bottom friction issue...

[jchanut]

It seems that there is a conflict between bottom friction and the way the surface pressure gradient is computed in NEMO. This is the case whatever the method used for the free surface (explicit, filtered or time split) and as long an implicit scheme is used for vertical diffusion (ln_zdfexp=.FALSE.).

In this case indeed, the bottom stress is computed implicitely through the dyn_zdf routine but before the call to dyn_spg so that the contribution of the pressure gradient is missing in the evaluation of the "next" bottom velocities. As a consequence the tendency due to the bottom stress is wrong.

In the explicit free surface case, moving the call to dyn_zdf after dyn_spg would solve the problem.

In other cases, it is more difficult to fix it, except if one use an explicit form of the bottom friction (at least for the barotropic contribution). It is likely that this would be much less stable (by doing this, my model blows up, but I probably made a mistake). If one would like to keep an implicit form of bottom friction, a possible conflict with the implicit computation of the filtering force will also arise...

In the time splitting case, a solution could be to explicitely add the bottom stress in the barotropic equations.

Jerome

[rblod]

At this time we add an explicit computation with control on the coefficient

Update of the documentation follows
Note: tke_old had not been updated

Rachid