The main goals of work package 1 were the quantification of denudation rates at steep rock walls in the European Alps, assessing how they vary in space and time, and test if denudation rates are linked to temperature-sensitive processes. To reach these goals, we conducted several field expeditions during which we collected different types of samples for analysis of cosmogenic nuclides. Comparison of rock wall denudation rates with predictions based on frost-cracking models showed limited convergence, leading us to suggest that frost-cracking may not be the key process controlling denudation at these sites. Instead, permafrost thaw and associated recent acceleration of denudation rates may play a more important role. A recent acceleration of denudation rates is supported by results from our medial moraine samples that provide archives of denudation rates in source areas over time periods of decades to centuries (Wetterauer et al., 2022). Comparing denudation rates from multiple glaciers, we find faster denudation at steeper, north-facing and presumably colder rock walls (Wetterauer and Scherler, 2023). However, we also observed that temporal variations in 10Be and 14C are complicated by stochastic erosion processes (Dennis and Scherler, 2022), and can be strongly influenced by recent glacier retreat in various ways (Wetterauer, 2023).
The main goals of work package 2 were to develop a coupled ice and landscape evolution model that includes the production of debris on headwalls, its transport to and by the glacier, and its effect on glacial mass balances, and to apply this model to invert cosmogenic nuclide data. We implemented the described modeling approach using the existing numerical model iSOSIA and included a Lagrangian particle tracing module (Scherler and Egholm, 2020). Based on cosmogenic nuclide data from the Chhota Shigri Glacier, we showed that rather large changes in 10Be concentrations along medial moraines are difficult to explain with actual changes in denudation rates. Instead, a significant change in the contributing source area, for example by deglaciation, appears more plausible to lead to these changes. We currently work with another ice model that is orders of magnitudes faster than previous ice flow models, as it employs deep learning and can be parallelized on graphic processing units. Together with our colleague G. Jouvet, a similar particle tracking as in iSOSIA has been implemented and we plan to apply this model to our cosmogenic nuclide data sets.
In work package 3 we used the cloud computing platform Google Earth Engine to develop algorithms that allowed us to map supraglacial debris cover at planetary scale based on Landsat and Sentinel 2 imagery (Scherler et al., 2018). The results provide the first global assessment of supraglacial debris cover extents and the provided data sets are widely used by other researchers. We further explored thermal imagery, which is equally sensitive to debris cover on ice surfaces, but in different ways than optical imagery. To assess its potential and limitations, we equipped a consumer-grade drone with a light-weight thermal sensor and mapped parts of the debris cover of the Tsijiore Nouve Glacier (CH). When using satellite-derived thermal images however, the accuracy of the thermal data improves, but at the cost of spatial resolution (Scherler et al., 2023). We thus focused on exploiting the long time series of thermal imagery offered by the Landsat satellites to derived trends of land surface temperatures across the European Alps (Gök et al., submitted). The obtained trend values allow to decipher areas of exceptional surface warming in terrain that is prone to catastrophic failure and difficult to access.
