250-hPa Hemispheric Map using Python - Worked in August, not now....sigh

[fmarsik]

I am new to MetPy. In August, I was able to implement the Unidata Python Gallery's 250-hPa Hemispheric Map example presented on the following page:

https://unidata.github.io/python-gallery/examples/250hPa_Hemispheric_Plot.html

It worked brilliantly. I recently tried running this exact script again (not having done anything to the code), and I am now suddenly getting the following error. Any suggestions would be greatly appreciated.

peace,
Frank

---

IndexError Traceback (most recent call last)
~\AppData\Local\Temp\20\ipykernel_9536\4210448907.py in <cell line: 12>()
10
11 # Smooth the 250-hPa heights using a gaussian filter from scipy.ndimage
---> 12 hgt_250, lon = cutil.add_cyclic_point(data_hght.variables['Geopotential_height_isobaric'][:], coord=lon)
13 Z_250 = ndimage.gaussian_filter(hgt_250[0, 0, :, :], sigma=3, order=0)
14

~.conda\envs\metpy\lib\site-packages\cartopy\util.py in add_cyclic_point(data, coord, axis)
77 f'data.shape[{axis}] = {data.shape[axis]}.')
78 delta_coord = np.diff(coord)
---> 79 if not np.allclose(delta_coord, delta_coord[0]):
80 raise ValueError('The coordinate must be equally spaced.')
81 new_coord = ma.concatenate((coord, coord[-1:] + delta_coord[0]))

~.conda\envs\metpy\lib\site-packages\numpy\ma\core.py in getitem(self, indx)
3220 # mask of being reshaped if it hasn't been set up properly yet
3221 # So it's easier to stick to the current version
-> 3222 dout = self.data[indx]
3223 _mask = self._mask
3224

IndexError: index 0 is out of bounds for axis 0 with size 0

[dopplershift]

Sorry you've run into this problem. What I think happened is that in August we upgraded our THREDDS data server to its next major version, v5. That subtly changed some behavior this example was relying on. The problematic lines are:

gfsdata_hght.lonlat_box(0, 360, 0, 90)

and

gfsdata_wind.lonlat_box(0, 360, 0, 90)

On the new version of the server, 0 and 360 seem to be treated the same, so the request ends up getting a single longitude line of data. The quick fix is to not quite request all the way to 360, like:

gfsdata_hght.lonlat_box(0, 359.9, 0, 90)

and

gfsdata_wind.lonlat_box(0, 359.9, 0, 90)

With that change made, I get the expected image:

[dopplershift]

You might also want to use the current copy of the example which makes more use of the MetPy<->xarray integration to simplify it a bit.

[fmarsik]

Excellent. This did, indeed, fix things at my end, as well. Thanks also for directing me to the updated copy.