Numerical error

Discussion of how to use ROMS on different regional and basin scale applications.

Moderators: arango, robertson

Post Reply
Posts: 23
Joined: Wed Feb 28, 2007 8:00 pm
Location: Massachusetts Institute of Technology

Numerical error

#1 Post by logvinov » Wed Sep 26, 2007 3:53 pm

Hi All,
I consider a simple case of wind driven flow in the channel (Wall and periodic BCs) with sloping bottom (cross the channel). So there should be no variation in the flow along the channel. But the output of the run demonstrate significant variations in the flow along the channel.

I believe there is some kind of numerical error. I use domain depth x Ly(cross the channel) x Lx : 70-73m x 250m x 100m with number of points respectively 25 x 80 x 10.

The flow near the shallower part seams to demonstrate almost x-independent behavior and it is the deeper part that have strong along the channel variations. (Looks like it seats near the periodic boundaries occupying 2 grid points and then decreasing towards the area in the middle)

Could you give me any suggestions on this?

Thanks a lot,

User avatar
Posts: 3678
Joined: Wed Jul 02, 2003 5:29 pm
Location: IMS/UAF, USA

#2 Post by kate » Wed Sep 26, 2007 4:36 pm

It sounds a lot like the UPWELLING case which has been invariant along-channel when I've checked. Do you have more terms turned on than in UPWELLING? Is this serial or parallel? Is UPWELLING behaving for you? One thing to do to confirm your suspicions is to turn on the STATIONS option and put a line of stations across the periodic join. Then you can start playing games with turning off various terms until it behaves itself and you can tell us where the trouble lies.

Posts: 798
Joined: Wed Dec 31, 2003 6:16 pm
Location: USGS, USA

#3 Post by jcwarner » Wed Sep 26, 2007 4:57 pm


I copied the setup files you sent and i see the disturbance that you are talking about. I changed
#define MIX_GEO_UV
#undef MIX_GEO_UV
#define MIX_S_UV

and it seems to work correctly now.
Give that a try and post a reply here.
This should allow you to continue while we dig deeper into mix_geo_uv.

Posts: 23
Joined: Wed Feb 28, 2007 8:00 pm
Location: Massachusetts Institute of Technology

#4 Post by logvinov » Thu Sep 27, 2007 8:11 pm

Thanks a lot for the suggestions!

Now it does work when switched to
#define MIX_S_UV

Post Reply