这里有一个解决方案，它使用ax.annotate在每个条内绘制箭头。由于OP不太清楚箭头应该是什么样子，所以我把每个条分成矩形(在代码中我称它们为squares，但如果你确定了绘图的纵横比，它们就只是正方形)，并为每个矩形绘制了一个中心箭头，其方向由用户提供的角度给出(这里是一个称为wind_direction)。在

在代码中，我将Axes的纵横比设置为x-和y-极限的纵横比，这使得Axes呈正方形，因此可以轻松绘制长度相同的箭头，而不受其方向的影响。如果不需要，可以注释掉相应的行。如果箭头的长度必须相同而不受限制，则必须计算图形的纵横比，例如请参见here如何进行此操作。在

我还用度数标注了每个条上的风向，以便检查箭头是否与给定的风向一致。在from matplotlib import pyplot as plt

import numpy as np

fig,ax = plt.subplots()

x = np.arange(12)

wind_strength = np.random.random(12)

wind_direction = np.linspace(0,2*np.pi,12, endpoint = False)

colors = ['green', 'yellow', 'blue', 'pink', 'orange']

bar_cols = [colors[i%len(colors)] for i,s in enumerate(wind_strength)]

bars = ax.bar(x,wind_strength, color=bar_cols)

##computing the aspect ratio of the plot ranges:

xlim = ax.get_xlim()

ylim = ax.get_ylim()

aspect = (xlim[1]-xlim[0])/(ylim[1]-ylim[0])

##comment out this line if you don't care about the arrows being the

##same length

ax.set_aspect(aspect)

##dividing each bar into 'squares' and plotting an arrow into each square

##with orientation given by wind direction

for bar,angle in zip(bars,wind_direction):

(x1,y1),(x2,y2) = bar.get_bbox().get_points()

w = x2-x1

h = w/aspect

x_mid = (x1+x2)/2

dx = np.sin(angle)*w

dy = np.cos(angle)*h

##the top and bottom of the current square:

y_bottom = y1

y_top = y_bottom+h

##draw at least one arrow (for very small bars)

first = True

while y_top < y2 or first:

y_mid = (y_top+y_bottom)/2

ax.annotate(

'',xytext=(x_mid-dx/2,y_mid-dy/2),

xy=(x_mid+dx/2,y_mid+dy/2),

arrowprops=dict(arrowstyle="->"),

)

##next square

y_bottom = y_top

y_top += h

##first arrow drawn:

first = False

##annotating the wind direction:

ax.text(x_mid, y2+0.05, '{}'.format(int(180*angle/np.pi)), ha = 'center')

plt.show()

最终结果如下：

希望这有帮助。在