Replication Code and Data for Socio-ecological Drought and Smallholder Vulnerability in the Central American Dry Corridor

This repository provides data and replication code for the article Socio-ecological drought and smallholder vulnerability in the Central American Dry Corridor. (Submitted April 2025, Accepted January 2026) 1_hls_download.ipynb is a python notebook that queries and downloads the sentinel-2 imagery from the Harmonized Landsat-Sentinel (HLS) dataset. This is used both for extracting mean NDVI values per land cover class and is used to generate a median dry and wet season image for the land cover classification. Downloaded sentinel-2 HLS images were loaded into google earth engine, combined with Landsat HLS imagery and topographically-corrected in the following script: https://code.earthengine.google.com/400eebf6e2e0e7d9d78181cdf63bae61. These topographically-corrected images were used to generate the land cover classification in this script: https://code.earthengine.google.com/ac589cf10d8ac6e8cc0398587b879bf0 Figure_3.R takes in precipitation and temperature data downloaded for Zacapa Department, Guatemala and calculates 3-month SPEI. It produces plots on monthly temperature and Standardized Precipitation Index (SPI), which are displayed in SI Figure 1. 3-month SPEI is displayed, with a trendline (LOESS, with span of 0.75), as a time series between 1979 and 2024. We also filter to growing season months (May-Nov) and display monthly values and a trendline. Figure_5_and_6.R produces the heatmap figures 4 & 5, inputing NDVI values for land cover classes across time series, split into 4 'seasons': wet early (June - Aug), wet late (Sept - Nov), dry early (Dec - Feb) and dry late (Mar - May). Finally, we provide code and data for replicating SI_Figure_B3 which produces graphs of nighttime light trends for the two villages assessed in the ethnographic analysis.