Replication code for Schwarzwald et al., Large-scale stability and the Greater Horn of Africa long and short rains

This code replicates the analysis and figures in Schwarzwald et al. (in revision).

This code is to be used in conjunction with the replication data at doi.org/10.5281/zenodo.8092600.

To run:

1. Create a conda environment using the included environment_gha_stability.yaml file through conda env create -f environment_gha_stability.yaml or mamba env create -f environment_gha_stability.yaml (we recommend the latter for speed)
2. Update paths in dir_list.csv, to where you downloaded the replication data (set the path to climate_raw as raw, the path to climate_proc as proc, the path to aux_data as aux, and where you would like output figures as figs)
3. Step through complete_run.ipynb

Note that because of size constraints, only data needed to replicate main text figures were included in the replication data set. Some intermediate data (such as saturation specific humidity, or raw daily data for equatorial-wide maps such as Figures 10-11) have not been included, but can be regenerated using the replication data and the included code. Data needed for Supplementary Figures or for robustness checks against other reanalysis data products are (for the most part) not included, again primarily for space reasons. If you'd like to have access to the rest of these data, feel free to get in touch with the corresponding author (Kevin Schwarzwald, at kevin.schwarzwald@columbia.edu).

Troubleshooting

• import errors involving ESMF can usually be solved by explicitly specifying the latest xesmf version in installations

Notes on replication data

Replication data primarily consist of the MERRA2, CHIRPS, and OISST data needed to replicate the main text figures. Generally, these data have been pre-processed into a standard format, following (roughly) CMIP filename, variable, and dimension conventions. In some cases, data have been resampled to daily from higher temporal resolutions by taking the daily mean.