GIS-based erosion and landslide susceptibility assessment of the ayvalik and surroundings

The purpose of this thesis is to make an erosion and landslide susceptibility assessment of an a area of about 424 km2 around Ayvalık by Geograpical Information Systems (GIS) and remote sensing techniques using the field and laboratory based data. In this study, three main data set were used. They are 30 m spatial resolution Landsat TM/ETM+ and 15 m spatial resolution Aster 3A0 satellite images and 1/35.000 scale aerial photographs, 1/25.000 scale geological maps and topographical sheets. A landslide inventory of the investigation area was obtained using the aerial photographs obtained in 1995 and satellite images acquired in 1987, 2001 and 2004 years. Gully erosion sites were also detected using the same satellite images. Weathering grades map of the area was produced using the geological maps and the laboratory determined data. From the digital topographical data which was obtained by digitizing the topographical sheets, slope gradient, slope aspect, stream power index, topographical wetness index, drainage density and distance to drainage paths data were produced. Lineament density and distance to lineaments data were obtained from the geological maps and the topographical sheets. With the help of the satellite images, land-cover and vegetation density data were extracted. Unit weight, porosity, void ratio, water content, slake durability index, quick water absorption and mass loosing after the freezing-thawing cycles of the rock samples collected from the field were found in laboratory. The labarotory determined were converted to map type in a GIS environment using point interpolation method. In order to anlyse the all data together, they were converted into a 30x30 m raster data type. Fuzzy logic approach was used to standardize the data produced in a common scale changing from 0 to 250. The standardized data were assessed by Bivariate statistics (Likelihood Ratio), Multivariate Statistics (Logistic Regression) and Multiix Criteria Decision Analysis (Weighted Linear Combination) methods by constructing the different data sets and the areas susceptible to erosion and landslide were determined. The data obtained at the end of the landslide susceptibility assessments were compared with the landslide localities in the landslide inventory map, and so the validation process was carried out. The erosion suscepibility maps produced were also compared with the actual gully erosion sites, slake durability and mass loosing of the rocks after the freezing-thawing data and the maps obtained were verified. The 58 %, 42% and 8% of the slided areas in the area fall into the high and very high susceptibility classes determined using Multi Criteria Decision Analysis, Likelihood Ratio and Logistic Regression methods respectively. In the light of these results, the Multi-Criteria Decision Analysis method was concluded to be the most reasonable and the most realistic method among the other methods used for landslide susceptibility assessments. The erosion susceptibility maps obtained by Multi- Criteria Decision Analysis technique was found to be more acceptable method than the erosion susceptibility maps produced using the logistic regression analysis. According to the landslide susceptibility analysis results, the areas, where the rocks show high to very high weathering grades, the hillslopes having moderately to high slope gradients, NW facing hillslopes having relatively high precipitation, sparse vegetated areas and the areas having high drainage density were determined to be highly and very highly susceptibile to sliding. In addition, the areas where moderately to highly weathered rocks present in steep slopes, NW facing convex hillslopes and the areas having high stream power index were determined to be showing highly susceptibile to erosion. Keywords: Erosion, Landslide, Multivariate Statistics, Geographical Information Systems, Multicriteria Decision Analysis.