A multi-dimensional water transport model to reproduce preferential flow in the snowpack

• We developed a multi-dimensional water transport model for a snowpack.
• Our model simulates the formation of preferential flow paths.
• Our simulations reproduce the result of laboratory experiments well.
• Sensitivity experiments show the importance of estimating heterogeneity.
• The model shows that the water entry suction is important for preferential flow.

A multi-dimensional water transport model for a snowpack was developed to reproduce preferential flow. Darcy–Buckingham's law and the van Genuchten model were used to simulate water movement. Parameters for the van Genuchten model were determined as functions of snow density and grain size, based on a previous gravity drainage column experiment. Water entry suction was newly incorporated to simulate liquid water infiltrating dry snow, based on the results of a previous capillary pressure experiment. Heterogeneity of grain size and snow density were also implemented. Heterogeneity alone was not sufficient for the development of preferential flow paths. When both water entry suction and heterogeneity were implemented, the model could simulate the formation of a preferential flow path. Validation of the simulation results by comparison with the results of a previous laboratory experiment showed that the simulated preferential flow path reproduced the experimental result fairly well. The size of the preferential flow path was negatively correlated with grain size and positively correlated with the water supply rate in both the simulation and the experiment. Comparison of the time series of suction between the simulation and the experimental result suggested that some additional measurements need to be obtained experimentally to improve the accuracy of the model. Sensitivity experiments showed that grain size and snow density heterogeneity affected both the size of the preferential flow path and the arrival time of liquid water at the snow base. These results suggest that more measurements of small-scale spatial variability in natural snow are needed to improve the model.