Snow avalanche tracks mapping within Bâlea glacial valley (the Făgăraș Mountains) using semi-automated detection methods
Abstract: Mapping of snow avalanche tracks based on topographic maps, aerophotos and field data to achieve inventories for the whole mountaineous areas in Romania is an important step in snow avalanche risk assessment and other related geomorphic processes. This requires experience and it is a time consuming process. In the absence of field data, the process of snow avalanche tracks mapping is influenced by the subjectivity of those who digitize.
Thus, we propose a semi-automated method for detection of snow avalanche tracks based mainly on geomorphometric parameters that can be extracted from the Digital Elevation Model (DEM) like slope gradient, plan and profile curvature, mean curvature, runoff.
In this study we used an object based analysis to detect snow avalanche tracks in central part of the Făgăraș Mts. This approach has two steps, segmentation and classification. First, we segmented the area based on plan curvature (which is the most important parameter that describes these snow avalanche tracks) in order to obtain objects. In the process of classification we added other conditions such as fuzzy function for slope gradient, thresholds for altitude and runoff and a shape index of objects. The results obtained were very close to the mapped tracks using digitizing techniques. The maps resulted from the classification were compared to the those resulted from digitizing in both number of objects and spatial agreement of the class of objects. There was a very good fit in case of the number of objects and total area of objects. The method could be improved if we apply on high resolution DEMs and also on more case studies with different topography and existing vector database.