Three-dimensional canopy structure is necessary to model snowpack mass and energy budgets in montane forests

Modeling the effects of forest variability and change on snow is critical to resource management. However, many snow models are too coarse to represent fine scale canopy structure in heterogeneous forests, and therefore simplify its impact on snowpack mass and energy budgets. To quantify the loss of snowpack prediction skill from simplifications in forest canopy processes, we applied a high-resolution energy budget snowpack model at two forested sites at a fine (1 m²) and coarser (100 m²) resolution. Reductions in prediction skill from the coarser simulations were then predicted using three-dimensional forest structure metrics (calculated at a 1 m² resolution). In general, the coarser simulations predicted greater under-canopy radiation, faster snow ablation, and had both higher and lower interception amounts, depending on the simulation. These differences were best explained by a combination of forest structure metrics (canopy cover, tree height, and canopy edginess), accounting for an average of 61% of coarse model bias in peak SWE, 68% of bias in interception, 68% of bias in under canopy shortwave radiation, and 77% of bias in longwave radiation. On flat and north facing slopes, tree height and canopy cover were the best predictors of the coarse model bias, while on south-facing slopes, canopy cover and forest edginess were better predictors. Our results suggest that current land surface models that do not include information about fine-scale forest structure may contain simulation errors of snowpack in forests, potentially limiting their ability to reliably simulate snowpack changes due to forest disturbance.