[MITgcm-support] Orlanski radiation condition parameters

[Stanislav Martyanov]

Hi Paul,

Yes, the Orlanski open BCs are suitable for variable phase velocity, but it
works perfectly only in theory. In reality, a number of numerical studies
demonstrated some problems with the Orlanski scheme. For more details, you
may see, e.g., 'Revisiting open boundary conditions from the point of view
of characteristic variables' by E. Blayo *, L. Debreu, 2005, etc.

>From my own experience with MITgcm applied for a real object (with
realistic forcings, etc), the Orlanski scheme does not work very well, and
the Dirichlet+sponge condition is the best option available in the model,
even better than the Stevens OBCs implemented in MITgcm. It' s just my
experience, maybe in other cases it will be the opposite.

---
Kind regards,
Stanislav Martyanov

пн, 18 апр. 2022 г. в 09:28, Burns, Paul <P.Burns2 at exeter.ac.uk>:

> Hi Jody,
>
> Thanks for your message.
>
> My understanding is that the Orlanski scheme is suitable for cases where a
> spectrum of waves is excited.  In Orlanski (1976) it shows that the scheme
> uses a variable phase velocity (varying in time, location along boundary,
> and with each prognostic variable) to try and account for different waves
> reaching the boundary (shown in equation (2.2)).  Orlanski also performed
> tests, computing the evolution of a collapsing bubble in stratified flow,
> which excited many different internal waves, with the results showing only
> small distortion using the boundary condition (2.2).  In the same paper it
> is also said that the sponge condition is not perfect and may create small
> distortions.  It seems that neither boundary condition is perfect.
>
> My understanding is that the parameter Cmax that the users sets in MITgcm
> sets only the maximum phase speed that the Orlanski scheme accounts for,
> but that the scheme still computes variable phase speeds below this limit.
>
> I've performed some initial tests using the Orlanski scheme and I'm not
> seeing obvious reflections at the boundary, with flow structures apparently
> leaving the domain without big distortions.  But I wish to confirm things
> further than visual inspection, which isn't sufficient alone.
>
>
> Cheers,
> Paul
>
>
>
>
>
> Dr Paul Burns MInstP FRMetS
>
> Research Fellow
>
> Geophysical and Astrophysical Fluid Dynamics Group
>
> Department of Mathematics
>
> Harrison Building
>
> University of Exeter
>
> ------------------------------
> *From:* MITgcm-support <mitgcm-support-bounces at mitgcm.org> on behalf of
> Jody Klymak <jklymak at uvic.ca>
> *Sent:* 17 April 2022 17:43
> *To:* mitgcm-support at mitgcm.org <mitgcm-support at mitgcm.org>
> *Subject:* Re: [MITgcm-support] Orlanski radiation condition parameters
>
> CAUTION: This email originated from outside of the organisation. Do not
> click links or open attachments unless you recognise the sender and know
> the content is safe.
>
>
>
> Hi Paul,
>
> Others should correct me if I am mistaken, but I don’t think Orlanski will
> work particularly well for this case because you do not have one dominant
> wave speed in your problem.  Rather, you might try a gentle sponge.
>
> Cheers,   Jody
>
>
> On Apr 12, 2022, at 11:58 AM, Burns, Paul <P.Burns2 at exeter.ac.uk> wrote:
>
> Hi,
>
> This is a bit basic, but I would like to check my implementation of my
> lateral boundary conditions.  I am running a 2D simulation and prescribe
> (Dirichlet conditions) a Garrett-Munk spectrum of waves on my left
> boundary, which then propagate into my domain.  I would like to allow the
> motions in the interior solution to radiate away through my right boundary,
> so I've asked the model to use the Orlanski condition on the right
> boundary.
>
> In Orlanski (1976) it states that the (variable) phase velocity C must be
> less than the maximum numerical velocity, which for my setup is dx/dt =
> 25/1 = 25 m/s.  So in data.obcs I've set Cmax = 20 m/s.  This is quite
> different to the value used in the 'Gravity Plume On a Continental Slope'
> tutorial, which has a similar domain size to mine, but this is because of
> my much smaller timestep due to my fine vertical grid required for my
> features of interest.  I realise I am using large aspect ratio cells, but
> this is to keep the simulation feasible with the compute resources I have.
> I will try and test sensitivity of my results to horizontal resolution.
>
> Can someone confirm how you fix the remaining parameter of the Orlanski
> scheme? That is,
>  cVelTimeScale
>
>
>
> Many thanks,
> Paul
>
>
>
>
>
> Dr Paul Burns MInstP FRMetS
> Research Fellow
> Geophysical and Astrophysical Fluid Dynamics Group
> Department of Mathematics
> Harrison Building
> University of Exeter
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support
> <https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fmailman.mitgcm.org%2Fmailman%2Flistinfo%2Fmitgcm-support&data=04%7C01%7CP.Burns2%40exeter.ac.uk%7C361fb4c681ad47b6905908da20918ac9%7C912a5d77fb984eeeaf321334d8f04a53%7C0%7C0%7C637858106991748874%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=s8yjYgBQE1fdMLE562bef7JM2U4FDu6JXJizyQilCPk%3D&reserved=0>
>
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20220418/14ec6c08/attachment.html>