LambertConformal projection maximal longitude extent

[leonard-seydoux]

Rationale

The previous version of LambertConformal projection does not allow to control the longitudinal extent of the map (it was hardcoded at 180º). It was only possible to use the set_extent method of a GeoAxes instance, resulting in vertical (or horizontal) axes boundaries. Yet, a Lambert conformal projection can have arbitrary longitude extent. This version implements a control on this extent.

Implications

Being able to control the longitude extent of a GeoAxes instance with a single keyword argument added to LambertConformal.

Checklist

• I have read and signed the Contributor Licence Agreement (CLA) just before proposing this pull request.

• New feature: keyword argument longitude_extent added to the LambertConformal CRS. By default, the behavior of he projection is unchanged (i.e. longitude_extent=180). When specified, the extent of the GeoAxes with LambertConformal projection with have axes boundaries drawn within the range (central_longitude - longitude_extent, central_longitude + longitude_extent).

• The docstring was updated accordingly.

Example use (compare with and without controlling the longitude extent)

from cartopy import crs, feature
from matplotlib import pyplot as plt

# create figure
fig = plt.figure()

# original behaviour
proj = crs.LambertConformal(central_longitude=-100)
ax = fig.add_subplot(121, projection=proj)
ax.add_feature(feature.LAND, facecolor="0.9")
ax.gridlines()
ax.set_title("Default extent")

# new possibility
proj = crs.LambertConformal(central_longitude=-100, longitude_extent=60)
ax = fig.add_subplot(122, projection=proj)
ax.add_feature(feature.LAND, facecolor="0.9")
ax.gridlines()
ax.set_title("Controled extent")

# show maps
plt.show()

[dopplershift]

This seems reasonable since you can't replicate this by calling set_extent(). Can you add a test? Maybe checking the boundary when you pass longitude_extent?

[leonard-seydoux]

Sure, but I do not see what test you have in mind?

[marqh]

hello @leonard-seydoux

you can see the current testing of the
lambert conformal projection
in
https://github.com/SciTools/cartopy/blob/master/lib/cartopy/tests/crs/test_lambert_conformal.py

perhaps you could consider a new test case which explicitly uses this new longitude_extent parameter, verify that it exists and check the resulting longitude_extent as suggested by @dopplershift

does this pattern seem suitable for adoption into this PR?

marqh

[CLAassistant]

All committers have signed the CLA.