Surface hoar growth detection using a terrestrial laser scanner

Dry snow slab avalanches, responsible for over 90% of avalanche-related deaths in Switzerland, are triggered by the failure of weak layers in between cohesive snow slabs. Surface hoar, a prevalent weak layer, forms under clear skies, cold conditions with minimal wind, as outgoing longwave radiation cools the snow surface, creating a temperature gradient that facilitates water vapour deposition. Using remote sensing methods to detect surface hoar crystals could be beneficial to mountain resort operators in order to characterise snow surface grain type of potentially dangerous and inaccessible slopes. Mapping the spatial variability of surface hoar at a slope or basin scale, could help relate the resulting patterns to the snow and weather conditions which could potentially improve avalanche forecasting and current models. This study aims to replicate and apply the findings of Walter, Brouet, and Jaggi (2023) to field data in order to assess the potential of their method in segregating surface hoar events from non-surface hoar events at theWeissfluhjoch (WFJ) site above Davos. The hypothesis is that, depending on the surface hoar crystal orientation, some of the incoming laser beams will undergo stronger specular reflection (upon hitting flat crystal surfaces), thus modifying the bidirectional reflectance distribution function from a typically diffuse (Lambert) reflectance distribution to a distribution with a larger specular component, which increases the reflectance variability (rstd). Using laboratory experiments, we confirmed that LiDAR rstd increases with increasing crystal size. In addition to that, we showed that the same opposite trend is also valid during sublimation events. We also investigated the dependence of rstd values to the incidence angle at which the surface is scanned. We found that the rstd has a minimum around 40°, but the underlying cause could not be conclusively determined. The segregation potential using the WFJ scans did not prove to be very efficient. We used the slope of the rstd values during each night to identify whether surface hoar crystals were growing or not. The gap from laboratory to field was too large to obtain satisfactory results. The main reasons for this non-achievement were: unreliable ground truth (even enhanced with the model SNOWPACK), precipitation particles acting similarly to surface hoar in terms of rstd and potential effect of scanner-surface distance and incidence angle. In addition to the method suggested in (Walter et al., 2023), we detected surface hoar crystals by analysing the microscale surface roughness, calculated from the residuals resulting from fitting a polynomial to the scans. This method proved to be much more successful and displays great potential. It demonstrates a distinct difference between the residuals from scans during a surface hoar event and those from non-surface hoar events. Provided with an airborne LiDAR system, this method could be used to map surface hoar spatial var