time decoding problem with xarray and matplotlib contourf · python-questions

Stream: python-questions

Topic: time decoding problem with xarray and matplotlib contourf

Frank Bryan (Apr 16 2020 at 22:26):

I am running into a problem trying to plot a Hovmoller diagram of data form a Tropical Pacific mooring using contourf. Looking for suggestions for a work around. A simplifed version of the code follows:

ds_tao = xr.open_dataset('/glade/work/bryan/Observations/TAO/adcp0n140w_dy.cdf')
u = ds_tao.u_1205.squeeze().transpose()
u = u.where(u<1.0e35) # missing value=1.0e35

# This works fine and correctly decodes the dates
u.plot(robust=True).

# This creates a plot but does not decode the dates
ulev = np.arange(-100.,100.,5.0)
fig,ax = plt.subplots()
cf = ax.contourf(u,levels=np.arange(-100.,100.,5.0))

# This fails
fig,ax = plt.subplots()
cf = ax.contour(ds_tao.time,ds_tao.depth,u,levels=ulev)

The final part of the traceback is

/glade/work/bryan/miniconda3/envs/analysis/lib/python3.7/site-packages/matplotlib/dates.py in _from_ordinalf(x, tz)
    286                          'non-datetime values are passed to an axis that '
    287                          'expects datetime objects.'.format(ix))
--> 288     dt = datetime.datetime.fromordinal(ix).replace(tzinfo=UTC)
    289
    290     # Since the input date *x* float is unable to preserve microsecond

OverflowError: signed integer is greater than maximum

Gustavo M Marques (Apr 16 2020 at 23:46):

Try:

 cf = ax.contour(ds_tao.time.values,ds_tao.depth,u,levels=ulev)

Frank Bryan (Apr 17 2020 at 00:06):

Worked! Thanks Gustavo.

Anna-Lena Deppenmeier (Apr 17 2020 at 00:10):

Great!! That is way easier than my workaround.

Deepak Cherian (Apr 17 2020 at 16:27):

Alternatively ds_tao.u.plot.contourf(x="time", y="depth", levels=ulev)?

Frank Bryan (Apr 17 2020 at 19:09):

Also works. Thanks Deepak.

Last updated: Jan 30 2022 at 12:01 UTC