I have netcdf dataset that I think is regular gridded dataset using a Rotated pole mapping, but the rotated pole information is wrong.
For those not familiar with Rotated Poles, the cartopy documentation has a description of the projection. Check out the source code for the image for a clearer understanding - the Rotated North pole is at 37.5N, 175E (in standard lat/lon coorinates).
The dataset has lat/lon value per grid point, but rotating those into the Rotated pole using the pole data provided results in a nearly regular grid, but it is slightly curved, and rotated. This makes me think that the pole coordinates are wrong.
Is there any simple way to estimate a rotated north pole, given a grid of lat/lons? I guess it should be possible using a linear regression in spherical space over each column of the grid, and then finding the intersection point of all the grid lines, but that seems complex and I'm wondering if there are any pre-existing algorithms for this.