How to set 0 as the middle point in a matplotlib diverging colormap ?

Diverging colormaps are often used to visualize data that has a natural midpoint, such as temperature or depth. If the data does not have a natural midpoint, it is possible to use a diverging colormap by setting 0 as the middle point.

Diverging colormap

Let's construct a straightforward example using matplotlib with a diverging colormap:

    from pylab import figure, cm

    import matplotlib.pyplot as plt
    import numpy as np

    x1_min = -10.0
    x1_max = 5.0
    x2_min = -2.0
    x2_max = 2.0

    x1, x2 = np.meshgrid(np.arange(x1_min,x1_max, 0.1), np.arange(x2_min,x2_max, 0.1))

    plt.imshow(x1, origin='lower', cmap='seismic',extent=[x1_min,x1_max,x1_min,x1_max])

    plt.title("How to set 0 as the middle point \n in a matplotlib diverging colormap ?" , fontsize=12)

    plt.colorbar()

    plt.savefig("matplotlib_set_0_middle_point_matpotlib_diverging_colormap_01.png", bbox_inches='tight',dpi=200, facecolor='white')

    plt.show()

    plt.close()

Evidently, zero is not the middle point of the colormap.

Set 0 as the middle point

To set 0 as the middle point, a solution is to define a new norm:

custom_norm=colors.TwoSlopeNorm(vmin=x1_min, vcenter=0., vmax=x1_max)

Example:

    from pylab import figure, cm

    import matplotlib.pyplot as plt
    import numpy as np

    from matplotlib import colors
    x1_min = -10.0
    x1_max = 5.0
    x2_min = -2.0
    x2_max = 2.0

    x1, x2 = np.meshgrid(np.arange(x1_min,x1_max, 0.1), np.arange(x2_min,x2_max, 0.1))


    custom_norm=colors.TwoSlopeNorm(vmin=x1_min, vcenter=0., vmax=x1_max)

    plt.imshow(x1, origin='lower', cmap='seismic',extent=[x1_min,x1_max,x1_min,x1_max], norm=custom_norm)

    plt.title("How to set 0 as the middle point \n in a matplotlib diverging colormap ?" , fontsize=12)

    plt.colorbar()

    plt.savefig("matplotlib_set_0_middle_point_matpotlib_diverging_colormap_02.png", bbox_inches='tight',dpi=200, facecolor='white')

    plt.show()

    plt.close()

References