[Mitgcm-support] Re: Cannot reproduce Gaussian bump problem in Adcroft et al(1997)

[mitgcm-support at dev.mitgcm.org]

Yu-Heng Tseng wrote:
> 
> Hi
> 
> Does anybody reproduce the Gaussian bump problem in Adcroft et al. (1997)
> successfully? I still cannot get the 10th day relative vorticity field as
> reported in the paper. I couldn't get the large nearly circular cyclonic
> eddy which is very close to the downstream boundary.
> I don't know why.

I've modified exp4 as described in Adcroft et al., '97, and placed it at
http://mitgcm.org/~adcroft/expts/. After running the model the m-script
input/plotresults.m plots something like that shown in paper. I don't recall
how we calculated depth integrated vorticity - so I'm guessing. Perhaps your
diagnostic will be closer.

This set-up does not use OBCs: the position and shape of eddy is dependent
on the presence of channel walls. The structure of the tail is different and
this is likely because the discretization is different and I'm using a
free-surface here. I may also not be initializing exactly as we did in paper.
Nevertheless, the solution is essentially the same as in paper.

Radiation conditions do work but the eddy spins off earlier and leaves the
domain by day 10.

A.