[MITgcm-support] problem with KPPghat term being too large

[Dimitris Menemenlis]

Dear MITgcm-ers, has anyone ever experienced problems with the KPPghat 
term being too large?

Specifically in kpp_routines, there is following line:
       ghat(i,ki) = (1.-stable(i)) * cg / max(phepsi,tempVar)

In tracking down cause of a crash in a hi-res (llc2160) simulation, I 
found that the value of phepsi=1e-10 can lead to some very large 
non-local transport values.

If someone has experience with this issue and suggestion how to fix, 
please let me know.  In meantime, I will experiment with larger values 
for phepsi.

Thank you,

-- 
Dimitris Menemenlis