Improve & add worksheet questions

[hsteptoe]

Most worksheets need the questions block reviewing either to make the questions clearer (some are very vague) or to add questions to break up otherwise large chucks of code.

This should be fairly quick and can be done when the notebooks get to the review stage, so I've put his against the v1.0 milestone.

At this point it would also be useful to:

• Generally proof read the worksheet instructions
• Check flow of worksheet exercises (but bear in mind that no major content changes will happen whilst we're still working towards v1)
• Check learning aims are consistent with worksheet content

Other comments are welcome but might not be actioned until we have completed the v1 milestone.

[rosannaamato]

Worksheet 1:
Removed reference to ! as it's less powerful than !, as req'd to run code.
Note that the data doesn't live in the same directory as the worksheets, so from 'cd' command onwards, code won't run. Point to an explicit path.
Assumes we know UM date stamps... not ideal.

[rosannaamato]

S1.2. Data points to air temperature data, it may be more effective to use a precipitation cube, as you will be able to clearly able to see the rim when plotting, to better understand why the rim-removal stage (S1.3) i so crucial.

[rosannaamato]

Editing in fix/structure: ff79083 (commit at 16:22).

[rosannaamato]

Rim Removal code block could benefit from some output to confirm that it's completed. Perhaps a print or ls statement?
(Also, write permissions will need changing before Cathryn tests the code's functionality).

Added a question to review (basic sense check of) the final pre-processed files, to confirm what the code blocks did.

Updates: fix/structure: eddc555
Review of worksheet1 complete.

[rosannaamato]

Worksheet 1:

• Labelled instructions alphabetically to match worksheet 2 (clearer than bullet points).
• Is it necessary to include processing of cahpb at the end of Worksheet 1? We've only worked with capha up to this point, so it may cause confusion to add another runid without explanation. If the processed data already exists for use in worksheet 2 (and beyond) then I suggest removing cahpb from worksheet 1. Then the two runids can be properly introduced at the start of worksheet 2, where they are both used.

Worksheet 2a:

• Added and improved link naming (APRODITE: http://www.chikyu.ac.jp/precip/english/index.html, IRIS documentation.)

• Added a graphic to explain the difference between capha and cahpb.

• Removed reference to OND from Section 2.2 (unit conversion) as it's not calculated until S2.3.

• S2.3: is there any reason for creating two seasons (OND, JFMAMJJAS), instead of the standard 4 seasons? Might be more intuitive to make the 4.

• What does the '.a.' stand for in the file naming? e.g.
  outfile = os.path.join(outdir, 'capha.a.OND.mean.baseline.pr.mmday-1.nc')
  Could it be removed?

• Added 'aphro' to each variable name in the code used to make APHRODITE seasonal means, so it won't accidentally pick up variables from the cahp[a,b] loops.

• Section 2.4: when plotting, why does the APHRODITE colour bar not display it's units?

IMO, This feels like a natural break point to finish worksheet 2a.
Continue activities 2.5 etc. in another worksheet.

• Section 2.5 (domain extraction):

• added a question to order Kuala Lumpur lat-lon coords, (another reminder that it would be great if we could figure out how to hide & reveal answers in a Jupyter notebook!)
• Switched APHRODITE & CAPH[A,B] code blocks, so it's easier to explain the difference between reg lat-lon and rotated coords. (Could do with an image of where the rotated pole lies for this simulation...?)
• I imagine we're going to get questions about how to calculate grid lat-lon coords for extracting data from a rotated grid (and rightly so) - can we add a step to explain how this is done at some point? (i.e. extract the rotated pole coords from the cube, use to adjust reg lat-lon coords).

• Section 2.5 split into 2.5 (domain extraction) and 2.6 (multi-annual monthly means)
  Latter sections are now numbered +1.

2.7 (comparing models and obs):
Why can Iris' subtraction not be used here? Is there an instance when it can be used? Perhaps point to some documentation, or consider removing reference to it, if it causes added confusion.

