<div dir="ltr"><div dir="ltr">Hi Martin!<div><br></div><div>Thanks for your reply!</div><div><br></div><div>Right to the business:<br></div><div><br></div><div>>>I would use viscAhGrid or viscAhGridMin for this (something small)<br></div><div><font face="arial, sans-serif">To be honest, I hardly understand the nature of viscAhGrid parameter... While viscAh can be seen as some kind of constant physical background viscosity not resolved by the main horizontal turbulence parameterization used (Smagorinsky, Leith, etc),
viscAhGrid seems like some numerical trick. From its formula in MOM_CALC_VISC it can be concluded that it<span style="background-color:rgb(243,246,246);color:rgb(64,64,64)"> has </span>something <span style="background-color:rgb(243,246,246);color:rgb(64,64,64)">to do with grid size (just like Smagorinsky)</span>.</font></div><div><font face="arial, sans-serif"><br></font></div><font face="arial, sans-serif">>>this is only 2D Smagorinsky. I would use 2D Leith instead.</font></div><div><font face="arial, sans-serif">The only reason I chose Smagorinsky parameterization was its dependence on grid resolution and dynamical regime (velocity gradients are involved). But your alternative (viscAhGrid +
viscC2leith), perhaps, does the same in a combined manner. </font></div><div><br></div><div dir="ltr"><font face="arial, sans-serif">>> setting viscAhGridMax (something a little smaller than 1)</font></div><div dir="ltr"><font face="arial, sans-serif">>> may be necessary</font></div><div><font face="arial, sans-serif">In MOM_CALC_VISC the recommended value for viscAhGridMax is 1. What value would you use?</font></div><div dir="ltr"><br></div><div dir="ltr"><font face="arial, sans-serif">>> depending on your advection scheme, horizontal diffusivities</font></div><div dir="ltr"><font face="arial, sans-serif">>> can be zero (I would use 77, 33 or 7 for advection)</font></div><div>I use the advection scheme 33 (3-order DST with Sweby flux limiter) for T, S and seaice.<br></div><div><br></div><div><div><font face="arial, sans-serif">Does the next set of parameters seem better than the previous one?</font></div><div><font face="arial, sans-serif">viscAhGrid = 0.5, viscAhgridmax = 1.0, <span style="background-color:rgb(252,252,252);color:rgb(0,0,0)">useFullLeith = .TRUE., </span><span style="color:rgb(0,0,0);background-color:rgb(252,252,252)">viscC2leith = 1.0, </span><span style="background-color:rgb(252,252,252)">viscC2LeithD = 1.0, </span><span style="color:rgb(0,0,0)">viscAr = 1.0e-5.</span></font></div></div><div><font face="arial, sans-serif"><span style="background-color:rgb(252,252,252)">Diffusivities: </span></font><span style="color:rgb(0,0,0);font-family:arial,sans-serif">diffKhT = 1.0, </span><span style="color:rgb(0,0,0);font-family:arial,sans-serif">diffKhS= 1.0, </span><span style="color:rgb(0,0,0);font-family:arial,sans-serif">diffKrT = 1.0e-5, </span><span style="color:rgb(0,0,0);font-family:arial,sans-serif">diffKrS = 1.0e-5<span style="background-color:rgb(252,252,252)">.</span></span> <font face="arial, sans-serif"><span style="color:rgb(0,0,0)"><br></span></font></div><div><div><span lang="EN-US" style="font-family:arial,sans-serif;line-height:115%;background:rgb(252,252,252)"><font color="#000000">Here I additionally set </font></span><span style="font-family:arial,sans-serif">viscC2LeithD as it is </span>recommented<span style="font-family:arial,sans-serif"> in the manual (</span><span style="font-family:arial,sans-serif;background-color:rgb(252,252,252)">a
damping specifically targeting purely divergent instabilities near the
