LANDSLIDE SUSCEPTIBILITY MODELING OF NORTHEASTERN PENNSYLVANIA

As population grows and land use trends change, the study of mass movements (slope failures) and their potential effects on life and property has become essential. To mitigate the loss of lives and costs associated with mass movements, susceptibility maps - maps that show the occurrence probability of landslides under certain geo-environmental conditions - are some of the most useful tools available to planners and the public. The development of such susceptibility maps can be driven by the expertise and knowledge of individuals, the behavior of materials in varying physical conditions, or data. Knowledge- and physically-driven approaches are limited by size and scope, whereas data-driven methods are not, especially given computational advances in geospatial and big data. Among the most commonly used data-driven methods to produce landslide susceptibility maps is the frequency-ratio method, that calculates weight values for classes of individual landslide related factors (i.e., elevation, bedrock geology, etc.). The popularity of this method is based on its clear principles, ease of use, and that it allows for investigation of vulnerability to landslide failure of individual factors. In developing a susceptibility model for northeastern PA (NEPA), we use a new landslide inventory developed for the region using high-resolution digital terrain models (DTMs), as well as python scripting in ArcMap to process the dataset. We explored elevation, slope, aspect, mean annual precipitation (MAP), bedrock geology, and landcover as factors, which were classified based off natural jenks breaks, as well as user selected criteria based on data trends/behavior. A Landslide Susceptibility Index (LSI), or susceptibility factor, was calculated for each class to explore the correlation between geospatial landslide distribution and each factor. The process of calculating LSI also produces an LSI map to visually depict variations. An average of LSI maps, for all or subset combinations of factors, produces a susceptibility map which is tested using a receiver operating characteristics (ROC) curve of a 5-class natural jenks susceptibility classification: very high, high, moderate, low, and very low. We present and detail the results of our final susceptibility map for NEPA.