[MITgcm-support] Using OBCS in a 1D model setup

[Nicholas Rogan]

Ah ok thanks for your help Martin. 
I’ll go through this this afternoon. 

Sounds good though

Nick

On 12 Mar 2015, at 12:44, Martin Losch <Martin.Losch at awi.de> wrote:

> Nick,
> 
> the model should conserve tracers perfectly in your 1-column setup, because, whatever leaves the domain on one side, enters again through the opposite side. In fact with periodic boundaries this should work perfectly and with OBCS these conservation issues are more subtle (see the documentation for more details).
> 
> I would first check, if your tracer is conserved without your reaction-terms on the right hand side. Then there is a potential issue with the surface boundary conditions. Do you use the exf-package (useEXF=.TRUE. in data.pkg)? If you do you don’t specify any rain, but the package will probably compute evaporation, so that you are loosing freshwater in your domain (does the sainitiy change?).
> What does your data.ptracers look like? I have had surprises with the parameter PTRACERS_EvPrRn(:). If it unset, any fresh water that you add or remove (precip or evap), will have the concentration of your tracer in the surface layer. If you only evaporate, then you will loose tracer through the surface.
> 
> Martin
> 
> 
> On 12 Mar 2015, at 11:44, Nicholas Rogan <nrogan at geomar.de> wrote:
>> 
>> Hi Martin, 
>> 
>> Thanks for your response. My previous experiments running the model with Nx=Ny=1 have resulted in sensible seasonally cycling nutrient profile development. The only issue is that there is a “creep” in the total nutrients (phosphate and iron) resulting in a gradual reduction of the concentrations. I gauge this by the concentration once the winter mixing has occurred (the column is entirely mixed). This is expected for iron, but phosphate should be conserved.
>> 
>> So, results-wise, my model setup with the Nx=Ny=1 is performing well. However, we had a discussion in my group about the model setup and the issue of the model not being “closed” was raised. There was some concern about heat fluxes and the general robustness of using the model as I had set it up. I’ll admit that I didn’t fully understand the concern, but that’s probably because I am slightly naive when it comes to using idealistic models. Is there perhaps a requirement for using OBCS or RBCS in order to make the model more robust?
>> 
>> Sorry for evolving the thread into a more general model “Dos and Don’ts”.
>> 
>> 
>> On 12 Mar 2015, at 10:21, Martin Losch <Martin.Losch at awi.de> wrote:
>> 
>>> Nick,
>>> 
>>> the default of the MITgcm is to have (double) period boundaries, ie. with your configuration (Nx=Ny=1), you have an infinite ocean that is homogenous in the horizontal. Closed boundaries are not possible with this configuration (you’d have to put a land point at your only horizontal grid point).
>>> 
>>> OBCS: The open boundaries are always at grid point that belong to the computational domain, i.e. they have to be a i = 1 (or larger) and at i = Nx (or smaller), and similarly for j. In your case Nx=Ny=1, so your single horizontal grid point belongs to 4 boundaries. You can imagine that this is not sensible. A 3x3 domain should work, in the sense that the central grid point will not be directly prescribed (although the normal velocities will), but I am not sure what you want to achieve with this configuration that you cannot with a single column (without OBCS). Even for the 3x3 domain the centrail point will be totally determined by the "open boundaries”, because there will be little dynamics that can develop.
>>> 
>>> I think it would be simpler and more straightforward to use RBCS to restore your fields to your data, but then you also have only very little dynamic development.
>>> 
>>> Martin
>>> 
>>> 
>>>> On 11 Mar 2015, at 17:53, Nicholas Rogan <nrogan at geomar.de> wrote:
>>>> 
>>>> Hi all,
>>>> 
>>>> I have recently been using a 1D model setup to explore the effects of different aerosol depositions on a water column ~ 500m deep.
>>>> 
>>>> Previously the model was set up with closed boundaries and was not a closed system (removal out of the base of the water column is gradually draining the conserved tracers).
>>>> 
>>>> For a more robust model I wanted to add open boundaries on all of the faces of my single column, for which I have diagnosed time varying daily T, S, U, V and six ptracers from a Global 3D model field. 
>>>> 
>>>> I am now trying to implement these velocity fields and tracer concentrations in order to have the diagnosed fluxes at the boundaries of my model.
>>>> 
>>>> The model output from my run with closed boundaries and run with open boundaries is drastically different and I’m not convinced it is solely down to the change in dynamics. 
>>>> I’d greatly appreciate if anyone could have a quick look at my input data files and identify an obvious error, or tell me if using OBCS with a model with a horizontal grid size of 1 x 1 is possible? 
>>>> I’ve thought that I might have to change to a 3 x 3 grid and have the central grid point as the single water column? 
>>>> Or maybe the daily OBCS fields period/cycles do not match with the half hour time step?
>>>> 
>>>> I’ve provided my SIZE.h, data, data.exf and data.obcs files.
>>>> 
>>>> Look forward to hearing from someone!
>>>> 
>>>> Nick
>>> 
>>> 
>>> _______________________________________________
>>> MITgcm-support mailing list
>>> MITgcm-support at mitgcm.org
>>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>> 
>> 
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-support
> 
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support