Help on ROMS Blowing before even running first time step

[shikhar.rai]

Hello everyone. I am new to ROMS. I set up my initial and boundary conditions and tried to ROMS, but I got the following error. It ran a few hundred time-steps before, but now it does not even start. Attached is the log file. Any help is really appreciated.

[bilge.tutak]

Hi,
Although it says blow up I think there is something else.

- At your initial time step, Potential Energy and Total Energy is NaN?
- You have many MPI/Intel Fortran Library issues (as far as I can see many warnings on that).
- You max speed at time 0 is ~1.5 m/s (which is quite high for start up, unless you have a really good balance and initial condition),

You should check your grid and input files. Find what causes the NaN.

Good luck,

[wilkin]

That looks like an issue with your libraries or compiler. If it were just ROMS the blow-up message would give more guidance as to where in the code the issue arose. This is an outright crash. Make sure you are linking to MPI libraries built with the same compiler you use for ROMS.

That said, the NaN at the start would seem to implicate something bad in the initial conditions, though as has been noted your initial speed is quite high. You could just set initial velocity to zero and see if that makes a difference (inertial oscillations will quickly bring velocity into consistency with density).

By the way, your grid dimensions 1021 x 349 are both prime numbers, so you are never going to have even sized parallel tiles. That's why they all have different sizes ...

Tile partition information for Grid 01: 1021x349x30 tiling: 6x6

tile Istr Iend Jstr Jend Npts

0 1 169 1 57 288990
1 170 340 1 57 292410
2 341 511 1 57 292410
3 512 682 1 57 292410
4 683 853 1 57 292410
5 854 1021 1 57 287280
6 1 169 58 116 299130
7 170 340 58 116 302670
8 341 511 58 116 302670

It's not wrong, just not optimally efficient use of the parallel tiling.

[robertson]

You might also consider updating your ROMS source code; the revision you are using is over 3.5 years old.

[shikhar.rai]

Thank you everyone for your help. It was spurious NaN values of salinity at some points in the IC file. When I changed that to the nearest value of salinity, it worked.