paper

Jupyter notebooks for generating figures used in Guillaumin (2021)

Ensure to read the readme in the folder above this one first.

Figure 1

generate-paper-figure-1.ipynb generates figure 1b. The forcings it uses can be generated by running the data step with the following configuration:

python src/gz21_ocean_momentum/cli/data.py \
--config-file resources/cli-configs/data-paper.yaml \
--ntimes 4000

Figure 6

generate-paper-figure-6.ipynb, which generates figure 6b, requires the above forcing data, plus another set of forcings generated using the 1% annual CO2 increase CM2.6 dataset. Use the same command as above, with --co2-increase.

Figures 4, 5, 7

test-global-fig-4-5-7.ipynb generates figures 4, 5 and 7, as well as D4 and D5. For this, the inference step with the trained neural network has to be run both on the data with and without --co2-increase, and then the notebook needs to be run once with each set. (The neural net may be trained only once, on data without --co2-increase.) The paper figures referring to piControl are those without --co2-increase (the control simulation with pre-industrial CO2 levels), and the figures referring to 1pctCO2 are those with --co2-increase (a 1% increase per year in CO2 levels for the first 70 years, after which they remain constant).