snow deposition

In the current implementation of SI3, snow deposition is disabled under the condition evap_ice < 0. This is due to a constraint in forced simulations, where the latent heat term qla_ice is not allowed to take negative values. In contrast, coupled simulations lift this constraint, thereby allowing deposition to occur. However, while the mass flux associated with deposition onto the ice/snow surface is correctly handled, the enthalpy of the deposited snow is not. Specifically:

• If snowfall occurs, the snow enthalpy ze_s(0) correctly adopts the enthalpy of the incoming precipitation.
• If no snowfall occurs, the enthalpy defaults to zero, which leads to an unphysical virtual temperature of Lfus/cpi (approximately 150°C?), due to the absence of any thermal energy in the deposited snow.

Proposed Solution: In cases of deposition without snowfall, a physically consistent value for ze_s(0) must be defined. Possible approaches include:

• Assigning ze_s(0) = rhos * Lfus, assuming the deposited snow forms at 0°C.
• Reusing the existing snow enthalpy: ze_s(0) = ze_s(1).

The optimal choice is still to be determined

See also the forum: https://nemo-ocean.discourse.group/t/ice-deposition-the-inverse-of-sublimation-in-coupled-simulations