It is important to realize that snow is really a three phases medium: it might contain at the same time water in its solid phase (the ice crystals matrix), in its liquid phase (interstitial water) and in its gaseous phase (water vapor in the pore space). As the liquid water can move through the ice matrix, it transports mass as well as potentially energy. Depending on the conditions, it might also significantly alter the microstructure of the snow pack. Therefore it is very important to be able to simulate how this liquid water moves through the snow layers.

Matrix flow and preferential flow

Liquid water can move through the snow pack in two distinct ways: through matrix flow or through preferential flow. They are fundamentally different and have very different time scales.

Matrix flow

Matrix flow represents how the liquid water moves through the pore space of the ice matrix. This capillary motion is dominated by surface tension effects. Such a flow is highly dependent on the tortuosity of the ice matrix, the water column pressure head and changes of such properties which can lead to capillary barriers. The kind of flow moves the bulk of the mass in a gradual process.

Preferential flow

A second kind of liquid water transport mechanism is through preferential flow. This is a 2 dimensional effect where at some places the liquid water is able to locally flow much deeper into the snow pack. Although highly relevant for its impact on the snow microstructure and for its impact on snow stability, this transport mechanism only carries a minority of the liquid water mass. But by providing liquid water in deeper layers much faster, it can contribute to triggering a weak layer or accumulate liquid water at a capillary barrier (ponding) that could later refreeze and build an ice layer.

Modeling

In Snowpack, water transport can currently either be modeled with the bucket approach or by solving the Richards equations.

In the bucket approach, each snow layer has a given water storage capacity that can be filled by liquid water (thus similar to a bucket) and then overflows down to the next layer when full. This is computationally efficient but not a very accurate representation of the physical phenomenons involved in the liquid water transport.

On the other hand, the Richards equation describes the flow of a liquid in a porous media and is therefore a much more adequate representation. The novelty of the Richards equation solver in Snowpack is to use such equations in a media where the matrix is just a different phase of the liquid. This is computationally much more challenging than the bucket approach and needs to be much more carefully configured.

Configuration keys