Physical Drivers of Thin Snowpack Spatial Structure from Unpiloted Aerial System (UAS) Lidar Observations

Abstract:

Snow depth spatial variability is a function of interactions among static variables, such as terrain, vegetation, and soil properties and dynamic meteorological variables, such as wind speed and direction, solar radiation, and soil moisture that occur over a range of spatial scales. Unpiloted Aerial System (UAS) lidar allows for the ability to map high spatiotemporal resolution snow depth observations at watershed scales with a high degree of vertical accuracy. Using modest cost, commercially available components, UAS lidar surveys were used to map snow depth and vegetation with high vertical and horizontal resolution in open terrain and mixed coniferous and deciduous forests at the University of New Hampshire’s Thompson Farm Research Observatory, New Hampshire, United States. Despite the shallow, ephemeral snowpack, snow depth exhibited coherent spatial patterns. This spatial structure was analyzed using parametric and nonparametric methods. Maximum entropy and random forest models are used to identify the most important static and dynamic variables that explain the location of the shallowest and deepest snowpacks as well as the consistency of snow depth patterns across seasons. The results indicate that vegetation type and terrain roughness contribute most to predicting variations in snow depth across the landscape. In open fields, soil characteristics and illumination were also important controls on spatiotemporal snow depth variations.