search

Unidata / python-gallery

Repository of examples showing off the use of Python for meteorology.
http://unidata.github.io/python-gallery
88 stars 31 forks source link

Meteogram Example - Move to Canned Data #51

Closed jrleeman closed 4 years ago

jrleeman commented 7 years ago

Looks like the METAR data is giving us all NaN for pressure data and that's causing the y-limit setting to go sideways. Hazards of using "live" data I suppose. Maybe we should point the gallery at the Irma datasets? That would be interesting and consistent while still showing how to access data from THREDDS.

This looks to be a consequence of a regression in mpl https://github.com/matplotlib/matplotlib/issues/9361

dopplershift commented 7 years ago

Given the discussion that happened around the change for matplotlib (matplotlib/matplotlib#7460), I think we're on the hook for fixing this on our end.

jrleeman commented 7 years ago

Should we be manually setting the ylimits anyway?

dopplershift commented 7 years ago

Probably not. Matplotlib 2.0 improved default limits anyway (things like margins)

tacaswell commented 7 years ago

Might be worth adding a check to the Matplotlib code that ensures the auto-limits never blow up. What sort of data are you feeding in that has this issue?

dopplershift commented 7 years ago

@tacaswell In this case we have live data that (apparently) was completely missing. This lead us to send NaNs to set xlim. Before matplotlib would give us a useless panel, now we get an error--the more I think about this, I'm not sad at the new behavior.

jrleeman commented 5 years ago

Let's move to canned data (case study or IEM) for now as we don't have a reliable surface feed from Unidata currently.