Quantification of different flow components in a high-altitudeglacierized catchment (Dudh Koshi, Nepalese Himalaya)

Abstract. In a context of climate change and water demand growth, understanding the origin of water flows in the Himalayas is a key issue for assessing the current and future water resources availability and planning the future uses of water in downstream regions. This study estimates the relative contributions of rainfall, glacier and snow melt to the Khumbu River streamflow (Upper Dudh Koshi, Nepal, 146 km 2 , 43 % glacierized, elevation range from 4260 to 8848 m a.s.l.), as well as their seasonal variability during the period 2012–2015, by using the physically based glacio-hydrological model DHSVM-GDM (Distributed Hydrological Soil Vegetation Model – Glaciers Dynamics Model). One of the main issues in high elevated and glacierized catchments hydrology is the limited representation of cryospheric processes, which control the evolution of ice and snow, in distributed hydrological models. Here, the impact of different snow and glacier parametrizations was tested by modifying the original DHSVM-GDM snow albedo parametrization, by adding an avalanche module, and by adding a reduction factor for the melt of debris covered glaciers. Results show that this new version of DHSVM improves the simulation of the snow covered area and the glacier mass balances, thus improving the reliability of the overall hydrological simulation. In the presented case study, ice and snow melt contribute each more than 40 % to the annual outflow. 69 % of the outflow originates from glacierized areas. Our simulations also highlight that winter flows are mainly controlled by the release from the englacial water storage. In general, it is shown that the choice of a given parametrization for the snow and glacier processes has a significant impact on the simulated water balance. The sensitivity of the model to the glaciers inventory was tested, demonstrating that the uncertainty related to the glacierized surface leads to an uncertainty of 20 % on the simulated ice melt component.