[MITgcm-support] KPP and numerical stability

[Dimitris Menemenlis]

Entcho, are you sure that it is KPP that is to blame and not your initial T/S 
conditions?  When you turn off KPP, do you turn on an alternative convection 
scheme?  If problem is with your initial conditions, see related message by 
Martin Losch earlier today:
http://forge.csail.mit.edu/pipermail/mitgcm-support/2006-August/004202.html
That is, try to reduce time step until you get past initial instability and then 
you can increase time step once more.

Dimitris