gridscale).</span><br></div></div><div><span style="font-family:arial,sans-serif;background-color:rgb(252,252,252)"><br></span></div><div>>> Don’t use the CD scheme, if necessary use small biharmonic viscosity</div><div>>> instead. Implicit vertical advection is probably not required either<span style="font-family:arial,sans-serif;background-color:rgb(252,252,252)"><br></span></div><div>OK, the CD_scheme is omitted then, thank you! But can Implicit vertical advection significally alter the results? I just hoped to use it in order to somewhat increase the time step.</div><div><br></div><div>>> Why do you want to use the MY82 package?</div><div>The only reason is that I previously used it and familiar with it. Moreover, the MY is a well-known mixing scheme, and both its good and bad sides are also documented.</div><div><br></div><div><font face="arial, sans-serif">>> I’d prefer GGL90 or KPP (although at your resolution, KPP might be noisy (see this thread <<a href="http://mailman.mitgcm.org/pipermail/mitgcm-support/2020-April/012427.html" rel="noreferrer" target="_blank">http://mailman.mitgcm.org/pipermail/mitgcm-support/2020-April/012427.html</a>>).</font></div><div><font face="arial, sans-serif">Previously our team used the KPP for another configuration, but for now I decided to omit it. I read the thread you mentioned, and it was the last straw to find another scheme :-)</font></div><div><font face="arial, sans-serif"><br></font></div><div><font face="arial, sans-serif">As for GGL90, I did not previously used it at all. From the original paper by Gaspar et al, 1990, I know that it has only one prognostic equation (for TKE) and that the turbulent length scale is diagnosed rather than prognosed. From your experience, can you recommend it over the MY82?</font></div><div><font face="arial, sans-serif"><br> >> but it should (like GGL90) remove the instabilities,</font></div><div><font face="arial, sans-serif">>> because it’s also Richardson number based<br></font></div><div><font face="arial, sans-serif">Yes, I know, and that is why I was surprized by the comment in the code "convective adjustment might be needed even with ggl90". Do you use GGL90 with or without convAdj or ivdc_kappa?</font></div><div><font face="arial, sans-serif"><br></font></div><div><font face="arial, sans-serif">I also know that you are the author of the SEAICE package for MITgcm, so could you tell, is it OK for now to use the newly added feature <span style="background-color:rgb(252,252,252)">SEAICE_ALLOW_MOM_ADVECTION and corresponding run-time parameters </span><span style="background-color:rgb(252,252,252)">SEAICEmomAdvection=TRUE,</span><span style="background-color:rgb(252,252,252)"> </span><span style="background-color:rgb(252,252,252)">SEAICEuseMetricTerms
= TRUE, and </span><span style="background-color:rgb(252,252,252)">SEAICEuseFluxForm=</span><span style="background-color:rgb(252,252,252)">TRUE if my configuration uses curvilinear grid (and that's why I must use vectorInvariant for momentum)? The </span></font>vectorInvariant does not use the metric terms for momentum, should I use them for SEAICE on my grid? And as I understood, there is no alternative for
SEAICEuseFluxForm?</div><div><br></div><div>Stanislav</div><div><font face="arial, sans-serif"><br></font></div><div><font face="arial, sans-serif"><br></font></div><div><font face="arial, sans-serif"><br></font></div><div><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">вт, 21 апр. 2020 г. в 14:31, Martin Losch <<a href="mailto:Martin.Losch@awi.de">Martin.Losch@awi.de</a>>:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Hi Stanislav,<br>
<br>
generally your parameters look good, so comments below<br>
<br>
> On 21. Apr 2020, at 11:23, Stanislav Martyanov <<a href="mailto:martyanov.sd@gmail.com" target="_blank">martyanov.sd@gmail.com</a>> wrote:<br>
> <br>
> Hello, dear colleagues!<br>
> <br>
> I am configuring the MITgcm for the Kara Sea region. With the curvilinear grid, the model's resolution is: dx=500-1200 m, dz=2 m in upper layers, and up to 20 m in the deepest. The minimum model depth is set equal to 5 m. I use the pure r vertical coordinate (not r*) for now. Because of the curvilinear grid, I use vectorinvariant package. Other packages used are: seaice, my82, exf, cal, obcs.<br>
> <br>
