Assessing the impacts of vegetation heterogeneity on energy fluxes and snowmelt in boreal forests

Aims In the mid- and high-latitude regions, three quarters of the land surface is covered by boreal conifer forests, and snow lasts for 6–8 months of the year. Correctly modeling surface energy balance and snowmelt at mid- and high-latitudes has a significant influence on climate and hydrological processes. However, the heterogeneous and clumped forest structure exerts important control over the radiative energy at the forest floor, which results in large variations of underneath snow cover and snowmelt rate. The goal of this study is to investigate the impact of hierarchically clumped vegetation structure in boreal forest on snowmelt and exchanges of energy and water. Methods We used a simple Clumped Canopy Scheme (CCS) for canopy radiation transfer to characterize the impact of the clumped forest structure on net radiation at the snow surface underneath forests. The CCS was integrated with the Variable Infiltration Capacity macroscale hydrological model (herein referred to as VIC-CCS) to characterize the impact of clumped vegetation structure on surface energy balance and snowmelt during the snow season. A twin simulation, VICCCS and the standard VIC model, was performed to isolate the impact of CCS on the energy and water fluxes and snowmelt rates. The simulation results were compared to in situ measurements at four different forest stands: old aspen forest in the Southern Study Area (SOA), black spruce forests in the Southern and Northern Study Areas (SOBS and NOBS) and fen wetland in the Northern Study Area (NFEN) within the Boreal Ecosystem–Atmosphere Study (BOREAS) region in central Canada during 1994 to1996. Important Findings Simulations showed that the implementation of CCS has reduced incoming long-wave radiation at the underlying snow surface and, thereby, lowered the snowmelt rate. Comparison against ground observations of net radiation and surface flux rates showed a reasonable agreement while demonstrating implementation of CCS can markedly improve model surface energy budget and energy inputs computation for snowmelt. The modeled snowmelt matches reasonably well with observations with root mean square error (RMSE) ranging from 16.51 to 19.81 mm using VIC-CCS versus 29.86 to 32.61 mm for VIC only in the four forest sites. The improvement is the most significant for the deciduous forest (old aspen) site, reducing RMSE by16 mm. This study demonstrates that taking into account the effect of the clumped forest structure in land surface parameterization schemes is critical for snowmelt prediction in the boreal regions.