The reference is Overture Maps, which merges OpenStreetMap with machine-learning building footprints from Microsoft, Google and Esri. Each Overture feature records which sources it came from. Completeness asks: of everything Overture knows about in an area, how much is already in OpenStreetMap.
For each hexagon, buildings in OSM / total Overture buildings.
A high value means OpenStreetMap already has most of what exists. A low value
means buildings are present (from the machine-learning sources) that
OpenStreetMap is missing, a mapping gap.
The share of road segments in the hexagon that carry an OpenStreetMap source.
Caveat: Overture's road network is mostly OpenStreetMap, with only a small amount from TomTom and other sources. Because the reference is largely OpenStreetMap itself, road completeness reads high almost everywhere. A high value means the roads match Overture's reference, not that every road on the ground is mapped.
Population comes from the source selected in the panel (see Sources below).
The mapping gap weights people by how incomplete the buildings are:
population × (1 − building completeness). A cell with many people and
few buildings in OpenStreetMap scores high, that is where mapping helps most.
With WorldPop selected, a year slider covers 2015 to 2030, and the map, the population figure and the sparkline all follow the chosen year. Every year is a modelled estimate built from timepoints that differ by country, so read it as an estimate, not a count. Details and the per-country sources are in the release statement and the census sources. The slider moves population only; the mapping gap stays on its published year.
Features are counted in H3 resolution 8 hexagons (about 0.7 km² each). A dropped polygon is matched to the hexagons it covers.
Buildings and roads from Overture Maps (release ).
Population sources:
© OpenStreetMap contributors, Overture Maps Foundation.