Ocean Modeling Discussion

Open boundaries are not well posed. I found that I had to nudge at the open boundaries to conserve volume in the domain. I don't think this is a pressure gradient problem, but an open boundaries problem.

The same simulation with closed boundaries should not have this trouble and you can judge the ...

Sonaljit, I had similar trouble in my idealized simulations.

In my case, the problem happened where the vertical grid spacing was largest (and s-surfaces were sloping; i.e. classic conditions for pressure gradient error). Reducing vertical grid spacing increased the amount of time it took the ...

Hi Austin,

My background state is just flat isopycnals, N2 = constant.

You are right, I see the same trouble with no forcing and flat isopycnals. Increasing the number of vertical levels reduces the growth rate, but then increasing N2 increases the growth rate quite dramatically.

I've played with ...

Hi all,

I'm having trouble with what looks like a numerical instability that starts in the vertical velocity field.

My setup consists of an idealized flat bottom shelf + slope at the southern end of a beta-plane channel with a west to east (along-isobath) barotropic inflow (2cm/s). There is an ...

Thanks!

The changes I made agree with the code you posted. I like the "radiation cooling" analogy.

Deepak

RadNud means there's a nudging term added to the radiation condition. The timescale refers to the timescale of this nudging which is (generally) longer if the phase speed (determined for the radiation condition) is directed outward and shorter if it is inward.

The reference is this paper: http ...

I can now confirm that my application behaves a lot better when the two changes are made. Earlier, I had a bunch of waves running around my domain. Now, they exit the open boundaries very nicely.

Deepak

I think I found 2 bugs in the code for the new Shchepetkin boundary conditions. In u2dbc_im.F for the eastern boundary, the current code has

cff=1.0_r8/(0.5_r8*(GRID(ng)%h(Iend ,j)+ &
& GRID(ng)%h(Iend+1,j)))
cff1=SQRT(g*cff)
Cx=dt2d*cff1*0.5_r8*(GRID(ng)%pm(Iend ,j)+ &
& GRID(ng)%pm(Iend+1 ...

Also, the horizontal derivatives for the relative vorticity are calculated as

dUde(i,j)=om_u(i,j )*u(i,j ,k,nout)- &
& om_u(i,j-1)*u(i,j-1,k,nout)
dVdx(i,j)=on_v(i ,j)*v(i ,j,k,nout)- &
& on_v(i-1,j)*v(i-1,j,k,nout)
cff=pm(i,j)*pn(i,j)
rvor(i,j,k)=cff*(dVdx(i,j)-dUde(i,j))


Isn't this only ...

I just ran into this problem. The issue is that potential density, pden=0 unless you have DIAGNOSTICS or LMD mixing defined.

vorticity.F uses pden to calculate grad(rho) for PV. However, pden is defined in Nonlinear/rho_eos.F as

# if defined LMD_SKPP || defined LMD_BKPP || defined DIAGNOSTICS ...

Hi,

I think it would be very useful if we could write viscosities (visc2_p,visc2_r or visc4_p,visc4_r) and diffusivities (diff2, diff4) to the his / avg / dia files. This would greatly help debugging sponge layers coded in ana_hmixcoef.h.

Similar functionality for ana_nudgcoef.h might be useful to ...

To see how the Courant numbers are calculated, see Utility/metrics.F . To reduce it, reduce your time step. Right now you're using dt=432 seconds = 7.2 minutes which seems to be too big. The other way would be to increase ndtfast but your value seems pretty large already. Try dt=60, ndtfast = 30 and ...

I just wrote an answer for a similar question here: https://www.myroms.org/forum/viewtopic.php?t=2840#p10727

I've been implementing a similar setup viz., flow entering at 'top' (northern edge) and leaving at 'bottom' (southern edge). I'm using the RadNud option from the Marchesiello et al (2001 ...

Here, this should help: http://www.oceanwave.jp/research/ROMS/roms_obc.pdf . Looking at your analytical.F90 (Build/ folder) or the various *bc_im.F files in ROMS/Nonlinear/ will show you the code for the various options.

For more detail on the RadNud option, see Marchesiello et al. (2001): http ...