Regional / Coastal Model with Oceananigans or ClimaOcean?

[darosaj]

Hi,

I just wanted to say thank you to all of your helpful responses on all of the discussion posts! I'm enjoying learning the package, and its a great community to learn within.

I'm not sure if this question makes sense, but I'm hoping to get a starting point on where to look.

I'm wondering if I were looking to setup a coastal / regional model, would the best starting point be trying use the ClimaOcean package with its bathymetry regridding capability, ECCO and JRA capability? or would a more regional/coastal model be best to setup purely on Oceananigans without the added ClimaOcean functionality at this stage in development?

Again, apologies if this question doesn't make sense, or is ill-informed, but looking for what the best starting point in running a regional model using your amazing package!

Thanks again,

• Josh

[glwagner]

If you want to run an ocean model with a prescribed atmospheric state, you have to use ClimaOcean. ClimaOcean will compute fluxes between your ocean model and a prescribed atmosphere derived, for example, from JRA55.

If you would like to prescribe surface fluxes directly, rather than prescribing the atmospheric state, then it is possible to use Oceananigans. This is often a design used for LES.

The bathymetry regridding, ECCO, and JRA55 are convenience features that will (perhaps greatly) accelerate the process of setting up a regional model. Note, in other modeling systems, no such features are provided. So wrt to other modeling systems capability, the choice is purely between prescribing fluxes vs prescribing an atmosphere and computing fluxes. That said, it's obviously helpful if you want to use realistic bathymetry, an ocean state estimate, and JRA55, to use ClimaOcean.

Note right now ClimaOcean's JRA55 automatic utilities only support 1993-1994. To use other years you have to download the data yourself (like any other model).

Regional modeling is extremely difficult and complex so be sure you want to go down this route before starting. It will really help to have some experience in idealized modeling first before you do realistic modeling, so that you understand what you're getting out of the model and know what to do when things go wrong.

[darosaj]

Hi,

Thank you for the information and quick reply. I'll most definitely continue to run more idealized models and learn the package more before exploring the regional model endeavor. That is just my ultimate goal, and as I continue to learn more, the question popped into my head.

I guess part of my question is surrounding OpenBC's and with a regional model whether specifying these correctly would be possible and easier with ClimaOcean or Oceananigans.

I saw that this issue was open: #3828 (comment)

And I'm wondering if that issue still persists, and if it carried over to ClimaOcean via the HydrostaticFreeSurfaceModel?

Again, thank you for answering my questions and helping me to continue to learn.

All the best,

Josh

[glwagner]

Nobody has started work implementing open boundaries for the hydrostatic model as far as I know. But there is clearly interest, so a critical mass could form, but I don't know when/if this will happen. There is an experimental implementation of open boundaries for the nonhydrostatic model. @jagoosw can say more. Honestly, it is probably not extremely difficult to implement open boundaries provided you understand the algorithm that other codes use (since this is quite standard...) so really the hard part is knowing what to do, not writing the code, exactly.

"Regional modeling" of some kind is still possible without open boundaries for some cases, but may require a larger domain / larger sponge layers than would otherwise be needed.

[glwagner]

I noticed that @ali-ramadhan has also been making noise about implementing open boundaries in the hydrostatic model.