No slip vs. free slip around masked land areas

[debcox]

Just wondering whether I should expect much difference in the results around masked land areas using no slip (gamma2=-1) as opposed to free slip (gamma2=1) boundaries.

For a 2D test case - a 1km diameter cylindrical island in 100m water depth with uniform 50m grid spacing in a channel with a north to south flowing current - the velocity results look identical for the two land boundary conditions. The psi mask seems to be specified correctly in each case.

Perhaps it depends on whether the land boundary is straight and follows the direction of the grid or is staggered (as in the case of my cylindrical island)?

thanks,
Deborah

[kate]

My dad told me of someone else's experience in which a staircase mask acts as a no-slip boundary. I must say most of our free-slip/no-slip tests were done in rectangular basins, no masking.

[evan]

I've just completed a 2d island configuration, but my results for no-slip vs free-slip are different, and as expected: no-slip results in eddy generation, free-slip not. My island presents a staggered boundary to the oncoming flow.

Evan

[apaula_oceano]

I have the exact same issue as Deborah. I have two simulations, being the only difference between them the gamma2 variable in the .in, with values 1 and -1. And the results from both look exactly the same. Should I so something else, besides setting gamma2 to -1, to configure a no-slip simulation?
I am also using the masking option on the .h. Does that make a difference at all?

Thank you very much,

Ana

[kthyng]

In Utility/metrics.F, there is a code comment (at least in an older version I am looking at) that says

!-----------------------------------------------------------------------
! Set slipperiness (no-slip) mask at PSI-points.
!-----------------------------------------------------------------------
!
! Set no-slip boundary conditions on land-mask boundaries regardless of
! supplied value of gamma2.

This is also mentioned on the wiki: https://www.myroms.org/wiki/index.php/N ... Land_Areas (under Velocity Points).

My understanding is that masked areas are hardwired to be no-slip.

Kristen

[wilkin]

The no-slip option (gamma2 = -1) can only be activated along a model domain perimeter boundary (nominally i=0 or L, j=0 or Mp). It has no effect on interior land/sea boundaries at the edge of masked land. Search the code in ROMS/Nonlinear and you will see how limited is the use of gamma2.

All land/sea boundaries defined by masking are free-slip.

If you look in the code at ROMS/Nonlinear/uv3dmix2_s.h and reconcile this with Wiki ROMS entry for the stress tensor at https://www.myroms.org/wiki/index.php/Horizontal_Mixing you will see that some of the contributing terms are evaluated at tracer (or "rho") points and some at vorticity (or "psi") points. The "psi" points fall on the land/sea boundary.

[apaula_oceano]

Thank you very much, Kristen and wilkin.

However, I'm still confused.

On my Utility/metrics.F there is the same code comment as Kristen pointed out - so the masked areas would be hard-wired to be no-slip.

But on the link that she posted,

https://www.myroms.org/wiki/index.php/N ... Land_Areas (under Velocity Points)

it also says that some terms around landed masked areas are computed according to either free-slip or no-slip boundary conditions - so we should be able to choose either no-slip ou free-slip around masked areas. So I really don't understand if I would be able to choose that or not.

I do undestand what wilkin said, so is the version of ROMS that I'm using outdated concerning free-slip/no-slip around masked areas? I mean, are new versions hardwired to be free-slip and old versions to be no-slip (or for us to be able to choose)?

Also, if in my version it is indeed free-slip, how can I change it to no-slip? This is one of the main things that I would like to investigate in my study, no-slip vs. free-slip around land areas, so that would be important for me. I already just set undef MASKING, but that did not work - the model blew up. Also, what are the main advantages of using MASKING? Are we obligated to use the option MASKING whenever we have land in our domain?

I'm sorry for so many questions, but I'm just confused and this is pretty determinant for my work. I appreciate any help!

[kthyng]

It looks like the answer is that it depends, and this posting has lots of helpful information:

viewtopic.php?f=19&t=471

So the details of being able to set no/free slip at masked areas depends on the viscosity settings in your simulation, which should depend on your choice for horizontal momentum advection. In most cases, I think the simulations will act like no-slip. For example, I ran a simulation with the 3rd order upwind scheme for horizontal momentum advection and no explicit viscosity, and vorticity was generated at masked areas.

Kristen