[MITgcm-support] Re: pickup problem

[Tom Haine]

Dear Alistair and Jean-Michel, 

I think I've found the source of the pickup problem with my version of
the new horizontal viscosity: At pickup I removed the mean SSH from the
SSH field (to avoid problems with very stiff barotropic modes when
assimilating with an earlier version of this code - I should have cut
this a while ago, but I forgot: sorry if I wasted your time, J-M.). 

Without this adjustment (i.e., the regular code) I can integrate past
the pickup for several hundred more steps. Eventually, there's another
divergence however:

WARNING: r*FacC < hFacInf at: i,j,bi,bj,Thid,Iter=  71  20   1   1   1
5977
rStarFac,H,eta = -0.025522  5.000000E+00 -5.127608E+00
STOP in CALC_R_STAR : too SMALL rStarFacC !

in STDERR.0032. Again, I guess this is the Leith scheme diverging
despite the diagnostics about rStar (right?). I suppose my flow has
become too energetic (I see more or less monotonic increase in KE prior
to the crash). How should I control this?  Should I just increase
viscC4Leith (currently set to 1.0)?

Thanks again for your help. I agree with Dimitris - this is a fabulous
new scheme for the MITgcm.
Tom.