Analytical OBC Files?

[aryansafaie]

Dear All,

Perhaps I'm overthinking this, but I'm having trouble understanding what the purpose of the ana_Xobc.h files are (i.e. ana_fsobc.h, ana_m2obc.h, ana_m3obc.h, and ana_tobc.h), and whether I'm using them correctly.
In short, I'm running an idealized model forced only by analytical surface heat fluxes, and my northern boundary conditions are:
Free-surface: Chapman; 2D momentum: Fla; 3D momentum: RadNud; mixing TKE: Gra; tracers: RadNud.
This required me to define ANA_FSOBC, ANA_M2OBC, ANA_M3OBC, and ANA_TOBC. However, because this boundary is purposefully a large distance away from my (nearshore) area of interest, I have no intuition on what to set the values w/in these ANA_ files to, so I just set them to trivial values:
(in ana_fsobc) ...BOUNDARY(ng)%zeta_north(i) = 0.0_r8,
(in ana_m2obc) ...BOUNDARY(ng)%ubar_north(i) & vbar_north(i) = 0.0_r8,
(in ana_m3obc) BOUNDARY(ng)%u_north(i,k) = 0.0_r8 & v_north(i,k) = 0.0_r8,
(in ana_tobc) BOUNDARY(ng)%t_north(i,k,itemp) = T0(ng) & t_north(i,k,isalt) = S0(ng)

My questions relate to the numerical implications of these values. I was looking at the relevant lines of zetabc.F and similar files for u,v, and t, in /Nonlinear/, but it's not clear to me how, during time stepping, these possibly erroneous values that I'm assigning will cause the model to become unstable. On a related note, if I'm artificially constraining the free surface at the offshore end to equal 0, will that create some sort of artificial waves at other parts of the domain?

Any guidance or references to reading material will be greatly appreciated.

Thanks,
-aryan

[jcwarner]

Sometimes you just need to run the model to see what the impact of the boundary conditions are. if you created a domain with a boundary that is "purposefully a large distance away from my (nearshore) area of interest" that sounds like a good thing. The values you impose are not necessarily erroneous or trivial. Compare your BC values to your initial conditions. if you set zeta initial to 0.0 (or something), then a BC value of 0 would be a good choice.
It seems a bigger question is: why do you have any open boundaries to begin with if you are only interested in what seems to be a vertical heat flux exchange? if you are thinking to add tides later then this could be a good place to start.
-john

[aryansafaie]

Thanks for the guidance John.
Yes, the BC values are being set to their initial values.
We will be adding other forcing mechanisms, such as alongshore
currents with tidal variability, and indeed we would like
perturbations generated inside the model domain to leave offshore
without influencing the inner region numerical solution.

For the time being, the values I've set for these ana_Xobc files
seem to be giving a stable solution, but I'll be sure to post back here
if I try anything else that gives interesting results.
Thanks again!
-aryan