ASSESSING LANDSLIDE SUSCEPTIBILITY WITH GIS USING QUALITATIVE & QUANTITATIVE METHODS IN KNOX COUNTY, NEBRASKA

This thesis assesses landslide susceptibility using data from LiDAR DEMs, land cover, and soil surveys. Data were assessed quantitatively through Bayesian logistic regression within a geographic information system (GIS) and statistical software to produce a landslide susceptibility map. The study area exhibits moderate relief where bluffs along the Missouri River valley gradually recede into rolling loess-mantled hills further to the south and southeast in Knox County. The six factors used to determine susceptibility to landslides are: land cover, parent material, slope aspect, slope curvature slope degree, and soil series. My findings show an increase in slope is the most significant factor that causes landslides. In Knox County, for every one degree increase in slope, the odds of a landslide increase by 1.41. The next most important factors for determining landslides are soil series and parent material. The occurrence of a landslide in areas of Pierre Shale are five times as large as alluvium, a material that is not recognized as having a strong potential for landslides in Knox County. This finding is consistent with conclusions from previous regional studies. The last significant factor was land cover. The results of the land cover indicate construction of road networks found in developed land cover areas is statistically significant. I tested multiple combinations of models and found the best combined soil series and slope. The confusion matrix of this model has a predictive accuracy over 93%. The landslide susceptibility map created in this study distinguishes areas that are more likely to incur landslide activity or areas that have already experienced a significant amount of landslides. Advisor: Paul R. Hanso