Blowing up problem

[wzhu]

Hi Kate,

Thank you very much for your suggestion.

I have looked into the restart file and found the ranges of the following:
Salinity: -0.065-176.729;
Temperature: 0.1317-144.412 ℃;
Density anomaly: -19.1-210.97 kg/m3;
Free surface: -1.145-0.777 m;
u : -1.13857-2.08 m/s;
v : -2.213-1.2375 m/s.
Obviously, the salinity, Temp and density are abnormal, I also noticed that the the maximums appeared at the bottoms of two grids near the bay mouth almost since the beginning.

what should I check? the forcing files?
Thank you.

Wenwu

[kate]

Did you check your initial condition there? If that's clean, I would set up some STATIONS there to see how it goes bad. Is it two isolated grid points or a small area? If your STATIONS write every timestep, you can watch it go bad in time. Are you right on the edge of your timestep stability? Since it is early in the run, you can also try DIAGNOSTICS_TS and see if that's informative there.

[cflanary]

Hi Kate,

This topic has always been one I'm curious about and still feel unsure with diagnosing problems. The descriptions you gave the person that started the post are clear, but what does one do with the information they find out with the tests you suggested?

Is there a way to say definitively that ROMS blew up because of the bottom topography is too rough or if you're on the edge of the timestep stability or it's a boundary condition value.

Thank you,

Chris

[kate]

One of those is easy - change the timestep and see if the time of the trouble changes.

It's been a while since I've had an unambiguous blowup from steep topography. Steep topography causes unphysical currents, but not a blowup unless you have done no smoothing at all (there are those in here who try no smoothing at all...).

Troubles in T/S can be due to say vertical boundary conditions - surface heat/salt fluxes. I get troubles in u/v more commonly. Those can be surface, mid-depth or bottom, depending - I've had them all. I added print statements to diag.F to report whether troubles are in density or velocity.

The last time I had boundary troubles, it was a problem with my boundary file.

[wzhu]

Did you check your initial condition there? If that's clean, I would set up some STATIONS there to see how it goes bad. Is it two isolated grid points or a small area? If your STATIONS write every timestep, you can watch it go bad in time. Are you right on the edge of your timestep stability? Since it is early in the run, you can also try DIAGNOSTICS_TS and see if that's informative there.

Thank you, Kate.

I just checked the initial condition and boundary condition, and found that there were some strange errors with the parameters in initial condition file which probably caused the blowing up. They are two isolated grid points near the open boundary, but there seems to be not any abnormal values with the boundary condition. I corrected the values and let it run with timestep of 30 s now. How should I set the timestep to the right value? according to what?

[corvianawatie]

One of those is easy - change the timestep and see if the time of the trouble changes.

It's been a while since I've had an unambiguous blowup from steep topography. Steep topography causes unphysical currents, but not a blowup unless you have done no smoothing at all (there are those in here who try no smoothing at all...).

Troubles in T/S can be due to say vertical boundary conditions - surface heat/salt fluxes. I get troubles in u/v more commonly. Those can be surface, mid-depth or bottom, depending - I've had them all. I added print statements to diag.F to report whether troubles are in density or velocity.

The last time I had boundary troubles, it was a problem with my boundary file.

Dear All,

Is it still relevant to ask about this problem? I also have so much problem with the steep topography which produces unrealistically high current velocity (u,v,w). I tried to run using TS FIXED with no forcing, but the results are the same. Is there any suggestion to solve this problem?