在matplotlib中用两个Y轴（两个单位）绘制单个数据

Question

I'm trying to plot a single time series, but want to represent it in two units on left and right axes. Here's what I have done so far.

fig, ax1 = plt.subplots()
t = np.arange(1,50,1)
s1 = np.sin(t)*20000+40000 #synthetic ts, but closer to my data 
ax1.plot(t, s1, 'b-')

ax1.set_xlabel('time')
ax1.set_ylim(20000,70000)
ax1.set_ylabel('km3/year')

km3yearToSv=31.6887646*(1/1e6)

ax2 = ax1.twinx()
s2 = s1*km3yearToSv
ax2.plot(t, s2, 'b-')
ax2.set_ylim(20000*km3yearToSv,70000*km3yearToSv)
ax2.set_ylabel('Sv')

By adjusting the ylim(), I can get it to seem as a single line, but some aliasing can be seen. I would prefer if I don't have to plot the data twice.

Any suggestions?

UPDATE: Thanks, askewchan, for the perfect solution!

Answer 1

There is no need to plot it twice, this much should give you the result you want:

ax2 = ax1.twinx()
ax2.set_ylim(20000*km3yearToSv, 70000*km3yearToSv)
ax2.set_ylabel('Sv')

A more robust way to do it is to first extract the limits of the plot (in case you change them, and they're no longer 20000 and 70000 , or you want the plot to be able to automatically adjust the limits:

ax2 = ax1.twinx()
mn, mx = ax1.get_ylim()
ax2.set_ylim(mn*km3yearToSv, mx*km3yearToSv)
ax2.set_ylabel('Sv')

Altogether, with some other small tweaks:

import numpy as np
import matplotlib.pyplot as plt

mean, amp = 40000, 20000
t = np.arange(50)
s1 = np.sin(t)*amp + mean #synthetic ts, but closer to my data 

fig, ax1 = plt.subplots()
ax1.plot(t, s1, 'b-')

ax1.set_xlabel('time')
mn, mx = ax1.set_ylim(mean-amp, mean+amp)
ax1.set_ylabel('km$^3$/year')

km3yearToSv = 31.6887646e-6

ax2 = ax1.twinx()
ax2.set_ylim(mn*km3yearToSv, mx*km3yearToSv)
ax2.set_ylabel('Sv')

Answer 2

one small error in code above:

mn, mx = ax2.get_ylim()

should be:

mn, mx = ax1.get_ylim()

to scale new secondary axis based on original.