Analysis of georadar data to estimate the snow depth distribution

•Methods for fast mapping of spatial variability of snow depth•Statistical analysis of traveltimes for snow depth modeling•Enhancing integration of hydrological modeling and georadar results

We have performed extensive georadar surveys for mapping the snow depth in the basin of Breuil-Cervinia (Aosta Valley) in the Italian Alps, close to the Matterhorn. More than 9 km of georadar profiles were acquired in April 2008 and 15 km in April 2009, distributed on an hydrological basin of about 12 km2. Radar surveys were carried out partially on the iced area of Ventina glacier at elevation higher than 3000 m a.s.l. and partially at lower elevation (2500 m–3000 m) on the gently slopes of the basin where the winter snow accumulated directly on the ground surface.The snow distribution on the basin, at the end of the season, could vary significantly according to the elevation range, exposition and ground morphology. In small catchment the snow depth reached 6–7 m. At higher elevation, on the glacier, a more homogeneous distribution is usually observed. A descriptive statistical analysis of the dataset is discussed to demonstrate the high spatial variability of the snow depth distribution in the area. The probability distribution of the snow depth fits the gamma distribution with a good correlation. Instead we didn't found any satisfactory relationship of the snow depth with the main morphological parameters of the terrain (elevation, slope, curvature). This suggests that the snow distribution, at the end of the winter season, is mainly conditioned by the transport phenomena and re-distribution of the wind action. The comparison of the results of georadar surveys with the hand probe measurements points out the low accuracy of the snow depth estimate in the area by using conventional hand probing approach only, encouraging to develop technology for fast and accurate mapping of the snow depth at the scale of basin.