> Having carefully read (once again) the newest version of the online-manual and browsed the code, I have configured almost all necessary model parameters, but still some questions remain. They are qiute specialized, so I could not find the needed information in puplished papers. But for those experienced with MITgcm, they might be easy to answer. I will appreciate any advise!<br>
> <br>
> 1) The newly added feature with partial cells (#undef EXCLUDE_PCELL_MIX_CODE in CPP_OPTIONS). The parameter pCellMix_select = 0 by default, which means that the enhanced mixing is OFF in bottom and surface layers. The parameters intended for the enhanced mixing in the inner layers are also OFF by default: interViscAr_pCell and interDiffKr_pCell = FALSE. Are this option and the corresponding run-time parameters really important for such configuration (Kara Sea)? Any experience with them?<br>
I have no experience with this, but I think you can safely keep this turned off.<br>
> <br>
> 2) I use the following viscocity and diffusivity values, intending to use the Smagorinsky viscocity for lateral turbulent exchange:<br>
> viscAh = 1.0,<br>
I would use viscAhGrid or viscAhGridMin for this (something small)<br>
this is only 2D Smagorinsky. I would use 2D Leith instead.<br>
> viscC2smag = 3.0,<br>
setting viscAhGridMax (something a little smaller than 1) may be necessary<br>
> viscAr = 1.0e-5,<br>
depending on your advection scheme, horizontal diffusivities can be zero (I would use 77, 33 or 7 for advection)<br>
> diffKhT = 1.0,<br>
> diffKhS= 1.0,<br>
> diffKrT = 1.0e-5,<br>
> diffKrS = 1.0e-5,<br>
> But for diffusivity there is no Smagorinsky analogue. What approach can be the most suitable in this case?<br>
> <br>
> 3) In the paper "ECCO version 4 - an integrated framework... " by Forget et al., 2015, the authors decided to switch OFF the C-D sheme and to switch ON the Implicit vertical advection option. But they investigated the global scales, not high-res. From your experience, can their advise be applyed to the high-res ocean simulations?<br>
Don’t use the CD scheme, if necessary use small biharmonic viscosity instead. Implicit vertical advection is probably not required either<br>
> <br>
> 4) The MY82 package does not allow using the convective adjustment (cAdjFreq) and ivdc_kappa. Does it mean that the MY in MITgcm works fine with vertical instability cases? Actually, it should, as follows from J. Mellor's papers, but this question arose when I was investigating the GGL90 code, where I found the line "convective adjustment might be needed even with ggl90"…<br>
Why do you want to use the MY82 package? I’d prefer GGL90 or KPP (although at your resolution, KPP might be noisy (see this thread <<a href="http://mailman.mitgcm.org/pipermail/mitgcm-support/2020-April/012427.html" rel="noreferrer" target="_blank">http://mailman.mitgcm.org/pipermail/mitgcm-support/2020-April/012427.html</a>>).<br>
You can just comment out the stop-statement in the code and use MY82 with other convective adjustment, but it should (like GGL90) remove the instabilities, because it’s also Richardson number based. MY82 is not really used that much (at all?).<br>
<br>
Martin<br>
> <br>
> Once again, I will appreciate any advise!<br>
> <br>
> PS: Sorry for large wall-of-text here, but I tried to be concrete. :-)<br>
> _______________________________________________<br>
> MITgcm-support mailing list<br>
> <a href="mailto:MITgcm-support@mitgcm.org" target="_blank">MITgcm-support@mitgcm.org</a><br>
> <a href="http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support" rel="noreferrer" target="_blank">http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support</a><br>
<br>
_______________________________________________<br>
MITgcm-support mailing list<br>
<a href="mailto:MITgcm-support@mitgcm.org" target="_blank">MITgcm-support@mitgcm.org</a><br>
<a href="http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support" rel="noreferrer" target="_blank">http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support</a><br>
</blockquote></div>