An advanced liquid water flow scheme for firn models

For the Greenland ice sheet, runoff of liquid water has become the dominant factor of mass loss, however, the link between meltwater production and meltwater runoff remains unclear. Meltwater can be retained in the firn layer through long term storage in the liquid phase or by refreezing if the firn pack is cold. Liquid water in the firn layer has a large impact on its properties, for example: the firn densification rate is particularly affected by refreezing of percolating water, the temperature distribution by any associated heat transport plus latent heat released through refreezing, and grain size by wet snow metamorphism. As both the frequency and the intensity of melting and rain-on-snow events are expected to increase in the coming years, it is important that we are able to model these impacts in order to predict future Greenland ice sheet change with greater fidelity.

Vertical water transport through snow is achieved by homogeneous (matrix) flow and rapid flow through discrete 'pipes' (preferential flow). Such water flow has been documented extensively in field observations, but accurately representing liquid water flow in firn and snow models remains challenging. This is due to uncertainty in our understanding of key processes such as the distribution between matrix and preferential flow, the velocity of downward percolating water, the depth that liquid water can reach and the principles underlying the formation of preferential flow pathways. These processes play a crucial role in rapidly transporting water from the surface into deep, potentially subfreezing, layers and can affect feedbacks within the system e.g. on the structure of the firn.

To capture better these various effects in firn models, we implement a novel liquid water scheme in the Community Firn Model. Our approach is based on existing soil hydrology models accounting for the duality of water flow. We use observations from laboratory experiments of liquid water flow in snow to derive the parameters required in the model. From there, we aim at reproducing firn depth-density profiles, at constraining the potential of firn to retain meltwater and at predicting firn evolution according to climatic conditions. We assess the difference in performance between this advanced liquid water scheme and previous more simple approaches.