Regional Arctic simulation to 90°N: Grid configuration guidance needed

[ruoyan88]

Question: Extending Arctic simulation to North Pole (90°N)

Background

I am working on a regional Arctic ocean-sea ice coupled simulation using ClimaOcean. I started from the arctic_simulation.jl example, fixed several bugs in the original code, and now have it running successfully with the following configuration:

grid = RotatedLatitudeLongitudeGrid(arch, 
                                    size = (Nx, Ny, Nz),
                                    latitude = (65, 85),
                                    longitude = (-180, 180),
                                    z = z,
                                    north_pole = (0, -90),
                                    halo = (5, 5, 4),
                                    topology = (Periodic, Bounded, Bounded))

My Questions

1. Can I extend the current RotatedLatitudeLongitudeGrid to cover the North Pole (90°N)?

For example, if I change to latitude = (65, 90), would this work given that the grid north pole is rotated to (0, -90)? Or will I still encounter polar singularity issues despite the rotation?

2. Is it possible to configure a regional TripolarGrid for the Arctic?

I understand that TripolarGrid (from OrthogonalSphericalShellGrids.jl) can handle polar singularities by placing two grid poles over land. However, all examples I've seen use global domains. Can TripolarGrid be configured for a regional Arctic domain (e.g., 65°N-90°N) instead of requiring a full global grid?

3. What is the recommended approach for regional Arctic simulations including the North Pole?

My goal is to simulate the entire Arctic basin from 65°N to 90°N with higher resolution than would be feasible for a global domain.

---

System Information:

• ClimaOcean version: [0.8.4]
• Oceananigans version: [0.97.6, 0.98]
• Julia version: [v1.10.10]

Note: If needed, I can provide my complete working code (with the fixes I made to the original example) for reference.

[glwagner]

The parameter north_pole sets the pole of the grid. So for a polar simulation, you probably want to place the north pole at the equator, north_pole = (0, 0). You do not want to put the north pole on the other side. There are always two poles in a latitude longitude grid; therefore flipping the poles has the same singularity issue as an normal LatitudeLongitudeGrid. Your goal is to move the "grid pole" away from the "geographic pole".

The other arguments are associated with the latitude and longitude of the grid before rotation. Therefore you have to compute latitude and longitude such that your grid covers the arctic. I think if you use north_pole = (0, 0) you want something like latitude=(-20, 20) for 40 degrees of coverage. You also don't want longitude all the way around.

[ruoyan88]

Thank you for the clear explanation! I understand now.

The key point I was missing: the latitude and longitude parameters are grid coordinates (before rotation), not the geographic coordinates I want to simulate. I need to compute the grid coordinates that will map to my target geographic region (65°N-90°N) after rotation.

So for my Arctic simulation covering geographic 65°N-90°N, if I use north_pole = (0, 0):

Questions:

1. How do I calculate the correct latitude and longitude ranges (in grid coordinates) that will cover this geographic region after rotation?

2. Is there a utility function or formula I can use for this coordinate transformation? Or should I work backwards from the rotate_coordinates function in the source code?

3. For a typical Arctic regional simulation (65-90°N geographically), what would be a reasonable starting configuration? For example:

   north_pole = (0, 0)
   latitude = (-25, 25)  # grid coordinates - is this appropriate?
   longitude = (?, ?)     # what range would cover the Arctic?

I appreciate your help in understanding this correctly!