08.08.2022 @Soufiane Karmouche (SK), Institute of Environmental Physics (IUP), University of Bremen

• This repository presents a sample of the regime-oriented causal model evaluation (CME) in the form of jupyter notebooks. The main set of code is used to produce most of the figures on the paper "karmouche23esd_CausalModelEvaluation_Modes"

Karmouche, S., Galytska, E., Runge, J., Meehl, G. A., Phillips, A. S., Weigel, K., and Eyring, V.: Regime-oriented causal model evaluation of Atlantic–Pacific teleconnections in CMIP6, Earth Syst. Dynam., 14, 309–344, https://doi.org/10.5194/esd-14-309-2023, 2023.

Corresponding DOI:

I. INSTALLATION

1. Clone the repository:

git clone https://github.com/EyringMLClimateGroup/karmouche23esd_CausalModelEvaluation_Modes

2. Create conda environment from environment.yml:

cd karmouche23esd_CausalModelEvaluation_Modes

conda env create -f environment.yml

conda activate causalenv

cd ..

3. Install the TIGRAMITE package to use PCMCI+: please follow instructions: here

git clone https://github.com/jakobrunge/tigramite.git

cd tigramite   

python setup.py install

cd ..

It is the User's responsibility to install TIGRAMITE. Results of the paper have been produced using version: 5.0.1.17

After installing Tigramite in the conda environment, add environment to jupyter kernels:

python -m ipykernel install --user --name=causalenv   

---

II. DOWNLOADING DATA

The complete CVDP-LE diagnostic for the 1900-2014 historical CMIP6 comparison run can be found on: CESM CVCWG CVDP-LE Data Repository

- Alternatively through command line:

1. Download:

wget http://webext.cgd.ucar.edu/Multi-Case/CVDP-LE_repository/CMIP6_Historical_1900-2014/CMIP6_Historical_1900-2014.cvdp_data.tar

2. Extract:

tar -xvf CMIP6_Historical_1900-2014.cvdp_data.tar

After installing TIGRAMITE and downloading the data, it is time to slightly edit the jupyter notebooks before running them.

---

III. EDITTING JUPYTER NOTEBOOKS

OBS_CMIP6_pcmciplus.ipynb:

1. In the first code cell (Imports): Remove the hashtag (#) at the beginning of last four lines in the Imports cell (e.g. from tigramite...)
2. Change the path_to_data in Data PATH cell to where CMIP6_Historical_1900-2014.cvdp_data is saved.

F1_score.ipynb and Ensemble_graphs.ipynb:

1. In the first code cell (Imports): Remove the hashtag (#) at the beginning of last line in the Imports cell (e.g. from tigramite import plotting as tp)

Pattern_correlation.ipynb:

1. Change the path_to_data in Data PATH cell to where CMIP6_Historical_1900-2014.cvdp_data is saved.

---

IV. RUNNING JUPYTER NOTEBOOKS

First create a directory where results will be saved:

mkdir Results_DIR/

We recommend running the notebooks in the order below:
(some notebooks require results from other notebooks)

         1. Pattern_correlation.ipynb

         2. OBS_CMIP6_pcmciplus.ipynb

         3. F1_score.ipynb

         4. Ensemble_graphs.ipynb

---

V. RESULTS

Results from running the jupyter notebooks will be saved under the Results_DIR/ directory

          To refer to figures from the paper, please see overview_figures to locate the notebook needed to produce a specific plot.

---

REFERENCES:

Phillips, A. S., C. Deser, J Fasullo, D. P. Schneider and I. R. Simpson, 2020: Assessing Climate Variability and Change in Model Large Ensembles: A User’s Guide to the “Climate Variability Diagnostics Package for Large Ensembles”, doi:10.5065/h7c7-f961

PCMCI+: J. Runge (2020): Discovering contemporaneous and lagged causal relations in autocorrelated nonlinear time series datasets. Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence, UAI 2020,Toronto, Canada, 2019, AUAI Press, 2020.