Полярный контурный график с нерегулярными «ячейками»

Вот как я обычно это делаю:

import numpy as np
import matplotlib.pyplot as plt
import matplotlib as mpl

data = np.loadtxt(r"C:\Users\serge.degossondevare\Downloads\data.txt")

def create_custom_bins(data):
    azimuths = []
    values = []
    # Offset elevations such that bin edge is at tick for that value
    elevations = np.linspace(0, 90, data.shape[0]) + 90/data.shape[0]/2
    num_values = np.count_nonzero(data, axis=1) # Number of nonzero values in each row, array
    for i in range(data.shape[0]):
        # Create rows of bins that map to the shape of the elevation data - 30 rows
        # Each row has a number of non-zero entries... These are how many bins we want
        # To align the az bins on the left edge we need to subtract half the binsize
        az_bins = np.linspace(0, 360, num_values[i], endpoint=False) + 360/num_values[i]/2
        azimuths.append(az_bins) 
        values.append(data[i, :num_values[i]])
    return elevations, azimuths, values

elevations, azimuths, values = create_custom_bins(data)
min = values[values>0].min()
max = values.max()

fig, ax = plt.subplots(subplot_kw=dict(projection='polar'))

for el, az, val in zip(elevations, azimuths, values):
    theta, r = np.meshgrid(np.radians(az), [el, el + 3])
    val = np.tile(val, (2, 1))
    # Important to set min max since each row will have a different scale based on that rows min/max.
    im = ax.pcolormesh(theta, r, val, norm=mpl.colors.LogNorm(vmin=min,vmax=max), shading='auto')

ax.set_theta_direction(-1)
ax.set_theta_offset(np.pi / 2.0)
ax.invert_yaxis()

cbar = fig.colorbar(im, ax=ax, orientation='vertical')
cbar.set_label('PFD [W/m2/s/Hz]')

plt.show()

который дает

Если вам нужен более плавный график, вариант предлагаемого решения

import numpy as np
import matplotlib.pyplot as plt
import matplotlib as mpl
from scipy.interpolate import griddata

data = np.loadtxt(r"C:\Users\serge.degossondevare\Downloads\data.txt")

def create_custom_bins(data):
    azimuths = []
    values = []
    elevations = np.linspace(0, 90, data.shape[0]) + 90/data.shape[0]/2
    num_values = np.count_nonzero(data, axis=1) # Number of nonzero values in each row, array
    for i in range(data.shape[0]):
        # Create rows of bins that map to the shape of the elevation data - 30 rows
        # Each row has a number of non-zero entries... These are how many bins we want
        # To align the az bins on the left edge we need to subtract half the binsize
        az_bins = np.linspace(0, 360, num_values[i], endpoint=False) + 360/num_values[i]/2
        azimuths.append(az_bins) 
        values.append(data[i, :num_values[i]])
    return elevations, azimuths, values

elevations, azimuths, values = create_custom_bins(data)

azimuth_flat = np.hstack([az for az in azimuths])
elevation_flat = np.hstack([[el]*len(az) for el, az in zip(elevations, azimuths)])
value_flat = np.hstack(values)

azimuth_pad = np.hstack([azimuth_flat, azimuth_flat + 360, azimuth_flat - 360])
elevation_pad = np.hstack([elevation_flat, elevation_flat, elevation_flat])
value_pad = np.hstack([value_flat, value_flat, value_flat])

azimuth_fine = np.linspace(0, 360, 361)
elevation_fine = np.linspace(0, 90, 180)
azimuth_grid, elevation_grid = np.meshgrid(azimuth_fine, elevation_fine)

value_grid = griddata((azimuth_pad, elevation_pad), value_pad, (azimuth_grid, elevation_grid), method='linear')

fig, ax = plt.subplots(subplot_kw=dict(projection='polar'))

theta = np.radians(azimuth_grid)
r = elevation_grid

im = ax.pcolormesh(theta, r, value_grid, norm=mpl.colors.LogNorm(), shading='auto', cmap='viridis')

ax.set_theta_direction(-1)
ax.set_theta_offset(np.pi / 2.0)
ax.invert_yaxis()

cbar = fig.colorbar(im, ax=ax, orientation='vertical')
cbar.set_label('PFD [W/m2/s/Hz]')

plt.show()