[Fwd: RE: [MITgcm-support] grid-scale noise in 1/4-deg model]

[Ichiro Fukumori]

Alistair,

The biharmonic friction is set at 1.e-3.  The model blew up when we
tried 2.e-3 or 5.e-3.  Should we make it smaller?

 From the behavior of changing biharmonic friction and to kill the
noise, I thought of trying a large Laplacian friction which I thought
would have a wider ranger of stable frictional values.  We've ran the
model with Laplacian viscosity of 100 and 200 (m^2/s) with cosPower=2
without biharmonic friction.  Case with 200 is clearly much more diffusive than
biharmonic with 1.e-3, but the grid-scale noise is still present.

KPP parameters displayed similar grid-scale noise.  So we turned
horizontal smoothing of KPP parameters on, but the noise is still
present.  KPP noise must be a consequence of the grid-scale noise
and not the cause.

We do have bottom friction on;
  no_slip_sides=.TRUE.,
  no_slip_bottom=.TRUE.,
  bottomDragQuadratic = 0.002,
in the data file.   Could we have a problem in particular advection
scheme we are using?   We've got;
  tempAdvScheme=4
  saltAdvScheme=4
  multiDimAdvection=.FALSE.,
  staggerTimeStep=.TRUE.,
How about CD scheme?  We don't have it on, but we tried turning it on
cold turkey, setting D-grid velocity to zero, but the grid-scale noise
was still present after 30-days of integration.

Any suggestions?

-- Ichiro

PS. I am not on the mitgcm-support mailing list.  Can I be included?
Would I want to be included?

At 03:24 PM 12/2/2003 -0800, Dimitris Menemenlis wrote:
>-----Forwarded Message-----
>From: Alistair Adcroft <adcroft at MIT.EDU>
>To: mitgcm-support at mitgcm.org
>Subject: RE: [MITgcm-support] grid-scale noise in 1/4-deg model
>Date: Tue, 02 Dec 2003 17:10:56 -0500
>
>Ichiro,
>
>For the left column runs (bi-harmonic) were you using the maximum allowed
>viscosity?
>It's worth finding the upper limit on numerical viscosity before the model
>is unstable. My preferred recipe is to use 0.1 of this value - this damps
>the grid scale noise one e-folding scale every 10 steps and damps the 2-dx
>wave only every 160 steps. I think the relevent expression is something
>like:
>    A_4 * dt * cos(phi_max)^m * pi^4 [ 1 / dx(phi_max)^2 + 1/dy(phi_max)^2
>]^2 < 1
>assuming the expression is maximized at the northern edge of the domain. The
>choice of "m" can change that.
>
>Do you have bottom friction turned on?
>
>A.
>--
>Dr Alistair Adcroft            http://www.mit.edu/~adcroft
>MIT Climate Modeling Initiative        tel: (617) 253-5938
>EAPS 54-1523,  77 Massachusetts Ave,  Cambridge,  MA,  USA
>
>-----Original Message-----
>From: mitgcm-support-bounces at mitgcm.org
>[mailto:mitgcm-support-bounces at mitgcm.org] On Behalf Of Ichiro Fukumori
>Sent: Monday, November 24, 2003 9:41 PM
>To: cnh at mit.edu; adcroft at ocean.mit.edu; support at mitgcm.org; Armin Koehl;
>Detlef Stammer
>Cc: tlee at pacific.jpl.nasa.gov; bnc at pacific.jpl.nasa.gov
>Subject: [MITgcm-support] grid-scale noise in 1/4-deg model
>
>
>Dear Colleagues,
>
>We are experiencing grid-scale noise in our global 1/4-deg model simulation.
>It does not appear to be a lack of horizontal friction. Would you have any
>idea what may be the problem?  My suspicion is that this may have something
>to do with some vertical process.  Places with prominent noise are adjacent
>to coastal boundaries (where there's
>upwelling) and places with deep mixed layers.
>
>Attached are 2 plots comparing model output of two runs at two different
>locations.  Each page shows a 101x101 grid point area.  Plot n03d_1.ps is
>surface layer along the west coast of Africa (centered 13.8E, 19.4S) and the
>other (n03d_2.ps) is Kuroshio extension area (centered 170.1E, 28.9N) at
>350m depth.  The three rows are, from the top, zonal velocity (u, m/s),
>meridional velocity (v, m/s), and temperature (T, degC).  The left column is
>a run that employs biharmonic friction (1.e10 m^4/s) and is at day 123 of
>1992.  The right column used Laplacian friction (200 m^2/s) and is at day 69
>of 1992.
>
>In n03d_1.ps, the noise appears along the coast (land is shown in gray),
>especially in u & v of the right column (Laplacian friction run).
>
>In n03d_2.ps, the noise is severe in u & v in the left column in the
>northern half, just south of the Kuroshio.
>
>All results are 3-day averages.
>Could these noise be due to an absence of the CD scheme?  What else could
>cause such noise?  We would appreciate any suggestions you may have.
>
>Thanks, Ichiro
>
>
>===============================================================
>Ichiro Fukumori               | Jet Propulsion Laboratory
>e-mail: fukumori at jpl.nasa.gov | Mail Stop 300-323
>phone: +1 (818) 354-6965      | 4800 Oak Grove Drive
>fax: +1 (818) 393-6720        | Pasadena, California 91109, U.S.A.
>===============================================================
>
>
>_______________________________________________
>MITgcm-support mailing list
>MITgcm-support at mitgcm.org
>http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
>--
>Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
>Jet Propulsion Lab, California Insitute of Technology
>MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099
>tel: 818-354-1656;  fax: 818-393-6720

===============================================================
Ichiro Fukumori               | Jet Propulsion Laboratory
e-mail: fukumori at jpl.nasa.gov | Mail Stop 300-323
phone: +1 (818) 354-6965      | 4800 Oak Grove Drive
fax: +1 (818) 393-6720        | Pasadena, California 91109, U.S.A.
===============================================================