Tree sway monitoring for improved representation of canopy snow interception in cold, wet climates

In cold and wet regions, the forest canopy strongly influences the energy and mass balance of the snowpack by intercepting a large fraction of solid precipitation. Although commonly represented in land-surface models, snow interception remains poorly documented in the field because of the difficulties associated with directly measuring intercepted snow mass. Existing indirect measurement approaches include the use of accelerometers to quantify wind-induced tree sway and relate changes in sway frequency to variations in intercepted snow mass. However, this experimental method has so far been applied at only one site in the western United States, under climatic conditions that differ from those of the boreal forests of eastern Canada.

The objective of this study is to apply the tree sway method in eastern Canada to estimate intercepted snow mass and to improve the parametrization of canopy snow interception in the SVS2–Crocus land surface model.

A total of nine coniferous trees were equipped with accelerometers across three sites to monitor wind-induced tree sway and obtain estimates of intercepted snow mass. Results from the winter 2024–25 show that the sway method captures rapid loading and unloading events, with sway frequency responding to interception and release within a few hours, in agreement with hourly timelapse imagery acquired at each site. The resulting intercepted snow time series is then used to evaluate the canopy interception parametrization in the SVS2–Crocus model, which was forced using in situ meteorological measurements.

Sway-based observations and simulated intercepted snow mass show good agreement in the timing of interception and unloading, with rapid increases during snowfall and subsequent exponential decay. However, the model tends to overestimate slow, continuous unloading and often fails to accurately reproduce rapid unloading events associated with strong winds, warm temperatures, or rain-on-snow events. These results pave the way for improving the parametrization of canopy snow unloading in SVS2–Crocus and, in turn, for more accurately estimating snow cover in forested environments.