Change map projection in Python charts

Libraries & Data

For creating the charts, we will need to load the following libraries:

• matplotlib for creating the chart and the arrows
• pandas for loading the data
• geopandas for loading the world boundaries
• cartopy for changing the projections

import pandas as pd
import matplotlib.pyplot as plt

import geopandas as gpd
import cartopy.crs as ccrs

# set a higher resolution
plt.rcParams['figure.dpi'] = 300

# load the world dataset
url = "https://raw.githubusercontent.com/holtzy/The-Python-Graph-Gallery/master/static/data/all_world.geojson"
world = gpd.read_file(url)
world.head()

What's a projection

How can we make a 2D version of a 3D object like the Earth? This challenge is tackled through map projections, which transform the Earth's spherical surface into a flat map. Projections are crucial because they inevitably distort some aspects of the Earth's surface, such as shape, area, distance, or direction.

The choice of projection can significantly alter the message conveyed by a map; for instance, using a Mercator projection preserves direction but distorts the size of landmasses, making regions near the poles appear disproportionately large and potentially misleading viewers about the true scale of countries.

Default map

fig, ax = plt.subplots(figsize=(12, 8))

ax.set_axis_off() # remove border around the axes
world.plot(ax=ax)

plt.show()

This projection is called PlateCarree. The default Matplotlib map plot often shows countries with varying shapes. Countries near the equator appear relatively normal, while those farther from the equator, especially near the poles, look distorted.

This occurs because the projection stretches landmasses more as they move away from the equator, causing countries near the poles to spread horizontally along the x-axis.

The mercator projection

projection = ccrs.Mercator()
world_merc = world.to_crs(projection.proj4_init)

fig, ax = plt.subplots(figsize=(12, 8), subplot_kw={"projection": projection})

ax.set_axis_off() # remove border around the axes
world_merc.plot(ax=ax)

plt.show()

Our map looks way better, but the main problem with Mercator is that, for example, Greenland appears to be larger than Africa when in fact it is 14 times smaller!

Bubble maps and projections

A common map type is the bubble map, which overlays a scatter plot on a map, using longitudes as x-axis values and latitudes as y-axis values. However, simply changing the map projection doesn’t automatically adjust the scatter plot points accordingly.

Changing the projection alters the representation of the world's shape, thus shifting the actual locations of given latitudes and longitudes. Consequently, we must recalculate the new positions of each scatter plot point. Let's begin by loading a dataset with earthquake positions:

#Load data
url = "https://raw.githubusercontent.com/holtzy/The-Python-Graph-Gallery/master/static/data/earthquakes.csv"
df = pd.read_csv(url)
df = df[df['Depth (km)']>=0.01] # depth of at least 10 meters
df.sort_values(by='Depth (km)', ascending=False, inplace=True)
df.head()

fig, ax = plt.subplots(figsize=(12, 8))

ax.set_axis_off()
world.plot(ax=ax)
ax.scatter(
   x=df['Longitude'], y=df['Latitude'], s=df['Depth (km)']/3,
   color='darkred', alpha=0.3
)

plt.show()

projection = ccrs.Mercator()
previous_proj = ccrs.PlateCarree()
new_coords = projection.transform_points(previous_proj, df['Longitude'].values, df['Latitude'].values)
world_merc = world.to_crs(projection.proj4_init)

fig, ax = plt.subplots(figsize=(12, 8), dpi=300, subplot_kw={'projection':projection})

ax.set_axis_off()
world_merc.plot(ax=ax)

# transform the coordinates to the projection's CRS
ax.scatter(
   x=new_coords[:, 0], y=new_coords[:, 1], s=df['Depth (km)']/3,
   color='darkred', alpha=0.3
)

plt.show()

Other projections

The number of different projections is in fact very large. Below is an exhaustive list of the projections available in cartopy (the first ones are the most common ones, and recommended for most use cases):z

• PlateCarree(): most basic projection
• AlbersEqualArea(): preserves area, conical
• AzimuthalEquidistant(): preserves distance from center
• EquidistantConic(): preserves distance along meridians
• LambertConformal(): preserves shape, conical
• LambertCylindrical(): equal-area cylindrical
• Mercator(): preserves direction, exaggerates poles
• Miller(): compromise between Mercator and cylindrical
• Mollweide(): equal-area, elliptical
• ObliqueMercator(): rotated Mercator for oblique aspects
• Orthographic(): view from infinite distance
• Robinson(): pseudo-cylindrical, compromise
• Sinusoidal(): equal-area, pseudo-cylindrical
• Stereographic(): conformal, azimuthal
• TransverseMercator(): Mercator rotated 90 degrees
• UTM(): grid-based, preserves shape locally
• InterruptedGoodeHomolosine(): interrupted equal-area
• RotatedPole(): shifts pole position
• OSGB(): British National Grid
• EuroPP(): Europe polar stereographic
• Geostationary(): view from geostationary orbit
• NearsidePerspective(): view from finite distance
• EckertI(): pseudo-cylindrical, equal-area
• EckertII(): pseudo-cylindrical, equal-area
• EckertIII(): pseudo-cylindrical, compromise
• EckertIV(): pseudo-cylindrical, equal-area
• EckertV(): pseudo-cylindrical, compromise
• EckertVI(): pseudo-cylindrical, equal-area
• Aitoff(): modified azimuthal, compromise
• EqualEarth(): equal-area, pseudo-cylindrical
• Gnomonic(): straight great circles
• Hammer(): equal-area, elliptical
• LambertAzimuthalEqualArea(): equal-area, azimuthal
• NorthPolarStereo(): stereographic centered on North Pole
• OSNI(): Northern Ireland Grid
• SouthPolarStereo(): stereographic centered on South Pole

Going further

You might be interested in:

• learn more about maps in python
• how to create an interactive map with Folium
• how to create and customize bubble maps