Mini-Grid Nightlights

Code for “A Light in the Dark: Detectability of Mini-Grid Electrification with Remote Sensing in Sub-Saharan Africa.”
This repository contains the Python and R code used to extract geospatial time series, run staggered Difference-in-Differences (DiD) analyses, and generate figures for the paper and supplementary material.

Repository Structure (in numerical order)

Python / Jupyter Notebooks

• 01_GetCBILSites.ipynb – Clean and compile CBIL mini-grid site information.
• 02_GetNightlightValues.ipynb – Extract VIIRS DNB nighttime lights for CBIL sites.
• 03_InitialNightlightsAnalysis.ipynb – Exploratory analysis of brightness patterns.
• 06_GetCLUBNightlightValues.ipynb – Nightlights extraction for CLUB-ER sites.
• 09_SLGetDarkAreas.ipynb – Sample dark comparison sites for Sierra Leone.
• 11_AfricaGetDarkAreas.ipynb – Sample dark comparison sites across Africa.
• 13_GetSLPopulationValues.ipynb – Extract gridded population for Sierra Leone sites.
• 15_GetCLUBERPopulationValues.ipynb – Population extraction for CLUB-ER sites.
• 17_GetCBILPopulationValues.ipynb – Population extraction for CBIL sites.
• 19_OEMapsValidation.ipynb – Validate remote-sensing electricity access maps.
• 20_MGDistance.ipynb – Compute site distances and spatial metrics.
• 21_GetUsageInfo.ipynb – Usage and revenue data compilation.
• 22_UsageCorrelations.ipynb – Correlations between nightlights and usage.
• 23_MinusOceanDiD.ipynb – Background-ocean-adjusted DiD prep.
• 24_MovingAvgDiD.ipynb – Moving-average preprocessing for robustness checks.
• 25_GetBackgroundNightlights.ipynb – Global ocean background signal extraction.

R / R Markdown (DiD and Statistical Analysis)

• 05_StaggeredDinDSL.Rmd – Staggered DiD for Sierra Leone.
• 07_StaggeredDinDcluber.Rmd – Staggered DiD for CLUB-ER sites.
• 08_StaggeredDinDSLv2.Rmd – Revised Sierra Leone DiD model.
• 10_SLDarkAreaDinD.Rmd – Sierra Leone DiD using dark-area controls.
• 12_AfricaDarkAreaDinD.Rmd – Africa-wide dark-area DiD.
• 14_PopulationSLDinD.Rmd – DiD on population for Sierra Leone.
• 16_PopulationCLUBERDinD.Rmd – Population DiD for CLUB-ER sites.
• 18_PopulationCBILDinD.Rmd – Population DiD for CBIL sites.
• 26_BackgroundNightlightsAnalysis.Rmd – Analysis of global background signal.
• 27_ParallelTrendsTesting.Rmd – Parallel trends assumption tests.
• 28_HREAValidation.Rmd – Validation using external energy access products.

Getting Started

1. Clone the repository

git clone https://github.com/geohai/mini-grid-nightlights.git
cd mini-grid-nightlights

2. Python environment

Create the Conda environment:

conda env create -f environment.yml
conda activate CB_Lab_Mapping_Economic_Migration-env

3. R environment

Install required R packages (see headers of each .Rmd):

install.packages(c("tidyverse", "sf", "did", "ggplot2", "rmarkdown"))

Configuration

Copy and edit the sample configs:

cp user_config_sample.yml user_config.yml
cp .env.example .env

Add your local paths and any credentials required for data access.

Data Availability

• Public datasets (VIIRS, population grids, CLUB-ER) can be downloaded via the provided notebooks.
• Operational mini-grid data (PowerGen, CBIL) cannot be redistributed due to confidentiality.
• Scripts are provided so that users with legitimate access can reproduce the analysis.