Liquid water flow in snow and hydrological implications

Abstract

The presence of a snow cover has a strong impact on hydrological processes. In this thesis, the role of the snow cover as an interface between the atmosphere and the soil is assessed. An implementation of a solver for Richards Equation (RE) for water flow in variably saturated porous media in the one-dimensional, physics based snow cover model SNOWPACK is discussed, to solve water flow in snow and soil. The updated SNOWPACK model was validated by comparing simulations for the sites Weissfluhjoch (WFJ) and Col de Porte with an extensive data set of meteorological and snowpack measurements. Solving RE for snow appeared to improve the estimation of liquid water runoff from the snowpack, especially on sub-daily time scales and in deep, stratified snow covers. The onset of snowpack runoff in spring was better predicted when using RE instead of a bucket-type approach. On the sub-daily time scale, the timing of peak runoff was predicted more accurately using RE. For WFJ, the representation of the internal snowpack structure was assessed using snow profiles, snow temperature measurements and data from upward looking ground penetrating radar. In the main winter season, a high agreement was found for the temporal evolution of snow water equivalent (SWE), snow density, temperature profiles and, to a lesser extent, grain size and type. In the melt season, RE seems to provide a better simulation of the movement of the meltwater front through the snowpack than the bucket scheme, although large discrepancies between simulations, radar data and snow lysimeter measurements remain. The depletion rate of SWE in spring appeared to be overestimated in the model. Strong indications were found that the snow densification process in spring is not adequately simulated, as snow density in spring was consistently underestimated. The soil module of SNOWPACK was verified by comparing soil moisture measurements from 7 stations in the area of Davos, with distributed SNOWPACK simulations using the spatially explicit Alpine3D model. Soil moisture was simulated with varying degrees of success. By investigating cross-correlations between the different sites for measurements and Alpine3D simulations, it was found that the spatially distributed simulations were able to explain a part of the spatial variability in the measurements. This could be mainly attributed to the spatial variability of the snow cover and snow melt. Preliminary results also suggest that streamflow simulations in Alpine3D could benefit from accurate soil moisture simulations. Finally, SNOWPACK was used for simulations of 14 sites in two areas in Switzerland that encountered severe flooding in October 2011, when a large snowfall was followed by a significant warming and rainfall. Thick snow covers had significant capacity for liquid water storage, creating a time lag of several hours between the onset of rainfall and the onset of snowpack runoff, whereas shallow snow covers reacted much quicker. Interestingly, once thick snow covers produced runoff, peak runoff rates exceeded runoff rates from shallow snowpacks and also exceeded the sum of rainfall and snowmelt rates. This suggests that snow settling, wet snow metamorphism and liquid water flow provide an interaction with each other that potentially can enhance snowpack outflow. This model result could have important consequences for understanding the contribution of snowpack runoff in flood events.