problems in SSH

[hseo]

Hello all,
In my recent simulations SSH becomes unrealistically large values (like 100m). Although the temperature fileds look realistic, density and salinity look unrealistic. For example, salinity at surface changes from 28 psu to 38 psu.
I am using ROMS-2.0 and also turned OBC_VOLCONS option on to make sure mass is conserved. This seems to happen in several domains, indicating this is not domain-related problem either. Has anybody come across similar problem? Thanks!
Hyodae

[black]

I can't help you but I am having a very similar problem. My SSH goes to >150m then the model blows-up, all within half a model day. I am stumped.

[marcelcure]

I had similar problems with SSH in ROMS 2.1. I found that in the Make file, when compiling code with ifort, that no floating point checking was included as default. I added -fpe=0 to FFLAGS. What this does is it immediately aborts program execution when a floating point exception occurs, rather than set the floating point number to + or - inf, or to NaN - the user sees huge SSH's. You can then identify any problems with data input from NetCDF and rectify them, rather than see the problem after many more time steps and be completely in the dark.


Marcel Cure

[evan]

I had problems with unrealistically large SSH last year in ROMS2.1. The reason I found is that *_VOLCONS (* means N,S,W,E) is only activated if
*_M2NUDGING is also activated, (see globaldefs.h). In other words if you don't have *_M2NUDGING activated then *_VOLCONS is not activated, even though you might have flagged it in cppdefs.h.

Mark Hadfield indirectly drew attention to this aspect of *_VOLCONS in globaldefs.h in a post in ROMS Bugs entitled "Code to disable xxxx_VOLCONS", but unfortunately he never got any replies.

Anyway, either use *_M2_NUDGING, or edit the appropriate lines in globaldefs.h to ensure *_VOLCONS is switched on (but be aware that editing globaldefs.h is not recommended).

Hope this helps,

Evan