Weights in VSHAPE and layers thickness?

[Pysh]

Dear all,

If I need distribute river input among the layers, are the weights of VSHAPE (for example 0.5,0.3,0.2) take into account layers thickness or these are the specific intensities of input, t.e real disstribution among the layers are (0.5 * h1, 0.3 * h2, 0.2 * h3) / (h1 + h2 + h3), where h1, h2, h3 - layers thickness?

Thanks in advance,
Boris

[wilkin]

The weights are the fraction of the total discharge that goes in to each layer, without regard for how thick in meters that layer happens to be. Hence the weights must sum to one. They are dimensionless.

Therefore, if you alter the coordinate stretching without changing Vshape you would effectively alter the vertical distribution of the flow.

In very shallow water - where one usually places river sources (at the coast) - most stretching parameter choices will give you close to a uniform distribution of layer thickness. We don't see much sensitivity to the choice of Vshape in shallow estuary applications.

[Pysh]

Hi, John

It's clear now. I'am working at small program to prepare river input. And thank you for your matlab tools. It is very usefull.

Thank you very much,
Boris

[pmaccc]

I am using what I think are fairly common vertical stretching parameters:

THETA_S,4
THETA_B,0.8
TCLINE,0
N,30
VTRANSFORM,1
VSTRETCHING,1

and when I have a river source with 10 m depth the upper layer is rather thin, like 5 cm. So when I have very high river flow, and a Vshape that varied linearly from zero to 2/N over the layers (as recommended elsewhere) I was inadvertently putting a lot of transport through a very thin layer. This led to the model blowing up. By changing to Vshape = diff(Cs_w) you get constant velocity in each layer, which solved my blowup problem. My problem was that I thought Vshape would account for the layer thickness. I bet this incorrect interpretation has been pushing my DT to be unnecessarily low for many years.

[kate]

It has been many a year since I've heard a recommendation of VSTRETCHING=VTRANSFORM=1.

Parker, if you want to talk in person, I'm just down the street at U.W.M.C.

[pmaccc]

Kate - thanks for the tip. I realize that our choices about the vertical coordinate transform and stretching have not been updated for some time. I tried experimenting with VTRANSFORM = 2 and VSTRETCHING = 4, and can see that the results might be smoother than what I am using, but they won't be fundamentally different.

Two questions:

First, is it true, as Wilkin suggests, that in shallow water the s-distribution becomes more evenly distributed? If so, where in the code does this happen, and at what depth?

Second, what are recommended choices for parameters when using VTRANSFORM = 2 and VSTRETCHING = 4? Particularly for THETA_S and THETA_B.

Thanks, Parker

[wilkin]

Parker,

In WikiROMS at https://www.myroms.org/wiki/Vertical_S-coordinate there is a note to the effect that the coordinate tends to a uniform σ coordinate in shallow water:

Transformation (2) has been available in UCLA-ROMS since 2005. It is activated by setting Vtransform = 2 in ocean.in. Notice that,

S ( x , y , σ ) = { 0 , if σ = 0 , C ( σ ) = 0 , at the free-surface; − 1 , if σ = − 1 , C ( σ ) = − 1 , at the ocean bottom.
which is different to the behavior of the original functional in (1). Shchepetkin (personal communication) points out that (2) offers several advantages:

Regardless of the design of C ( σ ) , it behaves like equally-spaced sigma-coordinates in shallow regions, where h ( x , y ) ≪ h c . This is advantageous because it avoids excessive resolution and associated CFL limitation is such areas.

There are also recommendations about theta_s and theta_b values on that WikiROMS page.