Articles | Volume 10, issue 9
Geosci. Model Dev., 10, 3547–3566, 2017

Special issue: The externalised surface model SURFEX

Geosci. Model Dev., 10, 3547–3566, 2017

Model description paper 26 Sep 2017

Model description paper | 26 Sep 2017

Implementation of a physically based water percolation routine in the Crocus/SURFEX (V7.3) snowpack model

Christopher J. L. D'Amboise1,2, Karsten Müller1, Laurent Oxarango3, Samuel Morin4, and Thomas V. Schuler2 Christopher J. L. D'Amboise et al.
  • 1Norwegian Water Resources and Energy Directorate, Oslo, 0368, Norway
  • 2Department of Geoscience, University of Oslo, Oslo, 0316, Norway
  • 3Univ. Grenoble Alpes, CNRS, IRD, IGE, 38000 Grenoble, France
  • 4Météo-France – CNRS, CNRM UMR 3589, Centre d'Etudes de la Neige, Grenoble, France

Abstract. We present a new water percolation routine added to the one-dimensional snowpack model Crocus as an alternative to the empirical bucket routine. This routine solves the Richards equation, which describes flow of water through unsaturated porous snow governed by capillary suction, gravity and hydraulic conductivity of the snow layers. We tested the Richards routine on two data sets, one recorded from an automatic weather station over the winter of 2013–2014 at Filefjell, Norway, and the other an idealized synthetic data set. Model results using the Richards routine generally lead to higher water contents in the snow layers. Snow layers often reached a point at which the ice crystals' surface area is completely covered by a thin film of water (the transition between pendular and funicular regimes), at which feedback from the snow metamorphism and compaction routines are expected to be nonlinear. With the synthetic simulation 18 % of snow layers obtained a saturation of  >  10 % and 0.57 % of layers reached saturation of  >  15 %. The Richards routine had a maximum liquid water content of 173.6 kg m−3 whereas the bucket routine had a maximum of 42.1 kg m−3. We found that wet-snow processes, such as wet-snow metamorphism and wet-snow compaction rates, are not accurately represented at higher water contents. These routines feed back on the Richards routines, which rely heavily on grain size and snow density. The parameter sets for the water retention curve and hydraulic conductivity of snow layers, which are used in the Richards routine, do not represent all the snow types that can be found in a natural snowpack. We show that the new routine has been implemented in the Crocus model, but due to feedback amplification and parameter uncertainties, meaningful applicability is limited. Updating or adapting other routines in Crocus, specifically the snow compaction routine and the grain metamorphism routine, is needed before Crocus can accurately simulate the snowpack using the Richards routine.

Short summary
We present a new water percolation routine added to the Crocus model. The new routine is physically based, describing motion of water through a layered snowpack considering capillary-driven and gravity flow. We tested the routine on two data sets. Wet-snow layers were able to reach higher saturations than the empirical routine. Meaningful applicability is limited until new and better parameterizations of water retention are developed, and feedbacks are adjusted to handle higher saturations.