[MITgcm-support] is Orlanski radiation condition working in the rotating rank of MITgcm?

[Yi-Chih Huang]

Martin,

    Thanks much for the reply.  The .png picture is a vertical cross
section.  The initial temperature field is cold blubs at the top.  Suddenly
high-temperature disturbances occur around the lateral boundary in every
simulation with different grids of .bin data with which I have worked.
Thus, I believe something wrong with the lateral b.c.  If the setup for
Orlanski b.c. is right, why would the temperature patterns be the same in
two different b.c.?

    I turned off centrifugal force by setting amplCF=0.0 in data.  The cold
blubs are only testing simulations.

    I am new to MITgcm and would like to know about the relationship
between radius and grid information.  Do you think the given rSphere
or radius_fromHorizGrid
in the data file would change the grid information in CS, LLC, or the
curvilinear coordinates when OLD_GRID_IO is set in CPP_OPTIONS.h?
Certainly, I will try the Cartesian coordinates you suggested.

    Many thanks,

                Yi-Chih

Date: Wed, 19 Dec 2018 13:57:24 +0100
From: Martin Losch <Martin.Losch at awi.de>
To: MITgcm Support <mitgcm-support at mitgcm.org>
Subject: Re: [MITgcm-support] is Orlanski radiation condition working
        in the rotating rank of MITgcm?
Message-ID: <2ECB81F3-A681-4324-9108-D4065A7E778F at awi.de>
Content-Type: text/plain; charset="utf-8"

Yi-Chih,

I still don?t really know what the problem is. in your png file, which I
assume is a vertical section along a boundary, there are some smooth
variations of the otherwise homogeneous temperature (scaled density) field.
What do you expect to see? Constant temperature?

I would try without the ?centrifugal force? in apply_forcing.F, which I
assume is your addition (BTW, the masking in the ?centrifugal force? is not
correct. I should be maskW for gU_arr and maskS vor gV_arr, you have maskW
for both), and see if you get what you expect.

More general comments:

The obcs-settings appear to be OK although I still don?t see why you want
to use open boundary conditions for your simluation at all.

You seem to use z-coordinates (default). In your data you have very high
viscosities (1e6), but very low diffusivities (5e-6, 1e-7). Is that on
purpose? I would expect that this would put a strong contraint on your
timestep deltaT. Also you use rSphere = 60000.E3, rotationPeriod=38520.0,
so a planet rotating about twice as fast as the earth with a radius 10
times larger. Is the larger radius reflected in the grid input fields?

For your experiment, I would just use a cartesian grid centered at the
North Pole of the sphere (constant f0), or if the variation of coriolis is
important, you can read the (precomputed) coriolis parameter via files
fCoriC.bin, fCorCs.bin (for the 2*om*cos(lat)),fCoriG.bin, for that you
need to set selectCoriMap = 3 in data. It would make things much easier for
you to control.

Martin

>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20181219/f5ee2684/attachment.html>