Global Population Grids: Summary Characteristics
|
Dataset |
Source |
Concept |
Method |
Grid Cell Size |
Year(s) Represented |
Source for National Level Population Totals |
Distribution Policy |
|
Unmodeled Population Grids |
|||||||
|
Center for International Earth Science Information Network (CIESIN) - Columbia University |
Resident** population (population counted at place of domicile) |
Areal weighted such that population is spread across all grid cells covering a census unit based on the proportion of the grid cell that falls in the census unit. See table of input layers. Details. |
30 arc-seconds (1 km) |
2000, 2005, 2010, 2015, 2020 |
Two versions: 1) official country totals from census; 2) Country totals adjusted to United Nations Population Division (UNPD) estimates and projections |
Open access |
|
|
Lightly Modeled Population Grids |
|||||||
|
European Commission Joint Research Centre (JRC) and Center for International Earth Science Information Network (CIESIN) - Columbia University |
Resident** population (population counted at place of domicile) |
Dasymetric* using Sentinel-2 and Landsat-derived built up volume and function, and proportional allocation to distribute population data from subnational census data to built-up areas. See table of input layers. Details. |
100 m, 1 km in World Mollweide (EPSG:54009); 3 arc-seconds (~100 m), 30 arc-seconds (~1 km), WGS84 |
1975-2030 in 5-year steps |
United Nations Population Division (UNPD) estimates and projections |
Open access |
|
|
Center for International Earth Science Information Network (CIESIN) - Columbia University; International Food Policy Research Institute (IFPRI), The World Bank, Centro Internacional de Agricultural Tropical (CIAT) |
Resident** population (population counted at place of domicile) |
Binary dasymetric* using DMSP-OLS nighttime lights and proportional allocation to distribute population data from subnational census data to the settlement extents. See table of input layers. Details. |
30 arc-seconds (1 km) |
1990, 1995, 2000 |
United Nations Population Division (UNPD) estimates and projections |
Open access |
|
|
Highly Modeled Population Grids |
|||||||
|
History Database of the Global Environment (HYDE) Population Grids, version 3.1 |
Netherlands Environmental Assessment Agency (PBL) |
Resident** population |
Historical population, cropland and pasture statistics are combined with satellite information and specific allocation algorithms (which change over time) to create spatially explicit dasymetric* population maps. Details. |
5 arc-minutes (10km) |
10,000 BC to 2000 AD (10,000 BC, 5,000 BC, 0 AD, 500 AD, 1000 AD, 1500 AD, 1600 AD, 1700 AD, 1800 AD, 1900 AD, 1950 AD, 2000 AD) |
Population estimates are generally on the high end of the range of past estimates |
Open access |
|
Oak Ridge National Laboratory (ORNL) |
Ambient (24 hour average) population based on enumeration counts |
Spatial data, imagery analysis, and multi-variable dasymetric* modeling are combined with geospatial science (remote sensing and machine learning technologies) and subject matter expertise to distribute unwarned populations. Cells are weighted and tailored to match the data conditions and geographical nature of each country/region by leveraging the most recent and highest resolution ancillary data available. See table of input layers. Details. |
30 arc-seconds (1 km) |
annual releases 2000 - 2022 |
Individual country census normalized to the CIA World Factbook (source: U.S. Census Bureau's International Programs Center, International Database) |
Open access |
|
|
ESRI |
Mixed (45% of countries are resident** populations), based on available national population estimates, though progressing toward resident** with each new release |
Dasymetric* mapping based on four inputs: BaseVue 30-m landcover for urban classes; 75-m grid of presence/absence of road intersections; 75-m geonames populated place points buffered by 5 cells; landsat8 15-m panchromatic rugosity with regional threshold for levels of landscape disturbance equating to human settlement. Details. See table of input layers. |
150 meter (2016), 250 meter (earlier) |
2013, 2015, and 2016, with 2017 planned for fall 2018. |
Country-official estimates with 134 countries processed further by Michael Bauer Research GmbH. |
Commercial / Free to ArcGIS Users |
|
|
WorldPop |
High spatial resolution, temporally-explicit data on human population and demographic distributions |
Top-down approach that relies on a statistically-based weighting layer combined with a dasymetric* redistribution. See table of input layers. Details. |
3-arc seconds (100 meter) |
2000-2020 globally and country-specific years |
Two versions: 1) Country-official estimates, and 2) United Nations Population Division (UNPD) estimates and projections |
Open access |
|
* Dasymetric mapping approaches rely on ancillary data to spatially disaggregate census counts from administrative / census units in an effort to develop higher resolution data products that more faithfully represent population distribution on the ground. The simplest approach is binary dasymetric mapping, which uses one other data layer (such as satellite-derived built-up areas or urban extents) to move populations from census units (which are sometimes large) to areas identified as settlements. Other techniques use a variety of ancillary data, including urban extents, land cover data, and slope, as well as spatial “masks” to exclude populations from protected areas or military reserves, to move populations using statistical weighting algorithms to inform final gridded outputs.
**Resident population is linked to the primary residence where people live, rather than where they work.