Semi-automated calibration method for modelling of mountain permafrost evolution in Switzerland

Abstract

Abstract. Permafrost is a widespread phenomenon in mountainous regions of the world such as the European Alps. Many important topics such as the future evolution of permafrost related to climate change and the detection of permafrost related to potential natural hazards sites are of major concern to our society. Numerical permafrost models are the only tools which allow for the projection of the future evolution of permafrost. Due to the complexity of the processes involved and the heterogeneity of Alpine terrain, models must be carefully calibrated, and results should be compared with observations at the site (borehole) scale. However, for large-scale applications, a site-specific model calibration for a multitude of grid points would be very time-consuming. To tackle this issue, this study presents a semi-automated calibration method using the Generalized Likelihood Uncertainty Estimation (GLUE) as implemented in a 1-D soil model (CoupModel) and applies it to six permafrost sites in the Swiss Alps. We show that this semi-automated calibration method is able to accurately reproduce the main thermal condition characteristics with some limitations at sites with unique conditions such as 3-D air or water circulation, which have to be calibrated manually. The calibration obtained was used for global and regional climate model (GCM/RCM)-based long-term climate projections under the A1B climate scenario (EU-ENSEMBLES project) specifically downscaled at each borehole site. The projection shows general permafrost degradation with thawing at 10 m, even partially reaching 20 m depth by the end of the century, but with different timing among the sites and with partly considerable uncertainties due to the spread of the applied climatic forcing.

The European mountain cryosphere: A review of past, current and future issues

“…The triggering mechanisms of slope instabilities can be diverse, and depend on subsurface material (e.g., unconsolidated 25 sediments versus bedrock), its characteristics (fractures and fissures, ice and water content, slope angle, geological layering), and changes of these properties with time (Hasler et al, 2012;Krautblatter et al, 2013;Ravanel et al, 2013;Phillips et al, 2016). Water infiltration into newly thawed parts of permafrost is often mentioned as a possible triggering mechanism (Hasler et al, 2012), but only few observational data are available.…”

“…ground-based GPS monitoring; Wirz et al, 2014Wirz et al, , 2016, and for the estimation of sediment transfer rates between rock wall and creeping permafrost bodies (Müller et al, 2014). These monitoring methods 20 are, however, restricted to rock glaciers, rockfalls and subsidence (e.g., Deline et al, 2015, Phillips et al, 2016. Remote sensing has so far not enabled thermal changes in permafrost to be assessed.…”

“…These models should not be confused with permafrost distribution models 10 (Boeckli et al, 2012), which are mostly statistical and usually based on permafrost evidences such as rock glacier inventories or climatic variables such as potential incoming solar radiation or mean annual air temperature. Similar to land-surface schemes used for hemispheric permafrost modeling (e.g., Ekici et al, 2015;Chadburn et al, 2015;Peng et al, 2016), physically-based site-level models are used in combination with Regional Climate Models (RCM) for studies of long-term permafrost evolution (Farbrot et al, 2013;Scherler et al, 2013;Westermann et al, 2013;Marmy et al, 2016). Such models 15 are also used to explain the existence of specific permafrost phenomena in the European mountains (Scherler et al, 2014;Fiddes et al, 2015;Zhou et al, 2015;Haberkorn et al, 2016;Lüthi et al 2016,).…”

Towards improved monitoring of changing permafrost by estimating soil characteristics from ground temperature time-series

“…Although parameter fitting techniques have been used to calibrate soil properties for ground thermal models in permafrost environments, the results have chiefly been applied to the prediction of future permafrost states (e.g. Harp et al, 2016;Marmy et al, 2016). Little work has been done that uses the resulting parameters to estimate observed changes to subsurface heat or liquid water content at the monitoring sites.…”

“…(Atchley et al, 2015;Nicolsky et al, 2007b). However, Marmy et al (2016) report that using the coefficient of determination (R 2 ) as an objective function improves calibration results near the freezing point. They also suggest that the total energy content of the ground could be incorporated into the objective function in future research.…”