[Mitgcm-support] more KPP questions

[mitgcm-support at dev.mitgcm.org]

Hi,

I just browsed the questions about KPP of the last year. Some of these 
questions are relevant to me, so I'll asked them again, hoping that for 
some of these issues there are already answers. I need KPP in 
conjunction with
---GMredi, for that I plan to steal the parameters from the natl_box 
experiment, or the lab_sea experiment (although that's not quite as 
succesful as I had hoped)
---passive tracer advection, and because my tracers are going to be 
biological quantities, I probably need a positive multidimensional 
advection scheme, e.g. the 2nd order flux limited scheme (btw, how do I 
turn that on? tracerAdvScheme=2,77?, multiDimAdvection=.TRUE.?)
---passive tracer advection with KPP. I noticed that Tom's remark about 
STOP 'GAD_CALC_RHS: Ooops' is still valid. Has anyone got a fix for that 
already? Should look exactly like kpp_transport_s.F except for the 
surfaceTendencyS term and kappaRS, shouldn't it?
---most likely non-hydrostatic code (Jean-Michel, how are the non-linear 
free surface + non-hydrostatic code efforts? z^*?)

Will these components work with KPP? I am not sure, but it would save me 
a lot of time if anyone can tell what should already work and what 
shouldn't (GMredi+passive tracers+multidimensional positive advection 
scheme+KPP+non-hydrostatic).

Right now, I am playing with KPP with the global_ocean.90x40x15 
experiment (i.e., without all these fancy add-ons) but the model 
explodes after 20 timesteps, after the Southern Ocean becomes vertically 
homogeneous. Before this homogenization spreads across the equator, the 
model says "boom", with or without GM, looped cells, no_slip conditions.
Is is foolish to use KPP with the coarse vertical resolution in this 
experiment (surface layer is 50m)? Do I need to modify the time step for 
KPP?
(I have increased the overlap OLx,OLy to 4, or alternatively set #define 
FRUGAL_KPP, btw, what does that stand for? "frugal" in German is some 
sort of vegetable stock/broth)

Martin