[MITgcm-devel] conservation in seaice_growth

[Martin Losch]

Hi there,

at my usual, incredible pace I have been trying to figure out conservation properties in seaice_growth (it only took me 3months). I finally can say/report this about conservation of volume and heat (did not look at salinity) in seaice_growth.F
1. The "legacy" code conserves volume and heat AS LONG AS there is NO advection NOR diffusion
2. The new code conserves volume and heat with advection, but not with diffusion

So the diffusion of sea ice is non-conservative (and I am not talking about scheme 2: advect.F and diffus.F, which is probably even worse). I assume that this has to do with the awkward scaling of the diffusion coefficient by "MIN(dxF,dyF)". As most of the implemented schemes work without diffusion anyway, I do not think that this is a problem, so I am not going to touch this for now.

The "legacy" code looses conservation in resetting the values to positive (and maximum) values without taking into acount the corresponding volume change in the fluxes (as the new code does). If the advection schemes were truly posivite, then this wouldn't be a problem, but apparently, they are not and negative HEFF, HSNOW and AREA do appear.

Contrary to my previous understanding, the latent heat of sublimation is already part of the game. But it's not used correctly: latent heat of sublimation should be converted to (negative) thickness change with the "latent heat of sublimation" = 2.834e6 J/kg, but it is converted with "latent heat of fusion" = 3.34 J/kg, just as all other heat fluxes, that convert ice to water (and vice versa). So either we change that (I we should, but will change results), or when coupling to atmosphere, this has to be kept in mind.

Martin