[MITgcm-support] Layers and NLayers

[Ryan Abernathey]

Hi Madeline,

Sorry for the delayed reply--you caught me just before a vacation.

It would be nice to have some more quantitative information with which to
answer your questions. For now, I can give you the rules of thumb that I go
by.

The vertical resolution of your model is fundamentally limited by the
number of vertical levels (Nz). If the stratification in your domain is
fairly uniform, such as in a regional model, Nlayers ~= Nz is a good
choice. However, if your model spans regions with very different
stratification (e.g. a global model), you will want Nlayers > Nz.

As for the layer spacing, you can use either even or uneven spacing. If you
are using potential density and are interested in the deep MOC, you
probably want much closer spacing in the abyssal density ranges. For
example, if using sigma0, most of the ocean is volume between 27 and 27.82,
but near-surface densities can be as low as 20 in the tropics. I have used
various strategies to devise layer spacing, the most complicated being
constructing the density PDF and then spacing the layers such that each
contains approximately equal volume.

If you have an eddy-resolving model, you can get away with just using a
very high number of layers, evenly spaced, since the model's internal
variability will smooth out the distribution. (The downside is that it is
more computationally expensive.) This won't work well with a
coarse-resolution model, however, because if your layer spacing is too
fine, you might end up "skipping" layers.

I am slowly working on a paper about layers, which I hope will provide more
detailed answers to your questions.

Cheers,
Ryan







On Fri, Jul 22, 2016 at 5:26 PM, Madeline Miller <
madelinemiller at fas.harvard.edu> wrote:

> Hi Ryan and other Layers experts,
>
> Is there any rule-of-thumb for choosing Nlayers as a function of Nr? I
> know this depends on the density variations in my experiment, but have you
> run into any limitations on the high or low end?
>
> I'm trying out the Layers package for the first time, which makes my
> Friday very exciting.
>
> Best,
>
> Madeline
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20160808/eddecd3b/attachment.htm>