Snow is one of the most complex materials occurring in nature. In clouds, it can be seen either as a crystal with various shapes or as a freshwater resource that will fall down to the Earth's surface. Once on the ground, it can be described as a thermodynamically unstable matrix or as a layered granular material with inherent structural weaknesses that ultimately result in avalanches. At high elevations or latitudes, snow slowly transforms into pure ice and supplies mass to glaciers and polar ice caps. Finally, it can also be seen as a relatively warm material close to its fusion point that will experience a phase change, melt, and infiltrate into the ground, ultimately feeding a river system. In this dissertation, we investigate several aspects of the latter definition by following the water particles from the hillslope to the river network. The playground of these experiments is a high Alpine catchment, the Dischma river basin in Switzerland. By combining a physically based and spatially distributed snow model with experimental data (snow lysimeter, snow depth, discharge), we examine, in the first two chapters, the influence of the spatial variability of snow and liquid water transport within the snowpack on runoff dynamics. The analysis, conducted from point scale to watershed scale, highlights the importance of having a realistic snowpack at peak accumulation to accurately model the hydrological response at the basin outlet. Additionally, we show how the energy fluxes driving the snow ablation change during the course of a year. In the third chapter, we put the modeling part aside and investigate experimentally the spatio-temporal variability of snow and its melt. Using an ultra-long range Terrestrial Laser Scanner (TLS), we measure the evolution of the snow cover at a very high spatial resolution on different hillslopes of the upper Dischma valley during the 2015 ablation season. Data analysis reveals that the ablation dynamics at the slope scale follow a bi-modal distribution of ablation rate with diverging behavior during the course of the melt season. The emergence of this bimodality is explained on the basis of associated limiting factors: mass and energy. To conclude this dissertation, we move away from the cryospheric world but still study the energy exchanges between the Earth's surface and the atmosphere. Towards this goal, a low-cost sensible heat flux sensor was developed. The instrument was tested and validated against a state-of-the-art reference. The sensor shows promising results by giving good estimates over different surface types (grass, gravel). In the future, it can be used to measure the spatial variability of sensible heat flux within a Wireless Sensor Network.