Refreezing does not conserve column water content

[gampnico]

Issue: Refreezing does not conserve column water content when $\theta_{i} + \Delta\theta_{i} > 1$.

In the following snippet in refreezing.py:

theta_r = 0.0
[...]

if ((GRID.get_node_ice_fraction(idxNode) + dtheta_i + theta_r) >= 1.0):
    GRID.set_node_liquid_water_content(idxNode, theta_r)
    GRID.set_node_ice_fraction(idxNode, 1.0)
else:
    [...]

So when $\theta_{i} + \Delta\theta_{i}+\theta_{r} \geq 1$, the new liquid water content $\theta_{w}$ is set to $\theta_{r} = 0.0$, and the new ice fraction $\theta_{i} = 1.0$.

However, this means water is not conserved as the excess ice is discarded, therefore the layer regains less mass from refreezing.

This also leads to an overestimate for the refreeze parameter ($\theta_{i}h$) and an underestimate for the returned refrozen water ($\sum{-\Delta\theta_{w}z}$).

Possible Solutions:

• Thicken the layer height in response to the excess ice: this may create a discrepancy with subsurface melt.
• Convert excess ice back into liquid water, and percolate in the next timestep: this lets liquid water exist below freezing temperatures and requires yet another fix if the new $\theta_{w} \geq 1$.
• Refreeze during percolation (Wheler & Flowers, 2011.; Vionnet et al. 2012 etc.): this forces percolation.py to depend on refreezing.py.