'ValueError: equally-spaced latitudes are invalid (non-global?)' when calculating VectorWind(u, v) about windspharm HOT 5 CLOSED

I think this because the IPSL-CM5A-LR model has a grid that includes the poles but has an even number of points in the latitude direction. The underlying software assumes that girds with an even number of points do not include the poles (and conversely grids with an odd number of latitude points must include the poles).

Regridding your data to a suitable grid may be an easy option since you are using iris.

from windspharm.

Thank you!
I tried regridding using: u.regrid(v, iris.analysis.Linear())

according to the example
rotated_air_temp = global_air_temp.regrid(rotated_psl, iris.analysis.Linear()) in http://scitools.org.uk/iris/docs/latest/userguide/interpolation_and_regridding.html,

but I am not sure this is correct. The code seems to take ages or to be stuck doing this...

from windspharm.

It may take a long time to regrid if you have a lot of data. You may want to look into caching the interpolator.

from windspharm.

Thank you. I tried interpolating..

        delta_latitude = 180/96.0
        sample_points = [('longitude', u.coord('longitude').points),('latitude',  np.linspace(90 - 0.5 *  
                                     delta_latitude,-90 + 0.5 * delta_latitude,96))]
        result = u.interpolate(sample_points, iris.analysis.Linear())
        print(result.summary(shorten=True))

which gives me as result
eastward_wind / (m s-1) (time: 13140; latitude: 96; longitude: 96)
northward_wind / (m s-1) (time: 13140; latitude: 96; longitude: 96)

but now I get the following error message when I want to compute w = VectorWind(u,v) (for u = result):
raise ValueError('u and v cannot contain missing values')
ValueError: u and v cannot contain missing values

How should I proceed?

from windspharm.

I've taken a look at the data on JASMIN, it looks like the horizontal interpolation step is successful (and necessary), but the problem is that this data set really does contain missing values. Try plotting the first level at the first time, you'll see what I mean. Windspharm cannot handle data with missing values.

The missing values are there because the data are already interpolated onto pressure levels, and the lower pressure levels intersect the surface in regions of high topography. Some model output chooses to extrapolate below ground (e.g., ERA) but in this case that has not been done.

If I leave out the lowest few pressure levels (the ones that intersect the surface) you can successfully create the VectorWind instance. If you really want the vorticity on the lower levels too, you'll either need to interpolate to fill in the gaps yourself or use a finite difference method to compute the vorticity.

from windspharm.