Not quite ready. The ax args don't work at the moment, but figsize does. Perhaps there should be a windrose.subplots function to return the appropriate axes objects and provide a similar API?

You need to pass the projection like subplot_kw={"projection": "windrose"}. Here is a working example:

from windrose import plot_windrose
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

N = 500
ws = np.random.random(N) * 6
wd = np.random.random(N) * 360

df1 = pd.DataFrame({"speed": ws, "direction": wd})
df2 = pd.DataFrame({"speed": ws, "direction": wd})
fig, (ax1, ax2) = plt.subplots(ncols=2, subplot_kw={"projection": "windrose"})

plot_windrose(df1, ax=ax1)
plot_windrose(df2, ax=ax2)

I rebased this to run against the latest test and I'll add a subplot test to avoid regressions.

@scls19fr another one 😬

(Almost ready for the new release. I just want to check some issues and minor docs fixes.)