Replication materials for "Quantifying the contributions of climate change and adaptation to mortality from unprecedented extreme heat events"

This repository provides code and data for "Quantifying the contributions of climate change and adaptation to mortality from unprecedented extreme heat events," currently in review. 

## Organization

The repository is organized into Scripts/, Figures/, and Data/ folders.

- Data: This folder includes intermediate and processed summary data that enable replication of most the figures and numbers cited in the text. The raw mortality data and E-OBS/ERA5 observations are quite large so they are not provided here but are publicly accessible. 

- Scripts: Code required to reproduce the findings of our work is included in this folder. Scripts are written in Python/Jupyter and R. The "France_counterfactual_HW_analysis" subfolder within this folder contains the code necessary to reproduce the machine learning training and predictions, adapted from Trok et al. (2024), https://www.science.org/doi/10.1126/sciadv.adl3242. There is a separate README within this folder which contains more specific details for reproducing this component of the analysis should you wish to do so. 

- Figures: This is where figures will be saved if you run the scripts.

## Data

Much of the intermediate data required to reproduce the final figures and numbers in the paper are provided in the various folders within the Data directory, including the overall panel dataset that includes departement-level mortality and temperature. However, much of the initial/raw data is too large to be hosted here. They are all publicly available at various locations:

- The E-OBS station-based observations are available here: https://surfobs.climate.copernicus.eu/dataaccess/access_eobs.php#datafiles

- Daily France mortality data is available here: https://www.insee.fr/fr/information/4769950

## Scripts

Each script performs one component of the analysis. Many of the initial scripts are wrapped into the France_Heat_Mortality.Rproj R project, so check the .Rprofile file for details on the required packages and file paths that are then loaded into each R script.

- France_Mortality_Regressions_R1.R performs all the heat-mortality regressions.

- France_Mortality_Predictions_R1.R performs the excess deaths calculations and the predictions of heat-related mortality from the 2003 heat wave.

- France_Mortality_Attribution_R1.R calculates counterfactual mortality at various levels of global mean temperature using the same heat-mortality relationships.

- Plot_Main_Figures_PNASR1.ipynb plots the main text figures and Plot_SM_Figures_PNASR1.ipynb plots the supplementary figures.

- Process_CNN_Deltas.R does some preprocessing on the machine learning predictions of the event just to make the final analysis more streamlined.

- Process_France_Mortality.R creates the initial panel dataset using the E-OBS data and France mortality observations. You won't be able to run this unless you download those data from the public archives.