The combination of upward-looking ground-penetrating radar (upGPR), automatic weather station (AWS) and lysimeter allows for continuous monitoring of bulk volumetric liquid water content (�w) within the snowpack and direct comparison with measurements of the corresponding outflow. The AWS data can be utilized to calculate energy fluxes between atmosphere and snowpack at the location of the station. While combining all data sets, we were able to quantify diurnal and seasonal changes in residual water content and relate modeled energy fluxes to water outflow. Since upGPR is a non-destructive monitoring technique, it is possible to continuously observe the snowpack and results are not biased through spatial variability of pit locations. Data analysis conducted for three consecutive years at the flat test site Weissfluhjoch, Davos, Switzerland showed that diurnal �w variations never exceeded 2%. Without regard to days with new snow accumulation or refreezing, the diurnal patterns in �w were very similar, with always daily peaks in the late afternoon (at about 17:00h) at the site. Although �w values varied during a day up to 2%, the gradients during the season were very small. In 2012, for the whole melting period (>100 days), increases in �w from day to day were 0.4% liquid water content on average. After the snowpack has become isothermal, positive energy fluxes result in outflow and increase the residual water content (�r). Our data showed that as long as potential melt - calculated for the determined energy fluxes - was exceeding measured outflow, �r values were increasing but only until reaching a certain threshold. For all three years, the thresholds were similar at about �r=4–5%. Only shortly before full ablation, these thresholds were surpassed. In two sloped test sites (about 22 degree slope angle) in Boise, Idaho, USA and above Davos, we installed upGPR systems as well. AWS data and energy-flux calculations for both slopes were extrapolated for the respective aspect and slope angle. Our data showed that snow stratigraphy highly influences �r in slopes. As long as e.g. crusts ponded the vertical water flow, residual �w of the whole snowpack was fairly low (<1%) over weeks. Diurnal variations in �w were in accordance with values observed at the flat site. Once the ponding capabilities of the stratigraphy disappeared, residual �w values were in a comparable range to the ones of the flat site. The applied measurement setup is able to monitor continuously the behavior of liquid water in snow and record almost exactly the date when �r thresholds are reached or surpassed. Data thereof can be used to assimilate model outputs and may help to better predict water outflow and storage capacities.