Abstract
Rock glaciers are complex geomorphic units of high-mountain landscapes. They are composed of a mixture of ice and debris at varying degrees, whose deformation enables a steady creep behavior. Glacial, periglacial and paraglacial mechanisms were proposed to be drivers of their origin. The present-day persistence of rock glaciers in the landscape however is enabled by cryotic environmental conditions. The internal deformation of rock glaciers influences directly the amount of debris they are transporting. Hence, rock glaciers can be considered to be active conveyors of sediment in high mountain areas. In the context of climate change, rock glaciers increasingly gained attention as their downwasting driven by rising atmospheric temperatures has implications for the management of natural hazards and freshwater resources. The decay of rock glaciers is often accompanied by an acceleration of their creep behavior, which can even end in their partial or complete destabilization. The importance of meltwater derived from presently ice-containing rock glaciers will increase
in relevance as they were found to be more resilient to melting than clean glaciers. However, approaches to address these pressing and timely issues are still rare and often not transferable.
This thesis aims to respond to this need by proposing and applying novel assessment methods to evaluate rock glacier ice presence, sediment transfer and front stability. Following a state of the art review, three thematic chapters are responding step-by-step to this overarching aim. Firstly, a regional scale assessment method for rock glacier ice-presence is presented and applied in the region of South Tyrol (northern Italy). The paper shows strategies to exploit the increasing availability of appropriate data sources such as rock glacier inventories, high resolution terrain data or satellite data products and computational methods such as tatistical and machine-learning based algorithms. Three classifiers of various complexity (logistic regression, random forest and support vector machines) are compared and evaluated on their ability to spatially predict the likelihood of rock glacier ice presence. Results show that elevation and vegetation presence are the most powerful covariates to assign a status of activity to rock glaciers. Surprisingly, the applied statistical classifier performed better than the more complex machine-learning based models.
This study highlights also the need to apply a rigorously spatial perspective for training, testing and tuning the parameters of data-driven models. Second, the capacity of ice-containing rock glaciers to displace and transfer sediment at their front was estimated for the Schnalstal and Ultental catchments, both located in South Tyrol. Given the absence of validation data, a simple and easily reproducible heuristic method is proposed. The capacity of a rock glacier to transfer sediment at its front is considered to be determined by (i) the annual rate of transported debris by the rock glacier itself and (ii) the potential travel range of mobilized
material at its front. Both variables are then combined within a matrix-based integration scheme to an index that expresses on a qualitative level the capacity of a rock glacier to transfer sediment at its front (i.e. low, moderate, elevated). In a consecutive step, empirical relationships between the computed sediment transfer classes and independent topographic and environmental control variables are build.
Finally, at the example of two recent failures of rock glacier fronts observed in South Tyrol, a local-scaled assessment scheme for rock glacier front stability is presented. Besides the triggering rainstorm, medium-term and predisposing destabilizing factors are identified and quantified. A multi-method approach including
geotechnical testing and modelling, movement tracking, climate, borehole and meteorological data analysis is applied to provide a holistic view on the aspects affecting the stability of rock glaciers front slopes. Results for the study cases indicate that predisposing and preparatory destabilizing factors must not be neglected. Both landforms experienced a strong acceleration of their snouts over last decades, which might have affected their front stability. An exceptionally snow-rich winter, followed by a rainy spring and summer season indicate a critical role of water on slope stability, which is confirmed by geotechnical site investigations and modelling.