Improving the degree-day method for sub-daily melt simulations with physically-based diurnal variations

This paper proposes a new extension of the classical degree-day snowmelt model applicable to hourly simulations for regions with limited data and adaptable to a broad range of spatially-explicit hydrological models. The snowmelt schemes have been tested with a point measurement dataset at the Cotton Creek Experimental Watershed (CCEW) in British Columbia, Canada and with a detailed dataset available from the Dranse de Ferret catchment, an extensively monitored catchment in the Swiss Alps. The snowmelt model performance is quantified with the use of a spatially-explicit model of the hydrologic response. Comparative analyses are presented with the widely-known, grid-based method proposed by Hock which combines a local, temperature-index approach with potential radiation. The results suggest that a simple diurnal cycle of the degree-day melt parameter based on minimum and maximum temperatures is competitive with the Hock approach for sub-daily melt simulations. Advantages of the new extension of the classical degree-day method over other temperature-index methods include its use of physically-based, diurnal variations and its ability to be adapted to data-constrained hydrological models which are lumped in some nature.

► The improved degree-day approach is based solely on daily maximum and minimum temperature data. ► It assigns degree-day factors at the subcatchment scale rather than at the grid-scale. ► A spatially-explicit hydrological model demonstrates the melt method is robust and flexible. ► It provides accurate snowmelt responses and is applicable to a wide range of hydrological models. ► It is a valuable approach in data-limited mountainous which require sub-daily melt computations.