This directory contains data and Python scripts for reproducing results presented in the paper titled "Conditional pathways-based climate attribution" by Christopher R. Wentland, Michael Weylandt, Laura P. Swiler, and Diana L. Bull. The scripts are released under SCR#3080.
Global average impact (centered about counterfactual ensemble mean) time series data files are included in the data/ directory, and subdirectories with the format data-<region>-<period> for the spatial regions glob, nh, and na, and the time periods all, jja, and jfm. These are derived from the "CLDERA E3SM-SPA Simulation Ensembles" dataset, which are released under SCR#3051. These span a range of climate variables, Mt. Pinatubo stratospheric SO2 injection mass, and ensemble member numbers. The general format of each file follows
{variable}_{mass}Tg_ens{member_num}_avg_monthly.nc
These data files are formatted as E3SMv2 NetCDF files, and are processed here with the netcdf4 and xarray Python packages.
These scripts have only been tested with Python 3.8.19, and the package versions specified in requirements.txt. To guarantee version satisfaction, a fresh conda environment is recommended. For example,
conda config --add channels conda-forge
conda create --name multistep-attrib python=3.8 --file requirements.txt
conda activate multistep-attrib
Alternatively, one may attempt to simply satisfy package version requirements with pip via
pip3 install -r ./requirements.txt
There are five scripts for generating plots: plot_time_series.py, plot_fingerprinting.py, plot_regressions.py, plot_likelihood_dists.py, and plot_lr_pvals.py. Utility functions are held in utils.py, and constants.py contains global data and plotting parameters.
All scripts are assumed to be executed from this root directory, and generate images in a figs/ directory. They can be run by, for example,
python3 plot_lr_pvals.py