[MITgcm-devel] conservation in seaice and ocean

[Martin Losch]

Hi Gael,

the diffusion is a mess, because when I implemented it tried to stick closely to the original advect/diffus routines. These routines are only sensibly used for adjoint simulations and for a long time they where only ones one could use with the adjoint. This has changed, right? Do we still need the second order central difference scheme for seaice?

(1) If not, I suggest to remove it entirely (along with diffus) and replace "diff1" with something more sensible (e.g., SEAICE_diffKh) that defaults to zero, at the same time the strange factor "MIN( _dxF(I,J,bi,bj), _dyF(I,J,bi,bj))" vanishes.

(2) If we want to the 2nd order scheme (more likely), I still suggestion to replace diff1 and remove the weird factor, replace advect with something more useful (e.g. call a routine, that implements just the flux-form of the advection and then call seaice_diffusion)

(3) If we want to keep advect/diffus as it is for backward compatibility, then I still would like to fix seaice_diffusion, introduce new flag and have the old default of diff1 only applied if the advection scheme is 2.