The most downloaded population dataset on Zenodo reveals three decades of migration, megacity growth, and the quiet emptying of the countryside — one square kilometer at a time.
In 1990, the world held 5.3 billion people. By 2022, that number had crossed 8 billion. But the raw count obscures the real story: not how many people there are, but where they went. GlobPOP traces that movement at 1km resolution across 33 consecutive years, revealing patterns invisible to national statistics. Cities that barely existed in 1990 — Shenzhen, Dhaka's satellite towns, the sprawl radiating from Lagos — now appear as dense population cores in the latest grids. Meanwhile, entire rural provinces in Eastern Europe and East Asia have faded to near-emptiness, their populations absorbed by distant urban centers.
The dataset's construction is a feat of statistical engineering. National censuses, conducted at different years and different granularities by 200+ countries, are harmonized and disaggregated using satellite-derived settlement layers and nighttime light imagery. The result is a globally consistent population surface that can be queried at any 1km cell for any year from 1990 to 2022. This consistency is what makes GlobPOP uniquely valuable: unlike snapshot products that capture a single year, it enables trend analysis. Researchers can ask not just "how many people live here" but "how has this place changed" — and get a 33-year answer.
With 273,000 downloads, GlobPOP has become foundational infrastructure for fields ranging from epidemiology to disaster response to climate adaptation. The World Bank uses it to estimate populations in areas between censuses. Humanitarian organizations use it to pre-position aid supplies before cyclone landfalls. Climate modelers use it to project future exposure to heat waves, flooding, and sea level rise. The dataset does not merely describe where people live — it has become the denominator in calculations that affect how resources, risks, and responses are allocated across the planet.
Billions of people in urban-classified vs rural grid cells, showing the urbanization crossover
| metric | this dataset | alternative | verdict |
|---|---|---|---|
| Temporal Coverage | 1990-2022 (annual) | Single year (WorldPop) | 33x more years |
| Spatial Resolution | ~1km (30 arc-sec) | ~1km (GPW v4) | Comparable |
| Update Frequency | Annual | Every 5 years (GPW) | 5x more frequent |
| Nighttime Light Integration | Yes (DMSP + VIIRS) | Not in all products | More accurate |
| Downloads | 273,000 | Varies | Most downloaded on Zenodo |
| Settlement Layer | Satellite-derived | Census boundaries (GPW) | Spatially explicit |
GlobPOP's temporal depth enables projection of future population exposure to climate hazards — showing not just where people are, but the trajectory of where they are heading.
For countries that conduct censuses only every 10 years, GlobPOP provides inter-censal population estimates at spatial resolutions finer than any national statistical office can produce alone.
Humanitarian organizations use gridded population data to estimate affected populations within hours of a disaster, enabling pre-positioning of aid and evacuation planning before events occur.
Share this story