Landslide susceptibility assessment in Kumluca (se Bartın) watershed by geoeraphic information systems and statistical analysis methods

ABSTRACTMsc. ThesisLANDSLIDE SUSCEPTIBILITY ASSESSMENT IN KUMLUCA (SEBARTIN) WATERSHED USING GEOGRAPHIC INFORMATION SYSTEMSAND STATISTICAL ANALYSIS METHODSTolga MazmanDEPARTMENT OF GEOLOGICAL ENGINEERINGINSTITUTE OF NATURAL AND APPLIED SCIENCESUNIVERSITY OF ÇUKUROVASupervisor: Assist.Prof.Dr. Tolga ÇANYear: 2005, Page: 110Jury : Assist.Prof.Dr. Tolga ÇANProf.Dr. Aziz ERTUNÇAssoc.Prof.Dr. Candan GÖKÇEOĞLUIn this study, landslide susceptibility assessments in the Kumluca Rivercatchment (340 km2), located in the SE of Bartın being one of the tributary of theBartın River, were carried out by using geographical information systems andlogistic regression analysis. The selected area has been adversely affected bylandslides and in the period of 1985 and 1998, total dwellings considered in thelandslide disaster area were determined as 370 of which 256 were located in theKumluca catchment. In the catchment 58 additional destructed dwellings have beenrecorded since 1998. The geological, morphological and land use maps weredigitised and slope and aspects maps, which all could be considered as preperatoryfactors of landslides, were derived from digital elevation model of the study area inGIS environment. Landslide susceptibility maps were evaluated by usingmultivariate (logistic regression) statistical methods. Slope units were used asmapping unit in the statistical evaluations. Susceptibility classes were grouped intofive categories according to the probability intervals of 20 % as very low, low,moderate, high and very high. 27 % of the area was found in high to very highsusceptible zones of which 77% of the landslides located therein according to theperformance evaluations. As a consequence, the landslide susceptibility mapproduced herein is considered a valuable tool for reducing the landslide hazard andfor the planning purposes.Keywords: Geographical information systems, digital elevation model, landslidesusceptibility, logistic regression, landslide inventory.II