[MITgcm-support] Rivers imposed as OBCs - problem with fully-open boundary domains (i.e., islands)

[Jean-Michel Campin]

Hi Stefano,

I looked at the 2 pannels on the figure you sent. And you are right, the current code does not
"work" for the left pannel, only for the right pannel.

And using the "insideOBmaskFile" could help in some cases: 
If you decompose your domain in tiles such as j=21 and j=42 are not on the same tile,
the mask would allow to define the correct location for the Northern OB, but you still
cannot specify OB values from files (this is more an I/O problem) but Orlanski, for instance,
should work.
It's possible that there might be some ways (but not that I am aware of) to get around this I/O issue.

And regarding this:
> P.S.: is it possible that there is an oversight in the documentation (Figure 8.6)? 
> The correct OB configuration in data.obcs should be like that, do you agree?
You are right, this needs to be fixed.

Cheers,
Jean-Michel

On Thu, Mar 20, 2025 at 09:01:19PM +0100, Stefano Querin wrote:
> Dear MITgcm-ers,
> 
> We are running high-resolution regional simulations in marginal seas of the Mediterranean (and sub-basins).
> At resolutions ranging from 1/128° to 1/768° (some examples, respectively, here <https://medeaf.ogs.it/adriatic> and here <https://medeaf.ogs.it/got>), we insert the river inputs as lateral open boundary conditions, since the momentum contribution is significant at those scales (and we can also easily control the thermohaline and biogeochemical inputs by prescribing OB files).
> The problem arises when we deal with domains fully surrounded by (water) open boundaries (e.g., regional focus on Sicily or Sardinia).
> The attached picture shows an idealized case study (island in the open sea), where we want to impose a river originating from the northernmost end of the channel and flowing southward. If we impose a single land point at the northern edge of the model domain (i_g=21 and j_g=41, plot on the right), we can set an OB point at i_g=21 and j_g=21 and the simulation runs smoothly. Conversely, the configuration on the left cannot work (as far as we know...).
> We also tried to use the "insideOBmaskFile" as specified in the documentation for the more complex example <https://mitgcm.readthedocs.io/en/latest/phys_pkgs/obcs.html#a-more-complex-example>, but we are a bit stuck...
> 
> Can you confirm that there is no way to prescribe "inner" OBCs (the river input in i_g=21 and j_g=21) in a domain like the one in the picture (left), while keeping the nesting model values all around the (nested) domain (also in i_g=21 and j_g=41)?
> In that case, we will insert the same runoff as surface flux by using the exf package. Most likely, we will add the freshwater flux in i_g=21 and j_g=21, to let the water gain some momentum along the channel, otherwise, we will remove the channel and simply put the runoff in i_g=21 and j_g=12.
> 
> Thank you!
> 
> Cheers,
> 
> Stefano and Fabio
> 
> P.S.: is it possible that there is an oversight in the documentation (Figure 8.6)? The correct OB configuration in data.obcs should be like that, do you agree?
> 
> OB_Iwest  = 1*0,1*5,1*2,141*0,
> OB_Jsouth = 2*3,3*2,115*0,
> 
> and not
> 
> OB_Iwest  = 1*0,1*5,142*0,
> OB_Jsouth = 2*3,3*2,115*0,
> 
> (or the OB routines automatically set i=2, j=3 as the OB position -red line- ?)
> 
> --
> Stefano Querin, PhD
> Sezione di Oceanografia
> Istituto Nazionale di Oceanografia e di Geofisica Sperimentale - OGS
> via Beirut n. 2
> 34151 Trieste - Italia
> Tel. +39 040 2140623
> Skype: stefano_querin
> www.ogs.it <http://www.ogs.it/>
> 

> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support