Assigning projections to plt.subplots()-generated axes generates rotated images about geoplot HOT 3 CLOSED

When you assign a projection to the figure at large, geoplot has no way of knowing where your "region of interest" will be once you start adding subplots. So you end up with...something.

If you calculate and pass the centroid as projection parameters:

square = shapely.geometry.Polygon([(41.5, 88), (41.75, 88), (41.75, 87.5), (41.5, 87.5)])
fig, ax = plt.subplots(1,1, subplot_kw={'projection': gplt.crs.AlbersEqualArea(central_latitude=87.75,
                                                                               central_longitude=41.625)})

gplt.polyplot(gpd.GeoDataFrame({'geometry':square}, index=[1]), projection=gplt.crs.AlbersEqualArea(), ax=ax)

You get what you're looking for:

The examples always hardcode the projection centroid for this reason.

from geoplot.

Ah, I see. Thanks for sorting me out!

from geoplot.

For future reference for anyone coming across this issue that stumbles across the GH issue: geoplot automatically determines the central latitude and/or central longitude of the projection you are using by taking the midpoint of the geopandas.GeoDataFrame.total_bounds of your input data.

When you pass a projection to matplotlib instead via plt.subplots, geoplot doesn't get to perform this intermediate step because the value is being set someplace else. You have to manually input the central coordinates for the projection yourself. For example:

proj = gcrs.AlbersEqualArea(central_longitude=-98, central_latitude=39.5)
f, axarr = plt.subplots(1, 2, figsize=(12, 4), subplot_kw={
    'projection': proj
})
gplt.polyplot(contiguous_usa, projection=proj, ax=axarr[0])
gplt.polyplot(contiguous_usa, projection=proj, ax=axarr[1])

This is discussed in the documentation: Working with Projections#Projections on Subplots.

from geoplot.