[rosannaamato]

2.7
When creating difference cubes, should we also rename the cube to acknowledge that the data has been subtracted (cahpa-aphro), especially as the files are being saved?
cube.rename??

[rosannaamato]

Saving 2D data:
At an earlier stage in the worksheets, values were time averaged, creating a cube with just 1 time coordinate (plus lat, lon). Subsequently, we continue to reference e.g. cahpa_obs_diff[0] with the wordy comment:

Note this is where cube slicing is needed as you can only plot 2-coordinate dimensions with qplt.contourf, so here we have selected time[0] as there is only one timestep (the baseline 1961-1990 mean)

Perhaps it would be neater to save the original code as cube[0], to save this repeated explanation and make the code more intuitive?

[rosannaamato]

Review of worksheet 2a complete, pushed: 2775277

Suggest splitting worksheet 2a into worksheets 2 and 3, which would require:

• a worksheet summary after section 2.4
• new intro and objectives at the beginning of a new 'worksheet 3'
• explicitly specify that the final exercise of worksheet 3 is optional (an extension exercise)
• rename section and task labels from 3.1 (a) in worksheet 3
• rename worksheets 2b and 3 to worksheets 4 and 5.

[rosannaamato]

Worksheet 4 review (old 2b)

• Preamble - suggest removing comment # /data/users/ssadri/pyprecis/monthly
• This worksheet does the job of creating useful plots but is very passive. It could be finished in 5 mins if people didn't read the code. I'll have a think about how to make it more engaging next week.
• Plotting precipitation anomalies at the end of worksheet 4 - need to label capha and cahpb, otherwise the plots are indistinguishable. What's the learning aim of this comparison? Perhaps they could be plotted on the same graph if appropriate? Also, better to ensure consistent labelling of the x axis (top plot starts at 2020, bottom plot at 2022).
• No need to reference APHRODIR or CRUDIR in this worksheet; 2b only compares baseline and future model data.
• • It'd be nice to organise the data directories a little more;
  • DATADIR = '/project/precis/worksheets/data'
CLIMDIR = os.path.join(DATADIR, 'climatology')
    These make sense as data locations. Therefore, could MODELDIR be pointed to something more intuitive - perhaps /project/precis/worksheets/data/future?
  • MODELDIR currently points to /project/precis/worksheets/data/netcdf, which is full of pre-processed baseline data.
  • Also, CLIMDIR contains historic, future and fut-his processed data; these should be saved to a more sensible location in the processing scripts.
• labelled actions alphabetically for consistency with previous worksheets.
• To create more engagement with the worksheet, I have removed sections of code for the participants to fill in. This occurs in section c, where they simply need to copy code from the above block (precip) into the empty block (temperature).
• Suggest removing comment about being inefficient in 2.11 (f): # Apply land mask. Probably not most efficient method. baseline / future data arrays are changed to masked type. # Apply land mask is sufficient.
• Corrected some of the comments (copied and pasted without changing, I think): # Save the area averaged monthly differences -> Save future & baseline area averaged monthly data (time series).
• One method for making 2.11 (f) more interactive could be to ask which order the actions should be completed. Provide (jumbled version of ) this list of tasks:
  • load files
  • apply land mask
  • calculate mean values over land points
  • Save future & baseline area averaged monthly data (time series)
  • subtract baseline from future
  • save the area averaged monthly differences (time series)
    Ask the participants to decide the most sensible order for these tasks.
    Then get them to construct the code block to fit their order (provide discrete chunks of code for them to copy & paste in in the correct order).
• g): plot time series of pr anomalies for both models - can these be plotted on a common axis for easier comparison? These have limited value separately. We should then ask a question that requires the participants to interpret the plots.
• WS2b should be renamed WS4, after WS2a has been split and renamed.

[rosannaamato]

Plotting precipitation anomalies at the end of worksheet 4 - need to label capha and cahpb, otherwise the plots are indistinguishable. What's the learning aim of this comparison? Perhaps they could be plotted on the same graph if appropriate? Also, better to ensure consistent labelling of the x axis (top plot starts at 2020, bottom plot at 2022).