ice embedded/vvl in NEMO3.5

[dupontf]

Hi all,

I have a philosophical problem with the notion that the ice mass-embedded in NEMO is an option independent of VVL.

Right now in NEMO, VVL is associated with the exact salt flux approach (versus the virtual salt flux approach of the linear free surface). However, the exact salt flux approach assumes that the associated change in water volume is taking place: water is removed during ice formation and added during ice melt. But if mass is removed from water in order to fill the ice compartment above, should it not be accounted in the surface pressure? If yes, this means that the mass embedding is occurring whenever true salt flux is considered, i.e. mass embedding is not an option in VVL. On the other hand, I don't see why virtual salt flux could not be applied in VVL: it would be equivalent to having the levitating ice over an otherwise level-varying ocean. For example: I want to do high-frequency dynamics over some shelf but don't care too much about the ice-ocean interactions.

Comments?

Fred.

[dupontf]

OK, thanks to tips from Chris Harris, I am getting the idea that nn_ice_embd was indeed intended to work independently of having or not VVL.

As an addendum, I would like to raise the issue that, from the point of view of nn_ice_embd, only LIM2 is doing the right thing by introducing a switch (depending on nn_ice_embd) that includes or not the mass exchange between ice and ocean in EMP, and do a true salt or a virtual salt flux accordingly.

CICE and LIM3, not having such a switch, includes the mass exchange in EMP, regardless of nn_ice_embd and vvl. In the non-vvl case, this means that the mass exchange is "on" although a virtual salt flux is applied. To me, this is inconsistent.

So, the best would be to include the simple switch present in limsbc_2.F90, in LIM3 and CICE. Things would be at least consistent from the point of view of nn_ice_embd, regardless of what vertical coordinate one thinks is appropriate to his problem.

Fred.

PS: This said, I need Gurvan to explain to me how nn_ice_embd=2 and non-vvl is viable. To me, the presence of the artificial term wn*tsn in non-vvl tracer conservation forbids any high frequency or large amplitude signal at the top surface, such as ice melt signals (in this case wn~emp/rho).

PPS: and since I am it, the meaning of nn_ice_embd=1 fails me. Why having a true salt flux and mass exchange but not the associated pressure gradient? This is the base of the original ticket.

PPPS: there are going to be fun issues with OBC in regional models with nn_ice_embd=2! We will need to know if the external conditions were based on a model and if yes which nn_ice_embd was applied, or steric height...