Unexpected ROMS blow-up

[lanerolle]

I am running ROMS and it suddenly blows up saying :

Blowing-up: Saving latest model state into RESTART file

WRT_RST - wrote re-start fields (Index=1,1) into time record = 0000001

When I look at the contents of the restart file or the screen output log file (I also write-out the maximum CFL number and its u-, v- and w- components), I do not see anything abnormal. Hence, I am totally confused as to why ROMS is blowing up.

My application only uses the Nonlinear model and I am not using any 4DVar, Tangent, Representer, etc. options and is hence quite a straightforward and simple application.

A reduction in time step causes it to blow-up some time later also.

What is the criterion built into ROMS to detect a blow-up?

[nganju]

I've found the best way to diagnose a blow-up like the one you describe is to save restart output very frequently and look at the fields of zeta, u, v; and also the change in those fields between the last time step and the second-to-last time step. This usually shows you where problems may be.

[kate]

You can look in diag.F to see the criterion for blowing up. I've added code there to make it more verbose about its blowing up.