Colombia Environmental and Socioeconomic Satellite Dataset (2016–2021)
Description
This dataset compiles a rich set of environmental, climatic, and socioeconomic indicators for Colombian municipalities spanning the years 2016–2021. The data combine multiple satellite-based and administrative sources, enabling spatially explicit analyses of environmental and economic dynamics across the country. The dataset includes the following main variables: PM2.5 concentration (log): Logarithm of the annual mean concentration (µg/m³) of particulate matter smaller than 2.5 micrometers, derived from global atmospheric reanalysis models as described by Morcrette et al. (2009) and Benedetti et al. (2009). SO₄ concentration (log): Logarithm of the annual mean concentration (µg/m³) of sulfate aerosols, calculated from the NASA Global Modeling and Assimilation Office (GMAO, 2015) datasets. Total value added (log): Logarithm of the total municipal value added, obtained from the Colombian National Administrative Department of Statistics (DANE, 2025). Sum of Nighttime Lights (log): Logarithm of the annual sum of VIIRS Version 2 Nighttime Lights (NTL), used as a proxy for economic activity and urban development (Elvidge et al., 2021). Mean of Nighttime Lights (log): Logarithm of the annual mean VIIRS V.2 NTL, reflecting average light intensity within each municipality (Elvidge et al., 2021). Precipitation (log): Logarithm of total annual precipitation, obtained from the Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS; Funk et al., 2015). Temperature (log): Logarithm of the annual mean land surface temperature, derived from the MODIS Land Surface Temperature and Emissivity dataset (Wan et al., 2021). Dense vegetation area (log): Logarithm of the proportion of municipal area covered by dense vegetation, estimated using the Normalized Difference Vegetation Index (NDVI) following the approach of Sulla-Menashe and Friedl (2022). Burned area (log): Logarithm of the total municipal area affected by fire during each year, based on MODIS Terra Thermal Anomalies and Fire products (Giglio et al., 2021). Altitude (log): Logarithm of mean municipal elevation (meters above sea level), obtained from topographic datasets compiled by Universidad de los Andes. Population density (log): Logarithm of population density (inhabitants per km²), also sourced from Universidad de los Andes. Education expenditures (log): Logarithm of total central government transfers for education at the municipal level, based on administrative records compiled by Universidad de los Andes.
Files
Steps to reproduce
The dataset was constructed by integrating satellite imagery and administrative records through a combination of spatial processing and data aggregation techniques. Satellite-derived variables were obtained and processed in Google Earth Engine (GEE) using FAO’s geospatial feature collection for Colombia as the reference spatial unit. The analysis employed multi-temporal image collections covering the period 2016–2021, which were filtered, masked, and aggregated to generate annual composites representative of environmental and climatic conditions at the municipal level. The nighttime lights (NTL) data, derived from the Visible Infrared Imaging Radiometer Suite (VIIRS) Version 2 time series, were used as a proxy for local economic activity. These data have been validated in the Colombian context as reliable indicators of spatial and temporal variations in economic performance, particularly in areas with limited or inconsistent administrative records. Both the sum and mean NTL measures were computed to capture different aspects of luminosity intensity and distribution, and their logarithmic transformations were included to reduce skewness and facilitate econometric interpretation. In addition to NTL, the dataset includes other satellite-based variables such as precipitation, temperature, dense vegetation area, and burned area, derived respectively from the CHIRPS, MODIS, and related global datasets. Each raster product was spatially averaged over municipal boundaries to ensure consistency across years and variables. These remote sensing variables were then merged with socioeconomic indicators —such as total value added, population density, and education expenditures— compiled from the National Administrative Department of Statistics (DANE) and the Universidad de los Andes. This dataset was specifically developed to estimate the Environmental Kuznets Curve (EKC) for Colombia using a Bayesian Model Averaging (BMA) framework. The BMA approach allows for robust inference under model uncertainty by averaging across multiple model specifications, thereby identifying the most relevant determinants of environmental outcomes while accounting for spatial dependence and multicollinearity among predictors.
Institutions
- Nagoya Daigaku