Notes from 4/17 Millersville, PA Workshop

[jrleeman]

A place to collect notes for ourselves about improving the workshop materials.

Checkoff as we break these out into new issues - this is just a braindump!

Day 1

• Introduce website and write URL up in classroom everyday.

Git

• Add git (Windows) install notes
• Explain what a repo is before we create one - and what should be a repo
• Have editors with good syntax highlighting or no highlighting
• Use same editor as is available - probably not vi
• Add promise of making graphics before the end of the day to encourage getting through the hard start with git (start with setting up Python?)
• Add git pretty log
• bug calc.py initially so you can debug it later
• Maybe use plain text initially since they don't know python yet?
• Maybe explain workflow with graphics first?
• Magazine rack analogy

Conda

• Remove URL version of channel from environment.yml, just use conda-forge
• Make sure your local unidata-workshop environment doesn't exist before starting to teach!

Notebooks

• Rewrite our notebook intros to be more presentation quality
• Can then include metpy/siphon as examples
• Tab completion

Primer

• Pre-clear notebook outputs
• Question folks as to what the output should be before running it
• Maybe do this all in interactive shell?
• Point out that MATLAB is 1-based (and maybe why zero based is a thing?)
• Add units to speed example
• Pressure typo in exercises
• Use MetPy constants
• Talk about default values
• Kelvin issue in example
• Add tests for examples
• Add np.ndim
• Need a break before getting deep into time series stuff
• Add boolean arrays back
• Ran long

Timeseries

• Clarify subsetting exercise
• Missing links and resources
• About 40 minutes over (Ryan: WORSE.)
• Temperature_height_above_ground is all NaN

Upper Air

• Use from metpy.io import get_upper_air_data instead
• Clarify: use MetPy to download data
• Might want to solve gotcha:

get_upper_air_data(datetime(2017, 4, 5, 21), 'OUN')

to

my_date = datetime(2017, 4, 5, 21)
get_upper_air_data(my_date, 'OUN')

• Handle cases where there is no LFC
• Confusion alternating between filled-out code/plots and ones learners write themselves
• Add from metpy.units import units to last cell
• Was Hodograph really masking?

Day 2

• Put instructions for starting up again on the board
• Asked to post our completed example notebooks

Surface Data

• Weird failing for some lat/lon searches
• Pre-sort date data
• Ashville 35 C points that were unrealistic (good boolean filter)
• Why does filling the states result in funky polygons? (fills better with lakes) Just use lakes one to solve this.
• Why does the geo data look so bad from national earth? Rivers are especially bad.
• Request upper air data plots with dewpoint depression
• Plot of missing data currently shows an "X" which is obscuration. Should have an M

Satellite Data

• Maybe RGB example they can do?

Models

• Clean up some more oddities in end of example (no need to work with units)
• Rename Cartopy_mepty_hrrr to something better

Command Line

• People seemed engaged (this is good)
• Fix bugs?
  • Plot format
  • Set gfs level to int
• Try to simplify a bit more? Or at least have a simpler version with fewer bells and whistles to explain?

Other

• Test on 3.6 w/ Appveyor.
• Make basic numpy/matplotlib homework so we have more room for MetPy calc & units primer
• Maybe whack git so we have more time??
• Etherpad?
• Reformat
  • Have reference, filled notebook
  • Have separate exercise notebook, blank with only imports, markdown description cells, and comments in answer cells
  • Teach from blank notebook
• Add index notebook for workshop schedule

[jrleeman]

All broken up into new issues!