Hyper-resolution large-scale hydrological modelling benefits from improved process representation in mountain regions

Abstract. Many of the world’s major rivers originate in mountain regions and a large fraction of the global population relies on these regions for their water supply. The hydrological cycle of mountain regions and their dependent downstream regions are often studied using large-scale to global hydrological models (LHMs). The increasing spatial resolution of these models allows for improved representation of complex mountain topography, but existing model deficiencies in cold and high-elevation regions limit potential model performance gains. Such model performance gains might be realized by investing into a better representation of hydrological processes that are relevant in mountain regions such as snow-accumulation and -melt. However, how much improved process representation would increase LHM performance remains largely unquantified. Here, we set up the hyper-resolution global hydrological model PCR-GLOBWB 2.0 (PCRaster Global Water Balance) over the larger Alpine domain and implement several changes to make it better suited at representing hydrological processes in mountain regions. These changes include a.) the use of novel high-resolution meteorological forcing datasets; b.) an extended snow module based on a seasonally varying degree-day factor and an exponential melt function; c.) a regional calibration of the snow module against a snow reanalysis product; d.) a new integrated glacier module; and e.) increasing the contributions to the fast runoff components in the soil. Our evaluation of the effect of these different adjustments on model performance for discharge shows that while the meteorological forcing has a major effect on discharge simulations, its effect on performance is not unidirectional over the domain. In addition, the structural and parametric changes, i.e. the snow module modification, glacier representation and runoff partitioning, improve discharge simulations in mountain regions: the snow module modification leads to an improved representation of the snowmelt peak for high-elevation catchments, the glacier module supplies additional water to glacierized catchments, and runoff partitioning in the soil improves the representation of streamflow in flashy catchments at lower elevations. We use these insights to present a new setup of the large-scale and hyper-resolution PCR-GLOBWB 2.0 model that is better suited to study hydrological processes in and beyond mountain regions around the world.