[MITgcm-support] KPP and numerical stability

[Entcho Demirov]

Hi,

I am using the lab_sea set up and am trying to extend it towards
a regional North Atlantic model. When I use real topography, T/S
and surface forcing the code is stable only when useKPP=.FALSE.
When useKPP=.TRUE. the velocities become too large and the model
unstable. I have tried the two CPP options KPP_SMOOTH_VISC and
KPP_SMOOTH_DBLOC but the model is still unstable. What would be
the optimal way to improve the stability of the code with the kpp
option without smoothing too much the mixing coefficients?

Thanks

Entcho

----------------------------------------------------------------------
  Entcho Demirov                      e-mail: entcho at physics.mun.ca
  Physics and Physical Oceanography   Phone:  709-737-8834
  Memorial University
  St. John's, Newfoundland A1B 3X7    Fax:    709-737-8739
---------------------------------------------------------------------