Modelling High‐Alpine Karst Runoff With Coupled Snow Models of Different Complexity Levels at Mount Zugspitze

ABSTRACT

High‐alpine, snow‐dominated karst catchments are among the most challenging environments for hydrological modelling due to complex topography, heterogeneous snow distribution, and limited discharge observations. The representation of snow processes has significant implications for simulated runoff generation and water balance partitioning, yet direct comparisons in karst settings remain scarce. Two model configurations are compared for the Partnach Spring Catchment (PSC), a single‐outlet karst system at Mt. Zugspitze (Northern European Alps, Germany), over the hydrological years 2015–2025. A fully conceptual CemaNeige–GR4H setup is evaluated against a hybrid configuration combining Alpine3D with GR4H routing. Alpine3D snow simulations are independently validated using point‐scale snow depth and snow water equivalent (SWE) measurements, as well as Sentinel‐2‐derived snow‐covered area (87 scenes). Both configurations achieve strong discharge performance (calibration Kling‐Gupta Efficiency (KGE): 0.92 ± 0.01 vs. 0.91 ± 0.01; validation KGE: 0.82 ± 0.07 vs. 0.87 ± 0.03), though 41.3% of discharge observations are missing, concentrated in winter months, limiting seasonal evaluation. The hybrid configuration shows improved temporal transferability despite requiring fewer calibration parameters (2 vs. 5). Annual water balances align within 0.16%, yet the conceptual configuration compensates through up to 78% greater routing storage capacity, masking structural differences in internal process representation. These results show that lumped routing structures can reproduce discharge dynamics in complex, snow‐dominated karst systems regardless of snow and boundary layer model complexity. However, the hybrid approach provides clear advantages for representing spatial snow dynamics and for gaining process‐level understanding of high‐alpine hydrology.