[MITgcm-support] [EXTERNAL] Re: Query regarding horizontal strips in currents (Menemenlis, Dimitris (US 329B))

[Martin Losch]

Hi Pragnya,

I have these comments.
1. with your grid resolution of 3.7km and deltaT=90. the dimensionless viscosity parameter viscAhGrid that you would need to get viscAh = 0.7 is 0.7/ (0.25*3700**2 / 90) = 1.84e-05 (see pkg/mom_common/mom_calc_visc.F). I would try something larger, e.g. viscAhGrid = 0.01 or 0.001. I am quite confident that the noise will away with this and you can use hFacMin = 0.1 again. In general I usually use the parameter viscAhGrid instead of vischAh. Also for this high resolution you may want to consider biharmonic viscosity to stabilise the flow, e.g. viscA4Grid = 0.01 ( or smaller ) and then you can use viscAh or viscAhGrid to add some “physical” viscosity (as small as you like).

2. vertical grid: as a rule of thumb the vertical grid spacing should not vary abruptly (to preserve order of approximation). I normally use vertical grid spacings that do not exceed the condition dz(k+1)/dz(k) < 1.4. You have biggers jumps of factor 2 and 5. That may also be related to your observation that increasing the hFacMin helps.

3. Not relevant for your stability, but the combination of no_slip_bottom and a drag parameter is not very physical in my view. The no-slip BC already introduces a drag so that you now add drag at the bottom twice.


Martin
> On 15. Jun 2023, at 08:30, Pragnya Makar <asz218003 at cas.iitd.ac.in> wrote:
> 
> Hi,
> Thank you for your feedback. Well, I have increased the hFacMin value from 0.1 to 0.7; which has significantly reduced the instability in the current simulation, as shown in the attached image. I will also investigate the lateral boundary conditions as you suggested. Additionally, I will examine the viscosity aspect.Also, please find the attached data file.
> 
> Regards,
> Pragnya Makar