Legend overlaps when visualizing a DataArray in xarray

[mehmetgoktug]

Description

When visualizing a DataArray using xarray, the legend overlaps and becomes cluttered. This issue does not occur when I use the same script in Visual Studio Code.

Reproduce

import xarray as xr
import matplotlib.pyplot as plt

ds = xr.open_dataset('data.nc')
ds

ds_temp = ds['t']
ds_temp = ds_temp - 273.15
ds_min = ds_temp.min('valid_time')

ds_min.plot()
ds_min.plot()
ds_min.plot()
ds_min.plot.contourf()

Data: link

Screencast:

Screencast.from.2025-01-13.11-03-11.webm

System Information

Operating System: Ubuntu 24.04.1 LTS

Positron Version:
Positron: 2025.01.0 build 152
Positron SHA: 66aa3fb
Code OSS: 1.95.0
Architecture: x64

Interpreter: 3.12.3

[sharon-wang]

Hi @mehmetgoktug! Do you have a screenshot or video of the plot in Visual Studio Code? Are you expecting that a new plot is redrawn instead of the legend getting added to the existing plot?

---

I was able to reproduce with the following:

pip install matplotlib numpy xarray pooch

# based on https://tutorial.xarray.dev/overview/xarray-in-45-min.html

import matplotlib.pyplot as plt
import numpy as np
import xarray as xr

xr.set_options(keep_attrs=True, display_expand_data=False)
np.set_printoptions(threshold=10, edgeitems=2)

# load the data set
ds = xr.tutorial.load_dataset("air_temperature")

lat = ds.air.lat.data  # numpy array
lon = ds.air.lon.data  # numpy array
temp = ds.air.data  # numpy array

# plot the data
plt.pcolormesh(lon, lat, temp[0, :, :]);

# run this next line multiple times until the text starts overlapping
ds.air.isel(time=0).plot(x="lon");

CC @seeM for plot knowledge 😄

[seeM]

I think this is expected behavior but it would be helpful to see the same in VSCode. Was that in a notebook or interactive window? And was it executed line-by-line?

Positron's plotting is intended to be interactive line-by-line. The various plot() methods here might be using a matplotlib function that adds a new legend without checking if a previous exists. You could consider opening an issue in the xarray repo to allow repeated executions.

Or you can work around this by creating a new figure before each plot() call with plt.figure(). That's what would happen behind the scenes in a Jupyter notebook.

[mehmetgoktug]

@sharon-wang @seeM thank you for your reply. Using plt.close() after a plot or plt.figure() before a plot solved my problem. Since I am new to Python, I wasn’t aware of these functions. As far as I know, Jupyter and interactive console don’t require plt. functions. I have been plotting graphs directly in VS Code with these. I have also been using plt.show() in Python 3 and the IPython console. However, I used to close the window manually, so I hadn’t needed to use plt.close() before.