Featured Products

Source Cooperative is a Radiant Earth project

Walkthru-Indices · walkthru.earth · Source Cooperative | Source Cooperative

Global Population Projections (H3-indexed)

WorldPop SSP 1km population projections (2025–2100) aggregated to H3 hexagonal cells in optimized Parquet format. Eight H3 resolutions (1–8), one file per resolution, sorted by h3_index. v2 (recommended) uses BIGINT h3_index with 17 columns; v1 (legacy) uses VARCHAR h3_index with geometry/lat/lon/area_km2.


Source	WorldPop Global SSP Projections v0.2, 30 arc-second (~1 km), global coverage
Format	Apache Parquet (v2 optimized: BIGINT h3_index, no geometry columns)
CRS	EPSG:4326 (derivable via DuckDB h3 extension)
License	CC BY 4.0 by walkthru.earth
Code	walkthru-earth/walkthru-pop-index

Quick Start

Files

Compression: ZSTD level 3. Row groups: 1,000,000 rows.

Schema (v2, recommended)

Column	Type	Description
`h3_index`	BIGINT	H3 cell ID (integer representation)
`pop_2025`	FLOAT	Projected population count, 2025
`pop_2030`	FLOAT	Projected population count, 2030
`pop_2035`	FLOAT	Projected population count, 2035
`pop_2040`	FLOAT	Projected population count, 2040
`pop_2045`	FLOAT	Projected population count, 2045
`pop_2050`	FLOAT	Projected population count, 2050
`pop_2055`	FLOAT	Projected population count, 2055
`pop_2060`	FLOAT	Projected population count, 2060

17 columns total. Geometry, lat/lon, and area_km2 from v1 are derivable from h3_index via the DuckDB h3 extension (e.g., h3_cell_to_lat(h3_index), h3_cell_to_lng(h3_index)).

Sample values (res 5, most populated cells, SSP2 2025):

h3_index	pop_2025	pop_2050	pop_2100
599686542969856000	9,834,719	10,512,443	8,914,326
599262598287056896	7,126,541	8,021,334	7,432,105

How It Works

WorldPop 1 km rasters are read in 5° × 5° geographic windows
Each pixel is assigned to an H3 cell via h3.latlng_to_cell()
Population counts are summed per H3 cell (correct for count data)
Overlapping window-boundary cells are deduplicated via GROUP BY h3_index, SUM()
Final Parquet is sorted by h3_index (v2 stores BIGINT h3_index only; v1 legacy includes native GEOMETRY via DuckDB spatial)

All resolutions produce consistent world totals (~8.19 billion for 2025).

More Examples

Geometry / Coordinate Derivation

v2 files do not include geometry, lat, lon, or area_km2 columns — these are derivable from the BIGINT h3_index using the DuckDB h3 community extension:

v1 (legacy) files retain the original schema with VARCHAR h3_index, native Parquet 2.11+ GEOMETRY, lat, lon, and area_km2 columns. v1 is located at indices/population/v1/scenario=SSP2/h3_res={1-8}/data.parquet.

Source

WorldPop Global SSP Projections v0.2 — 30 arc-second (~1 km) gridded population projections for 2025–2100 under five Shared Socioeconomic Pathways (SSP1–SSP5). Float32, LZW-compressed GeoTIFF, EPSG:4326, coverage 60°S–84°N.

WorldPop. (2018). Global 1km-grid population projections, v0.2. University of Southampton. doi:10.5258/SOTON/WP00849

License

This dataset is licensed under CC BY 4.0 by walkthru.earth. The source WorldPop SSP Projections is licensed under CC BY 4.0 by the University of Southampton.