Yenice (Karabük) Bölgesinin Heyelan Tehlikesinin Değerlendirilmesi

Abstract

Particularly in recent years, there has been a significant increase in the number and quality of landslide inventory, susceptibility, hazard and risk studies as a result of significant developments in such studies. The fundamental approach around the world is the production and implementation of such maps in the selected areas by adopting an approach from the general to the specific in areas where the development of landslides is possible. In Turkey, up to date, there have been many losses of lives and properties because of the landslides. For this reason, it was aimed at conducting a GIS (Geographical Information System) based landslide hazard assessment in an area of 837 km2 between Karabük and Yenice at which such losses are frequently encountered. In the first stage of the thesis, a comprehensive literature research was conducted, and in the next step, the most appropriate approach for the landslide hazard analysis was presented. The existing archive data, the studies carried out by AFAD and MTA, orthophoto and aerial photography interpretations and in-situ mapping were performed and a total of 104 landslides were mapped. For the evaluation of landslide susceptibility, the conditioning parameters such as slope, aspect, topographical elevation, slope curvature, topographic wetness index (TWI), stream power index (SPI), sediment transport capacity index (STCI), lithology, distance to drainage, distance to roads and land use parameter maps with the pixel size of 20mx20m were prepared, and a total of 11 parameter maps were taken into account for the landslide susceptibility map produced by logistic regression method. According to this map, it was identified that 60.7% of the study area was very low, 10.9% was low, 3.5% was moderate, 5.7% was high and 19.2% was very high landslide susceptibility, and by performing a ROC analysis, the AUC value was calculated as 0.881. Due to the lack/deficiencies in the rainfall data, a multi temporal landslide hazard analysis approach was preferred instead of precipitation data analysis. In the determination of the temporal probability of the landslides in the study area, 164 landslide events occurred in the period between 1905 and 2013 were considered, and were analyzed with Poisson and Binomial distributions. Considering different temporal and spatial magnitudes, landslide hazard maps of the region were produced. In general terms, it was found that the landslide hazard is much higher for small-scaled landslides in the short term than that of the large-scaled landslides. It is recommended that landslide risk studies should be immediately carried out and put into practice by increasing the data quality of the region.