Process-based modeling of the streamflow generation in a highly glaciated Alpine headwater catchment since the last little Ice Age (i.e., 1850).

Abstract

The streamflow generation related to snow and glacier melt is particularly sensitive to temperature fluctuations and, hence, highly affected by global warming. However, the non-linear and complex interaction between streamflow contributions originating from snow melt versus glacier melt and being transferred to stream via the subsurface complicates the investigation of climate-induced changes in high-elevation catchments. We used the physically-based hydrological model WaSiM to simulate the climate-induced changes in the streamflow generation in the Kaunertal (Austria), a highly glaciated Alpine headwater catchment. The simulations extend from the last little Ice Age (i.e., 1850) to 2015. Large-scale climate processes of a general circulation model (GCM) were dynamically downscaled with the Weather Research & Forecasting Model (WRF) to the central Alpine region at a 2 km spatial resolution from 1850 to 2015. The WaSiM model parameters were transferred from a WaSiM configuration driven by station data and partly optimized by a manual calibration on observed streamflow. For model evaluation, a multi-objective approach was chosen considering streamflow, SWE, snow cover duration, and glacier mass balances. The hydrological model results showed a good representation of the individual components and seasonal streamflow generation. However, difficulties exist in the spatial representation of the heterogeneous and small-scale differences in the snowpack. In addition, there are limitations in the simulation of glacier evolution using WaSiM over long periods (> 30 years) in highly glaciated catchments, as WaSiM does not contain an ice flow routine that can simulate the glacier dynamics during advance or retreat. Despite the cascade of uncertainties in this complex model chain (i.e., GCM, WRF, WaSiM), the results of the long-term simulation show interesting dynamics and enable an analysis of streamflow generation for periods where no observational data is available. For instance, glacier melt indicates a high dependence on the development of summer temperatures (i.e., JJA). The rising temperatures led to an earlier onset of snow and glacier melt, which shifted the streamflow regime and increased the daily streamflow magnitude, especially from 1995 onwards. The next step will be the comparison of the hydrological model results with those from other headwater catchments in the eastern Central Alps with a different degree of glaciation. The novelty lies in comparing 165 years of simulated streamflow and the contributions from snow and glacier melt. This comparison will validate the transferability and generalizability of the complex model chain and the simulation results.