Modelling snowmelt in ungauged catchments

Temperature-based snowmelt models are simple to implement and tend to give good results in gauged basins. The situation is, however, different in ungauged basins, as the lack of discharge data precludes the calibration of the snowmelt parameters. The main objective of this study was therefore to assess alternative approaches. This study compares the performance of two temperature-based snowmelt models (with and without an additional radiation term) and two energy-balance models with different data requirements in 312 catchments in the US. It considers the impact of: (i) the meteorological forcing, by using two gridded datasets (Livneh and MERRA-2), (ii) different approaches for calibrating the snowmelt parameters (an a priori approach and one based on Snow Data Assimilation System (SNODAS), a remote sensing-based product) and (iii) the parameterization and structure of the hydrological model used for transforming the snowmelt signal into streamflow at the basin outlet. The results show that energy-balance-based approaches achieve the best results, closely followed by the temperature-based model including a radiation term and calibrated with SNODAS data. It is also seen that data availability and quality influence the ranking of the snowmelt models.