In some situations, we have several subplots and we want to use only one colorbar for all the subplots. How to do this in Matplotlib? Two ways can be employed.

The conventional method

The first method is like normal plotting: first draw the main plot, then add a colorbar to the main plot. Matplotlib provide different ways to add a colorbar: explicit or implicit way.

The explicit way

The idea is to adjust the existing axes manually to make room for an additional colorbar. Then explicitly add an axes where the colorbar resides. See the code below for details:

import matplotlib.pyplot as plt
import numpy as np

fig, axes = plt.subplots(nrows=2, ncols=3, figsize=(8.5, 5))

for ax in axes.flat:
    ax.set_axis_off()
    im = ax.imshow(np.random.random((16, 16)), cmap='viridis',
                   vmin=0, vmax=1)

fig.subplots_adjust(bottom=0.1, top=0.9, left=0.1, right=0.8,
                    wspace=0.02, hspace=0.02)

# add an axes, lower left corner in [0.83, 0.1] measured in figure coordinate with axes width 0.02 and height 0.8

cb_ax = fig.add_axes([0.83, 0.1, 0.02, 0.8])
cbar = fig.colorbar(im, cax=cb_ax)

 set the colorbar ticks and tick labels
cbar.set_ticks(np.arange(0, 1.1, 0.5))
cbar.set_ticklabels(['low', 'medium', 'high'])

plt.show()

In this way, we can control the position of colorbar precisely. The output image is like this:

The implicit way

Matplotlib also offers method which can adjust the existing axes and make room for a colorbar implicitly. See the code below for an example:

import matplotlib.pyplot as plt
import numpy as np

fig, axes = plt.subplots(nrows=2, ncols=3, figsize=(8.5, 5))

for ax in axes.flat:
    ax.set_axis_off()
    im = ax.imshow(np.random.random((16, 16)), cmap='viridis',
                   vmin=0, vmax=1)

# notice that here we use ax param of figure.colorbar method instead of

# the cax param as the above example

cbar = fig.colorbar(im, ax=axes.ravel().tolist(), shrink=0.95)

cbar.set_ticks(np.arange(0, 1.1, 0.5))
cbar.set_ticklabels(['low', 'medium', 'high'])

plt.show()

In this way, you have to manually tweak the shrink param of fig.colorbar method to make the main plot and the colorbar appear the same height. See the output image below

Both the two methods have an disadvantage that it is difficult to control the padding space between subplots. You have to adjust the figure aspect ratio and also the padding params to make the padding between the subplots appear the same. In fact, the padding in horizontal and vertical direction is not the same for the above two plots even after tweaking.

Using the axesgrid approach

Matplotlib also provides a AxesGrid toolkit to deal with padding and colorbar issues arising from plotting multiple subplots. By using axesgrid, the padding between subplots are guaranted to be the same. Also the colorbar have exactly the same height as the main plot. Following is a working example showing how to use axesgrid:

import matplotlib.pyplot as plt
from mpl_toolkits.axes_grid1 import AxesGrid
import numpy as np

fig = plt.figure(figsize=(6, 4))

grid = AxesGrid(fig, 111,
                nrows_ncols=(2, 3),
                axes_pad=0.05,
                cbar_mode='single',
                cbar_location='right',
                cbar_pad=0.1
                )

for ax in grid:
    ax.set_axis_off()
    im = ax.imshow(np.random.random((16,16)), vmin=0, vmax=1)

# when cbar_mode is 'single', for ax in grid, ax.cax = grid.cbar_axes[0]

cbar = ax.cax.colorbar(im)
cbar = grid.cbar_axes[0].colorbar(im)

cbar.ax.set_yticks(np.arange(0, 1.1, 0.5))
cbar.ax.set_yticklabels(['low', 'medium', 'high'])
plt.show()

See the output image below.

You can see that the padding between subplots are all the same, also the colorbar have the same height as the main plot. Conveniently, isn’t it?

Summary

Using the normal way is more flexible but also annoying because you have to adjust the paramters by trial and error. By employing the axesgrid, you can simplify the plotting of multiple plot with just one colorbar, significantly. In my opinion, the latter way is prefered.

References