Using forest canopy density to model beneath canopy snowpack

Background/Question/Methods

Measurements under varying degrees of canopy closure have indicated that the snowpack beneath a forest can be 33% to 80% less than in an opening. Most of this difference in accumulation is related to the interception capacity of the canopy, such that structural characteristics that vary by forest type can account for up to 50% differences in snow water equivalent (SWE). Prior studies have modeled the interception capacity of individual forest stands using two-dimensional (2D) measures of stand structure including percent cover and leaf area index (LAI). These approaches do not explain a large amount of the among stand SWE variation, because these 2D measures of canopy closure do not accurately represent the 3D nature of the canopy and its ability to hold snow during and after a storm. A more accurate proxy for 3D canopy structure is plant area density (PAD), which can be estimated by creating a bounding volume for a crown and ‘refoliating’ it using LAI. We measured winter PAD in nine forest types (aspen, oak, maple, red pine, white pine, white spruce, balsam fir, hemlock, and cedar) on the Bayfield Peninsula in northwestern Wisconsin to demonstrate that snow accumulation beneath a forest canopy is directly related to PAD and that snow accumulation relationships based on PAD are more accurate than those based on LAI or canopy cover. 33 stands were sampled with variable sized plots, in which we recorded tree location, species, DBH, gap fraction, and plant area index (PAI). For a subset of plots, we measured tree height, crown circumference, and height to the first live branch and the widest point of the crown. PAI and gap fraction were measured using leaf-off hemispherical photography. PAD was estimated for each plot by distributing PAI across the canopy envelope derived for that plot. During the winters of 2006-2007 and 2007-2008, each plot was visited on a biweekly basis to measure the SWE of the snowpack and was referenced to local meteorology and to SWE measurements in nearby fields. Each measurement of SWE was referenced to the local maximum snowpack for that sample period.

Results/Conclusions

We found that SWE decreased by as much as 55% as canopy density increased. These relationships between beneath canopy SWE, PAD, and gross snowfall were modeled using linear regression, and suggest that landscape models of snow accumulation can be created using PAD, independent of forest type, age, and site conditions.