W2D4 - tracking changes

Created on - 
Last updated on - 
Related project - @Link to project doc



Change Tracker Table

File
Decision
Comments
Responsible
Accountable
Consulted
Informed
Changed on
all
time series instead of timeseries
In line with other tutorials, e.g. W1D2
Jan 23, 2024
add estimated timing
cf. table  below 
Feb 14, 2024
TODO? random seed
 
TODO? unit convention for plot labels, so far e.g. (mm/day)
not sure if square brackets are the best solution , cf. Resource 1 
 
Drop down widget for definitions
TODO implement
 
Drop down markdown for plot descriptions
keep it short but with focus on interpretation
Jan 23, 2024
update mplstyle link
 
TODO replace Extremes & Vulnerability with Extremes & Variability in subheadings, artwork and
 
 
CouldDO Different colors for dropdown block titles depending on the topic
orange: Plot description,
green: Explanations
 
intro
Day Learning Objectives section
Jan 23, 2024
update sponsors
NFDI4Earth, CMIP ,...
 
tutorial 3
TODO dropdown detailed explanation of GEV
TODO T3 good as is, later necessary?
 
kept 'shape' parameter exercise in Coding Ex 1 and rephrased it acc to subsequent changes
Jan 26, 2024
create Widget for 'location' and 'scale' parameter instead of exercise in Coding Ex 1
TODO ask for review  Resource 1 
Jan 26, 2024
 
smaller semantic changes
Jan 26, 2024
tutorial 4
 
 
removed seaborn package
not used
 
tutorial 5
shorten plotting scripts
 
COULDDO formatted output instead of round()
f string instead of round(4)
 
tutorial 6
SSPXXX and hist scenarios, refer to W2D1 Intro Helene Witts  talk  21:30
 
removed pip install header as cartopy is not used
 
check whether Pangeo download is done in any other tutorials
 
added SHA256 of .nc file to pooch_retrieve
TODO discuss if this is too 'hardcoded'
 
renumbered sections and corresponding Q and C exercises, due to section 2 existing twice
 
updated resample(time="1Y") to resample(time="1YE")
undone, due to pandas conflicts see below
 
removed one of too many SSP-126 histograms in exercise and solution
 
update get_CMIP6 hyperlink
 
github, model, paper link updates in Resources
 
loop in exercise to allow for standard design, i.e.
data_src = ...
 
 
get_CMIP6
reviewed
 
removed intake and plotting libs
 
tutorial 7
removed SDFC library
 
added grid to plots
 
xlabel to ylabel in plot 1
reverse before
 
% formatting updated to f-string
state of the art in 3.X python
 
removed x and ylimits in QQ-plot
to show that black line is identity line
 
dropdown Plot descriptions for all 3 subplots
 
Latex formulae where adequate
 
added x- and ylabels in exercises
 
updated titles in exercises
 
replaced effective return level plotting by simple for loop
to trim down
 
added loop that break when non trivial solution was found
 
Tutorial 8
resample(time="1YE")
throws error due to old pandas version
 
in block comments
 
dropdown plot descriptions
 
TODO remove second os package import
 
 

Temporal summary

Name
Video length in min
Expected time to complete (including video)
Tutorial 1
8
30
Tutorial 2
9
30
Tutorial 3
7
30
Tutorial 4
7
30
Tutorial 5
8
25
Tutorial 6
5
35
Tutorial 7
11
Bonus
Tutorial 8
16
Bonus
Sum
71
180 + Bonus


Additional Review Notes

For future major updates
  • Speaker mispronounces GEV: GEF instead of GEV
  • T1 and T2 have sound issues after editing, low volume parts are cut-off
  • No estimated times in Tutorial headings
  • Where is Tutorial 7 video?? T7 has video T8, T8 has T9 video
  • T7 title contains 'EVT', however this does not appear on slides or within the tutorial itself
  • Future versions of T5 and T6 will need resample(time='Y') update, i.e. resample(time='YE') as of now pyleoclim forces us to use pandas 1.4, see  https://pandas.pydata.org/docs/user_guide/timeseries.html#anchored-offsets 


Survey Questions

Our Day Learning Objectives are:
    .1 Understand the relevance of the Generalized Extreme Value distribution to extreme events, apply this distribution to observation and model data, and assess the fit. [t3,t4,t5,t6,t7]
    .2Explain how the moments and parameters of the Generalized Extreme Value distribution vary with time. [t1,t3,t5,t7]
    .3Compute extreme event probabilities (return periods/levels). [t1,t2,t4,t6,t7,t8]
    .4Characterize extreme events (e.g. precipitation, sea level height, and heat) by these probabilities and prescribed thresholds. [t2,t5,t7,t8]

Corresponding Survey Questions can be found  here .


Backup Code snippets to discuss

For Tutorial 7 & 8 to measure the amount of time it takes to find non-trivial fitting parameters.
TODO: tune for alpha in QuantileRegressor() in sdfc_classes.py
# fit the GEV to the data, while specifying that the location parameter ('loc') is meant to be a covariate ('_c') of the time axis (data.index)
cnt = 0
import time
n_sample = 100
t_sample = np.zeros(n_sample)
for j in range(n_sample):
t = time.process_time()
for i in range(250):
law_ns.fit(data.ssh.values, c_loc=np.arange(data.index.size))
#print(law_ns.coef_)
if law_ns.coef_[0] != 0:
cnt+=1
nonzero = law_ns.coef_
break
elapsed_time = time.process_time() - t
t_sample[j] = elapsed_time
print(f'This took: {elapsed_time:.2f} seconds.')
print(cnt)
print(nonzero)
sns.histplot(t_sample)

Need some help getting started?  Read our guide to RACI.