Overture Maps Divisions
Point, line, and polygon representations of countries, regions, counties, and neighborhoods worldwide.
About the data
Overture builds the divisions theme by conflating two community sources, OpenStreetMap as the primary input and geoBoundaries alongside it, then deduplicating features and applying topological overlap rules that depend on subtype: overlap is permitted at lower subtypes such as locality but not at country. No cross-source matching or attribute merging step is documented, so the theme is closer to a filtered union of its two inputs than a reconciled one, and it inherits OpenStreetMap's licensing as ODbL.
That method sets the quality floor. Coverage is global and generally aligns with admin_level tagging in the source datasets, which makes country, dependency, region, and county dependable, but macroregion and macrocounty are currently miscategorized as other subtypes, and the subtypes below county are often spotty even though they should exist everywhere.
What's included
Three feature types describe the same places in different geometries: a division is a point giving the location most commonly associated with a place, a division_area is the polygon or multipolygon for its land or territorial extent, and a division_boundary is the linestring where two divisions meet. Together they hold more than 5.8 million representations, each labelled with a subtype that runs from country, dependency, and region down through county, locality, borough, and neighborhood to microhood, so one row is one representation of one place.
Division areas carry is_land and is_territorial flags, which let you pick a coastline-clipped geometry or the full extent including water for the same place. Every feature is assigned a GERS ID, which is the key that joins divisions to the other Overture themes.
Why we offer this
Most researchers who want this are not studying boundaries, they are attaching something else to a place: storefronts, listings, job postings, sensor readings, each of which has to fall inside a polygon before it can be counted. Divisions gives you one subtype vocabulary and one set of geometries covering every country, so the same point-in-polygon step works whether the unit of analysis is a Japanese prefecture, a Danish locality, or a Brooklyn neighborhood, and cross-country comparisons do not depend on stitching together national statistical agencies.
It is worth knowing where that breaks down. Treat country through county as solid, and check coverage in your own study area before committing to a design that rests on neighborhoods or localities.
More free datasets
Other open datasets hosted on Dewey