parameter for lateral boundary nudging?

[Mike Y.]

Hello HYCOM,

I am running a ATL0.10 experiment with southern and northern ends of the domain chopped off. I use the same baroclinic and barotropic time steps as the GLB0.08. However I keep getting "neg.dp (m)" errors near the southern or northern boundaries after a few days of run from a cold start. My first instinct was an violation of CFL criterion. So I reduced the time steps a little, and the model ran a little farther before blowing up again.

Since the "neg.dp" error mostly occurs near the south/north boundaries, I figure it might have somethign to do with the "Boundary Nudging" I defined in "ATLy0.10/relax/010/relax_rmu.com". In the "ATLb2.00" experiment, 'N and S boundary: 5 grid points with 20 to 120 day e-folding time' is used. But I guess when the resolution/domain size change, those numbers need to be changed accordingly.

I was just wondering what is the general rule for choosing the number of grid points and the e-folding time for the boundary budging in accordance with the domain size/resolution and run time?

Thanks a lot.

Mike Y.

[alan.wa...@hycom.org]

In ATLb2.00 5 grid points is 10 degrees.  This (5 grid points) is the absolute minimum practical relaxation zone but 20 grid points (which on a 0.10 grid is 2 degrees), or more, is probably optimal and at least 10 grid points should be used if at all possible.   The efolding strength is also usually higher than in ATLb2.00.

For ATLa0.08 we used:

N and S: 102 and 43 gps (~3 deg) with 10 to 50 day e-fldng time

The other possibility is that the rmu file is not correct.  It is very easy to get the maximum relaxation strength at the ocean facing edge of the relaxation zone rather than the closed boundary facing edge.           
Alan.