Seasonal and diurnal cycles of liquid water in snow

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