Plotting xarray datasets with variable coordinates

Question

I'm trying to use xarray to plot data on a variable grid. The grid that my data is stored on is changing over time, but keeps the same dimensions.

I'd like to be able to plot 1d slices of it at a given time. A toy example of what I'm trying to do is shown below.

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

time = [0.1, 0.2] # i.e. time in seconds

# a 1d grid changes over time, but keeps the same dims
radius = np.array([np.arange(3),
                   np.arange(3)*1.2])

velocity = np.sin(radius) # make some random velocity field

ds = xr.Dataset({'velocity': (['time', 'radius'],  velocity)},
            coords={'r': (['time','radius'], radius), 
                    'time': time})

If I try to plot it at different times, i.e.

ds.sel(time=0.1)['velocity'].plot()
ds.sel(time=0.2)['velocity'].plot()
plt.show()

But I'd like it to replicate the behavior that I can do explicitly using matplotlib. Here it properly plots the velocity against the radius at that time.

plt.plot(radius[0], velocity[0])
plt.plot(radius[1], velocity[1])
plt.show()

I may be using xarray wrong, but it should be plotting the velocity against the proper value of radius at that time.

Am I setting up the Dataset wrong or using the plot/index feature wrong?

jhamman · Accepted Answer · 2017-12-18 18:24:39Z

1

I agree this behavior is unexpected but its not entirely a bug.

Looking at the variables you're trying to plot:

da = ds.sel(time=0.2)['velocity']
print(da)

yields:

<xarray.DataArray 'velocity' (radius: 3)>
array([ 0.      ,  0.932039,  0.675463])
Coordinates:
    r        (radius) float64 0.0 1.2 2.4
    time     float64 0.2
Dimensions without coordinates: radius

What we see is that there is not a coordinate variable named radius which is what xarray is looking for when making its x coordinate for the plots you've shown above. In your case, you need a work simple around where we rename the 1-D coordinate variable the same name as the dimension:

for time in [0.1, 0.2]:
    ds.sel(time=time)['velocity'].rename({'r': 'radius'}).plot(label=time)

plt.legend()
plt.title('example for SO')

edited Dec 18, 2017 at 18:24

answered Aug 4, 2017 at 0:21

jhamman

6,5441 gold badge26 silver badges44 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

smillerc Over a year ago

Is there a better way to structure my Dataset to avoid this? This seems redundant...

jhamman Over a year ago

@smillerc - Not really. As was mentioned in the github issue you posted, we can probably make that happen in the xarray plotting code but my answer seems like the best approach for the current release.

Collectives™ on Stack Overflow

Plotting xarray datasets with variable coordinates

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related