Geomatics Engineering, Artvin Çoruh University, Artvin, Turkey), Cem Kılıçoğlu (Poplar Vocational School, Samsun, Turkey) and Emre Everan (Artvin Vocational School, Artvin Çoruh University, Artvin, Turkey)
Copyright: © 2015 |Pages: 18
DOI: 10.4018/IJ3DIM.2015040102
OnDemand PDF Download:
No Current Special Offers


Artvin is one of the provinces in Turkey where landslides occur most frequently. There have been numerous landslides characterized as natural disaster recorded across the province. The areas sensitive to landslides across the province should be identified in order to ensure people's safety, to take the necessary measures for reducing any devastating effects of landslides and to make the right decisions in respect to land use planning. In this study, the landslide susceptibility map of the Central district of Artvin was produced by using Bayesian probability model. Parameters including lithology, altitude, slope, aspect, plan and profile curvatures, soil depth, topographic wetness index, land cover, and proximity to the road and stream were used in landslide susceptibility analysis. The landslide susceptibility map produced in this study was validated using the receiver operating characteristics (ROC) based on area under curve (AUC) analysis. In addition, control landslide locations were used to validate the results of the landslide susceptibility map and the validation analysis resulted in 94.30% accuracy, a reliable outcome for this map that can be useful for general land use planning in Artvin.
Article Preview

1. Introduction

Landslides defined as the downward and outward movement of slopes forming materials composed of natural rock, soils, artificial fills, or combinations of these materials (Varnes, 1958). As one of the most common natural disasters in the world, landslides cause large-scale socio-economic devastation such as casualties, economic damage, and loss of cultural and natural heritage. For example, as a result of the landslide caused by heavy rainfall in the northeastern province of Badakhshan in Afghanistan on May 02, 2014, 256 people were killed and 300 houses were damaged (IFRC, 2014).

Landslides are known to be one of the leading natural disasters causing life and property losses in Turkey, too. When natural disasters between the years 1950-2000 in Turkey are analyzed, landslides are found to be the most frequent natural disasters with 45% (Gokce et al., 2008). Caused by heavy rains, one of the latest landslides with devastating effects in Turkey was recorded on August 26, 2010 in the town of Gundogdu at the center of Rize, killing 13 people.

Artvin is among the first ten provinces where natural disasters, with 658, have the most common occurrences in Turkey. In addition, when all these natural disasters in Artvin are analyzed, 471 of them, about 72%, were found to be landslides (Gokce et al., 2008). Natural (and technological) disasters with specific characteristics are included in the National Disaster Archive of Turkey managed by the Disaster and Emergency Management Presidency (AFAD) of the Republic of Turkey Prime Ministry. If, in the event of a disaster, at least 10 people are lost their lives, at least 50 people are injured, or at least 100 people are affected, that is recorded in the Archive. In this context, when this Archive is examined, one can see that total of 57 landslides -characterized as natural disasters- were occurred in Artvin, killing five people and affecting 6757 people up until today. Considering the fact that the landslides are the leading natural disasters causing losses of life and property in Turkey and high potential risk of landslides in Artvin, there was a great need for producing a landslide susceptibility map of Artvin’s central district in order to prevent the possible loss of life since it is inhabited by 20% of the whole population in the Artvin province.

Landslide susceptibility maps reveal sensitive areas against landslides in the future and refer to the tendency of any area to landslides (Dağdelenler, 2013). Although researchers have been using different parameters in landslide susceptibility analysis due to regional characteristics, in studies evaluating which methods and parameters should be used in the preparation of landslide susceptibility maps (Gökçeoğlu ve Ercanoğlu, 2001; Dağ et al., 2011), it has been expressed that slope, aspect, lithology and land cover are frequently used parameters. In parameter selection, one of the most important and effective factor is the fact that whether a reliable data set related with any parameter can be easily obtained. In this study, lithology, altitude, slope, aspect, plan and profile curvature, land cover, topographic wetness index, soil depth, proximity to the road and stream parameters were used.

In the literature, it occurs that there is no consensus among researchers on the methods and parameters used in the preparation of landslide susceptibility maps and there is a large number of parameters and methods in use due to the fact that researchers take into account parameters related to their study area (Gökçeoğlu ve Ercanoğlu, 2001; Dağ et al., 2011). Dağ et al. (2011) have pointed out that approximately 64% of all landslide susceptibility maps have been prepared by various statistical methods. In this study, Bayesian probability model was used since it is widely used in the literature, consists of simple and understandable statistical models, provides accurate results, and can be easily applicable.

Complete Article List

Search this Journal:
Open Access Articles
Volume 7: 4 Issues (2018)
Volume 6: 4 Issues (2017)
Volume 5: 4 Issues (2016)
Volume 4: 4 Issues (2015)
Volume 3: 4 Issues (2014)
Volume 2: 4 Issues (2013)
Volume 1: 4 Issues (2012)
View Complete Journal Contents Listing