Abstract. Snow interception by forest canopy drives the spatial heterogeneity of subcanopy snow accumulation leading to significant differences between forested and non-forested areas at a variety of scales. Snow intercepted by forest canopy can also drastically change the surface albedo. As such, accurately modelling snow interception is of importance for various model applications such as hydrological, weather and climate predictions. Due to difficulties in direct measurements of snow interception, previous empirical snow interception models were developed at just the point scale. The lack of spatially extensive data sets has hindered validation of snow interception models in different snow climates, forest types and at various spatial scales and has reduced accurate representation of snow interception in coarse-scale models. We present two novel models for the spatial mean and one for the standard deviation of snow interception derived from an extensive snow interception data set
collected in a spruce forest in the Swiss Alps. Besides open area snowfall, subgrid model input parameters include the standard deviation of the DSM (digital surface models) and the sky view factor, both of which can be easily pre-computed. Validation of both models was performed with snow interception data sets acquired in geographically different locations under disparate weather conditions. Snow interception data sets from the Rocky Mountains, U.S. and the French Alps compared well to modelled snow interception with a NRMSE for the spatial mean of 10 % for both models and NRMSE of the standard deviation of 13 %. Our results suggest that the proposed snow interception models can be applied in coarse land surface model grid cells provided that a sufficiently fine-scale DSM of the forest is available to derive subgrid forest parameters.
