Variations of snow surface and snowpack properties affect avalanche formation. In up to four north-facing slopes above the tree line near Davos, Switzerland, snow surface properties were characterized. Penetration resistance was measured with a snow micro-penetrometer. The sampling scheme was designed to allow a multi-scale approach covering the snowpack depth scale (0.5-5 m), the slope scale (5-100m) and the basin scale (100-1000m). Observations and measurements were compared to the data of a nearby automatic weather station (AWS). The AWS data were also used to model snow-cover stratigraphy and its evolution with the numerical snow-cover model SNOWPACK. Comparing the four slopes showed that surface properties observed manually were similar among the three slopes that were sheltered, and often different from the slope that was wind-exposed. However, the penetration resistance of the surface layer was in most cases significantly different among slopes, a although most values were <0.1 N, indicating very low hardness. These seemingly contradictory results follow from the different measurement support of the two methods. It is presently unclear which amount of variation at a given scale is relevant for avalanche formation. The geostatistical analysis and an analysis aimed at identifying the causes of variability were not conclusive. No patterns emerged that would allow conclusions regarding the effect on avalanche formation. Finding the causes of variability seems to require high-resolution terrain and weather models that are presently not readily available.