LANDSLIDE SUSCEPTIBILITY ASSESSMENT USING GEOGRAPHICAL INFORMATION SYSTEM BASED LOGISTIC REGRESSION SYSTEM BETWEEN ÇUBUK-KALECİK (ANKARA) AND ŞABANÖZÜ (ÇANKIRI) REGIONS

Abstract

Landslides occur frequently in particular is in certain parts in our country as well as in the world, is caused from time to time to return to natural disasters and signifi cant social and economic losses. In this study, landslide susceptibility analysis Çubuk-Kalecik (Ankara) between Şabanözü (Çankırı) were performed totally 2360 km2 area. According to the inventory map that represents about 5,16 % of the spatial distribution of landslide at study area, landslide movement is generally seen as the type of the existing rotational slip. According to the Turkish Landslide Inventory database prepared by MTA, 876 landslides covering 122 km2 were identifi ed in the study area. A total of twelve independent variables for the landslide conditioning factors were used in the susceptibility assessments, being as lithological maps, landform classifi cation, digital elevation model, slope, profi le, plan, and tangent curvatures, roughness index, slope / aspect ratio, stream power index, topographic position index. Landslide susceptibility assessment was carried out using multivariate logistic regression method, one of statistical methods. The mapping unit 25 *25 m pixels are used for statistical assessment. The obtained probability values of landslide susceptibility maps are very low and very high was evaluated in fi ve grades in the range. Performance evaluation of susceptibility maps were performed with using receiver operating characteristic (ROC) and prediction success rate curves. The area under the ROC curves were found to be in 0,794. Susceptibility map, high and very susceptible regions correspond to 27 % of the study area. 78 % of the landslides are in medium, high and very susceptible regions.The accomplished landslide susceptibility map with relatively high performance could be used during the medium scale planning strategies.