leach-et-al-2021

Recreate examples in FaIR2.0.0 paper

Reproducibility

With two exceptions, Python/Jupyter is used for everything. Firstly, clone the repository, navigate to the leach-et-al-2021 directory and run

pip install -r REQUIREMENTS.txt

Run the setup.sh script to download the auxiliary data (3.0 GB) and set up the necessary output directories:

./setup.sh

One script uses the EBM library written in R. This tunes the FaIRv2.0.0 parameters to CMIP6 climate respose using Cummins et al., 2020 MLE methodology. Then everything else is produced in the Jupyter notebooks (notebooks directory).

R script

Open an R prompt and run

install.packages("devtools")
devtools::install_github("donaldcummins/EBM")

This is run simply by

cd r_scripts/parameter-tuning
Rscript CMIP6-climresp_Cummins.R

Notebooks

These notebooks are listed in the approximate order they should be run in to generate all the data files and parameter sets required to fully carry out all the analyses within the paper. Figures and tables as they appear in the paper are listed after each description.

order	/path/to/notebook	description
1	/notebooks/example-nb.ipynb	Demonstration of FaIRv2.0.0-alpha I/O formats, plus some examples of how to run experiments.
2	/notebooks/parameter-tuning/CMIP6-aerosol.ipynb	Tuning FaIRv2.0.0 parameters to CMIP6 aerosol forcing from RFMIP. Fig5, Tab4
3	/notebooks/parameter-tuning/CMIP6-cc.ipynb	Tuning FaIRv2.0.0 parameters to CMIP6 ESMs from C4MIP. Fig4, Tab3
4	/notebooks/parameter-tuning/CMIP6-climresp-results.ipynb	Examining tuned climate response parameters from CMIP6-climresp_Cummins. Fig3, Tab2, TabS5
5	/notebooks/parameter-tuning/default-experiment-metrics.ipynb	Carries out a suite of experiments for the default parameter set, and computes associated metrics. Fig2, FigS2, FigS3, TabS2, TabS3
6	/notebooks/perturbed-parameter-ensemble/FULL-compute.ipynb	Creates a large parameter ensemble (FULL) & generates key metrics for each member over the SSPs.
7	/notebooks/perturbed-parameter-ensemble/global-warming-index.ipynb	Generate an estimate of the present-day Global Warming Index, and associated FULL ensemble selection probabilities.
8	/notebooks/perturbed-parameter-ensemble/FULL-constrain.ipynb	Constrains the FULL ensemble using the GWI-based selection probabilities, and explores the implications. Fig6, Fig7, Tab7, Tab8, Tab9, Tab10, TabS6
9	/notebooks/parameter-tuning/default-parameters.ipynb	Tunes or sets the default model parameter set.
10	/notebooks/perturbed-parameter-ensemble/DECK-simulations.ipynb	Carries out some basic CMIP6 experiments (abrupt-4xCO2 & 1pctCO2) for the FULL and CONSTRAINED ensembles. Fig8
11	/notebooks/perturbed-parameter-ensemble/SSP-simulations.ipynb	Carries out simulations of the SSPs for the CONSTRAINED ensemble. Fig9, Fig10, TabS4
12	/notebooks/RCMIP/RCMIP-experiments.ipynb	Carries out simulations submitted to RCMIP phase II.

Auxiliary data files

There are also a number of auxiliary data files and directories within this repo. Some are excluded from the repo due to limitations on file sizes, but all are described below.

/path/to/file/or/directory	description
/aux/FaIRv2.0.0-alpha_RCMIP_forcmap.csv	Maps between RCMIP forcing categories (pyam convention) & FaIRv2.0.0-alpha categories (AR5 convention).
/aux/FaIRv2.0.0-alpha_RCMIP_inputmap.csv	Maps between RCMIP input variables (emissions + concentrations) & FaIRv2.0.0-alpha inputs.
/aux/FaIRv2.0.0-alpha_RF_categories.csv	Maps each FaIRv2.0.0-alpha output forcing (default parameters) to an aggregated forcing category.
/aux/parameter-sets/CMIP6_cc_full.csv	Carbon cycle parameters tuned to C4MIP models with $r_a$ included in the fit.
/aux/parameter-sets/CMIP6_cc_reduced.csv	Carbon cycle parameters tuned to C4MIP models with $r_a$ not included in the fit.
/aux/parameter-sets/CMIP6_climresp.csv	Climate response parameters tuned to CMIP6 models.
/aux/parameter-sets/CMIP6_ERFaci.csv	Aerosol-cloud interaction parameters tuned to RFMIP.
/aux/parameter-sets/CMIP6_ERFari.csv	Aerosol-radiation interaction parameters tuned to RFMIP.
/aux/parameter-sets/CMIP6_f_params.csv	CO2 forcing parameters tuned to CMIP6 models.
/aux/parameter-sets/Complete_gas_cycle_params.csv	Full default gas cycle and forcing parameter set.
/aux/parameter-sets/CMIP6-climresp_Cummins/	Tuned EBM parameters to CMIP6 models following Cummins et al., 2020.
/aux/parameter-sets/perturbed-parameters/climresp_params/	FULL ensemble of climate response parameters in batches of 10,000 members.
/aux/parameter-sets/perturbed-parameters/extforc_sfs/	FULL ensemble of external forcing scaling factors in batches of 10,000 members.
/aux/parameter-sets/perturbed-parameters/gas_params/	FULL ensemble of gas cycle & forcing parameters in batches of 10,000 members.
/aux/parameter-sets/perturbed-parameters/gas_params/{FULL,ALT}_{ALL,ANT}.h5	Key output metrics for FULL (ALTERNATIVE) ensemble under all-forcing (anthropogenic-only) SSP2-45 simulation.
/aux/parameter-sets/perturbed-parameters/gas_params/ssp*.h5	Key output metrics for FULL ensemble under all-forcing SSP simulations.
/aux/parameter-sets/perturbed-parameters/gas_params/{FULL,ALT}_selection_probability*.h5	FULL (ALTERNATIVE) ensemble member selection probabilities based on anthropogenic-warming-index level/rate likelihood from various observational data products.
-------------	-------------
/aux/output-data/global-warming-index/	Outputs from the global-warming-index notebook.
/aux/output-data/RCMIP/	Model output submitted to RCMIP phase II.
-------------	-------------
/aux/input-data/Temperature-observations/	Global temperature data products used in global-warming-index notebook. Raw data as downloaded from corresponding sources.
/aux/input-data/RCMIP/	MAGICC7.1.0-beta and FaIRv1.5 data from RCMIP phase I used in several figures.
/aux/input-data/CMIP6/	CMIP6 data used throughout the paper. Downloaded from https://cmip6.science.unimelb.edu.au/. See Nicholls et al., 2021.