Merge pull request #2502 from rcomer/map-latitude

DOC: add example aligning Cartopy and Matplotlib axes

Author: greglucas

examples/gridlines_and_labels/map_latitudes.py

@@ -0,0 +1,82 @@
+"""
+Aligning Cartopy and Matplotlib axes
+====================================
+
+Sometimes it is useful to line up a standard Matplotlib axes with a Cartopy map, for
+example when plotting zonal mean data next to a global contour plot.  This example
+demonstrates one way to do that for projections where latitude lines are horizontal.
+
+"""
+
+# sphinx_gallery_multi_image = "single"
+
+import matplotlib.pyplot as plt
+import numpy as np
+
+import cartopy.crs as ccrs
+
+
+PC_PROJ = ccrs.PlateCarree()
+
+def map_and_latitudes(map_proj):
+    """
+    Create a map with the given projection, together with a rectangular axes
+    whose y-axis matches the map.
+    """
+
+    fig = plt.figure(figsize=(12, 5), layout='compressed')
+
+    # Create the map.
+    map_ax = fig.add_subplot(projection=map_proj)
+    map_ax.set_global()
+    gl = map_ax.gridlines(draw_labels=True)
+    map_ax.coastlines()
+
+    # Add the rectangular axes to the right of the map.
+    rect_ax = map_ax.inset_axes([1.1, 0, 0.3, 1])
+
+    # Set up scaling for the rectangular axes latitudes.  This part is
+    # necessary for projections with uneven latitude spacing, such as Mercator,
+    # but can be skipped for PlateCarree for example.
+    # Create functions to map between latitude and the projection's y values.
+    # Assume these are not x-dependent so we can just set x/longitude to zero.
+
+    def to_map(lat):
+        """Transform latitudes to projection's y-values."""
+        dummy_lon = np.zeros_like(lat)
+        projected = map_proj.transform_points(PC_PROJ, dummy_lon, lat)
+        return np.take(projected, 1, axis=-1)
+
+    def from_map(y):
+        """Transform projection's y-values to latitude."""
+        dummy_x = np.zeros_like(y)
+        projected = PC_PROJ.transform_points(map_proj, dummy_x, y)
+        return np.take(projected, 1, axis=-1)
+
+    # Use these function to define the y-scale on the rectangular axes.
+    rect_ax.set_yscale('function', functions=(to_map, from_map))
+    # Convert the map's y-limits to latitudes and set the y-limits.
+    rect_ax.set_ylim(from_map(np.array(map_ax.get_ylim())))
+
+    # y-axis visual choices.
+    rect_ax.grid(axis='y')
+    rect_ax.yaxis.tick_right()
+    rect_ax.yaxis.set_major_locator(gl.ylocator)
+    rect_ax.yaxis.set_major_formatter(gl.yformatter)
+
+    # Plot some dummy data.
+    y = np.linspace(-90, 90, 21)
+    y_rad = np.deg2rad(y)
+    rect_ax.plot(np.sin(y_rad), y, label='sine')
+    rect_ax.plot(np.cos(y_rad), y, label='cosine')
+    rect_ax.legend()
+
+    # Add the projection name as the title.
+    fig.suptitle(type(map_proj).__name__)
+
+
+if __name__ == '__main__':
+    for map_proj in [ccrs.PlateCarree(), ccrs.Mercator(), ccrs.Robinson()]:
+        map_and_latitudes(map_proj)
+
+    plt.show()