Using lidar to estimate snowpack formation under mixed-species forests in northern Wisconsin

Measurements under varying degrees of canopy closure show that the snowpack beneath a forest canopy can have 33% to 80% less snow than the snowpack in an opening. Much of the broad variation in the snow water equivalent (SWE) of the snowpack gradient measured under different forest stands can be attributed to stand-specific canopy structure. Differences in snow accumulation between stands and, by inference, differences in the interception capacity of stands, have been modeled using a variety of different measures of stand structure, but most often rely on canopy coverage. While these models have had good correlations with SWE they are limited to conifer stands with highly homogeneous age and species composition and have not been tested on a wide range of forest types or ages. We suspect that as more forest types are examined that have comparable canopy closure (or even LAI), but actually have very different three-dimensional (3D) canopy structures, these variables will be less related. We propose 3D structure should be an important factor for determining the amount of snow load that a canopy can carry. The more snow and the longer that it is in the canopy, the more water lost to ablative forces and the less on the ground. Because measurements of 3D structure are difficult to make in the field, time consuming, and virtually impossible to do for a large area, they typically have not been measured across large spatial extents. However, this has changed with the availability of airborne lidar platforms. We made biweekly snow measurements at 143 plots across 9 forest types in northern Wisconsin during the winters of 2007 and 2008 to assess how canopy structure controls snowpack formation. The measurements have indicated a decrease in SWE by as much as 55% as canopy density increases. These data are being correlated to multiple return discrete lidar data that includes return intensity that were collected for this site. We will use these data to create models that relate 3D forest structure to snowpack formation, and to demonstrate that snowpack SWE beneath heterogeneous forest cover can be modeled at a high resolution (20m) across a large (400 m2) spatial extent.