Plotting inconsistencies with Cartopy

[andreas-h]

I'm on xarray 0.11.3, and currently do not have the possibility to test on master. However, the commit history suggests that this issue still exists. Please accept my apologies if I missed something.

I'm following the plotting documentation at https://xarray.pydata.org/en/stable/plotting.html#maps

When plotting two maps in the same call (exactly as written in the documentation),

p = air.isel(time=[0, 1]).plot(transform=ccrs.PlateCarree(), col='time',
                               subplot_kws={'projection': ccrs.Orthographic(-80, 35)})

for ax in p.axes.flat:
    ax.coastlines()
    ax.gridlines()

everything works fine.

However, when I only want to plot one map,

p = air.isel(time=[0]).plot(transform=ccrs.PlateCarree(),
                            subplot_kws={'projection': ccrs.Orthographic(-80, 35)})

for ax in p.axes.flat:
    ax.coastlines()
    ax.gridlines()

I get the following error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-63-2c5a2a533878> in <module>
      1 p = air.isel(time=[0]).plot(transform=ccrs.PlateCarree(),
----> 2                             subplot_kws={'projection': ccrs.Orthographic(-80, 35)})
      3 
      4 for ax in p.axes.flat:
      5     ax.coastlines()

/home/jupyterhub/conda/envs/prod/lib/python3.6/site-packages/xarray/plot/plot.py in __call__(self, **kwargs)
    550 
    551     def __call__(self, **kwargs):
--> 552         return plot(self._da, **kwargs)
    553 
    554     @functools.wraps(hist)

/home/jupyterhub/conda/envs/prod/lib/python3.6/site-packages/xarray/plot/plot.py in plot(darray, row, col, col_wrap, ax, hue, rtol, subplot_kws, **kwargs)
    185     kwargs['ax'] = ax
    186 
--> 187     return plotfunc(darray, **kwargs)
    188 
    189 

/home/jupyterhub/conda/envs/prod/lib/python3.6/site-packages/xarray/plot/plot.py in newplotfunc(darray, x, y, figsize, size, aspect, ax, row, col, col_wrap, xincrease, yincrease, add_colorbar, add_labels, vmin, vmax, cmap, center, robust, extend, levels, infer_intervals, colors, subplot_kws, cbar_ax, cbar_kwargs, xscale, yscale, xticks, yticks, xlim, ylim, norm, **kwargs)
    852                              vmax=cmap_params['vmax'],
    853                              norm=cmap_params['norm'],
--> 854                              **kwargs)
    855 
    856         # Label the plot with metadata

/home/jupyterhub/conda/envs/prod/lib/python3.6/site-packages/xarray/plot/plot.py in pcolormesh(x, y, z, ax, infer_intervals, **kwargs)
   1106             y = _infer_interval_breaks(y, axis=0)
   1107 
-> 1108     primitive = ax.pcolormesh(x, y, z, **kwargs)
   1109 
   1110     # by default, pcolormesh picks "round" values for bounds

/home/jupyterhub/conda/envs/prod/lib/python3.6/site-packages/matplotlib/__init__.py in inner(ax, data, *args, **kwargs)
   1808                         "the Matplotlib list!)" % (label_namer, func.__name__),
   1809                         RuntimeWarning, stacklevel=2)
-> 1810             return func(ax, *args, **kwargs)
   1811 
   1812         inner.__doc__ = _add_data_doc(inner.__doc__,

/home/jupyterhub/conda/envs/prod/lib/python3.6/site-packages/matplotlib/axes/_axes.py in pcolormesh(self, alpha, norm, cmap, vmin, vmax, shading, antialiased, *args, **kwargs)
   6011         if (not isinstance(t, mtransforms.Transform) and
   6012             hasattr(t, '_as_mpl_transform')):
-> 6013             t = t._as_mpl_transform(self.axes)
   6014 
   6015         if t and any(t.contains_branch_seperately(self.transData)):

lib/cartopy/_crs.pyx in cartopy._crs.CRS._as_mpl_transform()

ValueError: Axes should be an instance of GeoAxes, got <class 'matplotlib.axes._subplots.AxesSubplot'>

It is not clear (from the user perspective) why one should work and the other should not.

Of course I can create the axes instance before using xarray's plot function, but it would be nicer if I wouldn't have to.

[shoyer]

I agree, this is unfortunate. I suspect the source of the discrepancy is that we use a totally different codepath if either col or row is supplied, creating our own subplot grid.

[ahuang11]

I think this is fixed in the latest master.

import xarray as xr
import numpy as np
import cartopy.crs as ccrs

da = xr.tutorial.open_dataset('air_temperature')['air']

p = da.isel(time=[0, 1]).plot(
    transform=ccrs.PlateCarree(), col='time',
    subplot_kws={'projection': ccrs.Orthographic(-80, 35)}
)

for ax in p.axes.flat:
    ax.coastlines()
    ax.gridlines()

p = da.isel(time=0).plot(
    transform=ccrs.PlateCarree(),
    subplot_kws={'projection': ccrs.Orthographic(-80, 35)}
)

However if you try iterating over the axes, it will crash because p is quadmesh type not a facetgrid type so I was wondering whether it'd be good to have an keyword like force_facetgrid_type=True which uses the FacetGrid class to plot a single map so that the user can keep the same code for single axes and facet axes?

p = da.isel(time=0).plot(
    transform=ccrs.PlateCarree(),
    subplot_kws={'projection': ccrs.Orthographic(-80, 35)}
)

for ax in p.axes.flat:
    ax.coastlines()
    ax.gridlines()

[mathause]

As a workaround you can use:

p = da.isel(time=[0]).plot.pcolormesh(
    col="time",
    transform=ccrs.PlateCarree(),
    subplot_kws={'projection': ccrs.Orthographic(-80, 35)}
)

Note the time=[0] - using time=0 squezes the time dimension. Also note the .plot.pcolormesh. Using just .plot also calls daarray.squeeze().

https://github.com/pydata/xarray/blob/8fab5a2449d8368251f96fc2b9d1eaa3040894e6/xarray/plot/plot.py#L163

[stale[bot]]

In order to maintain a list of currently relevant issues, we mark issues as stale after a period of inactivity

If this issue remains relevant, please comment here or remove the stale label; otherwise it will be marked as closed automatically