Improving a terrain-based parameter for the assessment of snow depths with TLS data in the Col du Lac Blanc area

- We compared a terrain-based parameter Sx to TLS snow depth data, at 1 m resolution.
- Using a DSM to calculate Sx improves results over using a DTM.
- When a DSM is used, changes in snow height can be approximated with linear regression.
- The slope of the regression line relates to magnitude of snow deposition.
- The search distance has substantial influence on the quality of Sx.

Wind and the associated snow drift are dominating factors determining the snow distribution and accumulation in alpine areas, resulting in a high spatial variability of snow depth that is difficult to evaluate and quantify. The terrain-based parameter Sx characterizes the degree of shelter or exposure of a grid point provided by the upwind terrain without the computational complexity of numerical wind field models. The parameter has shown to qualitatively predict snow redistribution with good reproduction of spatial patterns, but has failed to quantitatively describe the snow redistribution. By comparing the parameter with high-resolution snow surface data obtained through terrestrial laser scanning (TLS), we are able a) to identify areas of poor correlations between predicted and measured snow distribution and changes in snow depths, and b) to increase its ability to predict changes in snow depths by modifying the parameter, based on the TLS data and the terrain and wind conditions specific to our research site, the Col du Lac Blanc in the French Alps. We show how results improve if a snow surface model is used for calculating the parameter instead of a digital elevation model, and demonstrate the effects of changing the parameter's maximum search distance and of raster smoothing. Our analyses and results are important steps in the improvements of the parameter's ability to predict changes in snow depths.