In spite of considerable efforts dedicated to understanding and simulating the accumulation and ablation of the snowpack, there is arguably still a colossal amount of uncertainty in physics-based snow models. The fidelity of model representations of snow processes remains compromised by limited process understanding, and limited data. Many modeling decisions are often represented using empirical formulations with model parameters defined based either on order-of-magnitude considerations or by fitting a curve through a limited amount of experimental data. An outstanding challenge for the snow modeling community is to understand the impact of differences among existing snow models on their physical validity and predictive performance, and use this to identify and address fundamental model weaknesses.

We introduce an alternative methodology for understanding differences among snow models. We first define a single set of governing model equations – a “master modeling template” – from which many existing models can be reproduced, and new models derived. The master template is then implemented within a single robust numerical framework, and used to explore the impact of differences in the choice of modeling approaches and the choice of model parameter values. To keep the study tractable, we focus on a subset of modeling options available within the template, restricting attention to one-dimensional snow model applied over non-vegetated surfaces. Assessments of forest snow processes and spatial variability are deferred to a separate study.

The differences among existing snow models can be broadly classified into three categories: (i) estimation of fluxes at the snow-atmosphere interface, including the approach used to estimate the surface albedo, the turbulent fluxes of sensible and latent heat, and the partitioning of precipitation between rain and snow; (ii) internal processes within the snowpack, including heat conduction, penetration of shortwave radiation, vertical drainage of liquid water, and compaction of the snowpack associated with metamorphism of the snow crystals; and (iii) estimation of fluxes at the lower boundary associated with heat transfer in the soil. Results show that (in most cases) the impacts of differences in model structure are overwhelmed by uncertainty in a-priori estimates of model parameters, and suggest that careful specification of probability distributions of model parameters can be used to represent model uncertainty.