Datasets
The reference for each dataset: sources, methods, schemas, joins, caveats and citations. To compare selected module coverage and get files quickly, use the data catalog. Interpretation notes are included in each dataset reference.
DDL Global Urban Boundaries
City polygons with stable identifiers, published globally.
DDL Global Urban Boundaries ADM1 Splits
Administrative level-1 splits of parent city polygons.
DDL Global Urban Boundaries ADM2 Splits
Administrative level-2 splits of parent city polygons.
Historical Population
Population totals derived from GHSL population raster data, by year.
Urban PM2.5, monthly
Monthly mean surface PM2.5 concentrations derived from ACAG gridded data.
Urban Congestion Indices
Speed and congestion indices derived from Google Maps.
GHSL Built-up Zonal Statistics
Built-up surface, volume, non-residential use, and height derived from GHSL raster data.
World Bank Subnational Doing Business
Business regulation scores from the World Bank’s Subnational Doing Business surveys.
Our boundaries and datasets come at three different levels of granularity.
DDL Global Urban Boundary
Mega-cities are one unit.
e.g. Delhi + Gurgaon + Faridabad + etc. are all one unit
Unique identifier: ua_id
ADM1 splits
Mega-cities are divided at global ADM1 boundaries.
e.g. Delhi State, Delhi (Haryana), Delhi (UP) are each a separate unit
Unique identifier: ua_id + adm1_split_id
ADM2 splits
Mega-cities are divided at global ADM2 boundaries.
e.g. Each district of Delhi is a separate unit
Unique identifier: ua_id + adm2_split_id
Cities and administrative splits#
Each dataset is keyed to either a parent city boundary or to an administrative split of that boundary. ADM1 and ADM2 splits are derived by intersecting the parent city polygon with administrative boundaries from GADM.
ADM splits are only produced for cities where a meaningful partition exists. Specifically:
- An ADM1 split is included only when the city spans more than one GADM ADM1 unit. If the entire city falls within a single province or state, no ADM1 split entry is generated.
- An ADM2 split follows the same rule at the district or county level — the city must span at least two ADM2 units to produce a split.
As a result, small or compact cities may have a city boundary record but no corresponding ADM1 or ADM2 split records. This is expected, not a data gap.