Posted by Alan leonardi on December 18, 2001 at 10:09:26:
In Reply to: Poster comments posted by Rick Lumpkin on December 03, 2001 at 13:18:45:
Hi Rick. Great comments. Here is my response to comment #1.
Here we compare the flow in a series of similar simulations. All have the same horizontal resolution (1/16), and eddy viscosity (Ah=30 m^2/s). The main difference is in the vertical resolution: 1.5 layer linear reduced gravity, 6 layer finite depth, and 6 layer flat bottom.
The linearization process is not achieved through a change in the eddy viscosity. The nonlinear terms in the equations of motion are damped by multiplying them by a small coefficient (in this case by 1/100). This effectively removes the overall impact of the nonlinear features. As a result, the difference between the non-linear and linear reduced gravity experiments are primarily achieved via influences arising from non-linear features, or from the effects of the barotropic of higher baroclinic modes.
From the above, it shoud become clear that the linear runs are not more realistic, they simply provide us with a baseline experiment we can use to understand the differences arising from higher order dynamics. The corresponding boundary layer width scales in the linear and non-linear models correspond to the differing dynamics. In the linear model, the boundary layer width is the too wide (observationall) Munk frictional boundary layer, while the non-linear inertial boundary layer width scale more close mimics observations. Thus, it and the consistent NHRC speed represents the more realistic simulation. Indeed, the results match the observations very well (as Fred pointed out).
We also ran these simulations at differing horizontal resolutions. I can say thatwe did not compare the Alenuihaha channel flow in these simulations, but overall the flow was substantially improved in the higher resolution models. Part of this does stem from the fact that the higher resolution models more closely represents the topography of the islands themselves (as you allude to).
Given that the A-channel flow seems to be at least partially dictated by the non-linear dynamics, I don't think that increased horizontal resolution would provide the kind of change (0.5 Sv) that you are talking about, but without running the numbers, I can't confirm this.