9.7.1 — Population Intelligence — maturity: live

High-Resolution Population Mapping

Combining very-high-resolution optical imagery with building footprint extraction and ancillary data to produce sub-100-metre gridded population estimates across a national territory.

Auto-drafted reference page — content reviewed for shape, individual references not yet link-checked.

Every ministry that allocates schools, clinics, roads or disaster-response assets needs to know where people actually live — not where a census said they lived a decade ago. Traditional census data degrades fast: informal settlements grow overnight, peri-urban sprawl outpaces administrative records, and conflict zones simply go uncounted. A sovereign satellite capability breaks that dependency by generating fresh population grids on a rolling cadence, regardless of whether the census has been updated.

The satellite stack works in three layers. A constellation of optical microsatellites captures sub-metre pan and 3-metre multispectral imagery of the entire national territory at least quarterly. On the ground, a machine-learning pipeline extracts building footprints, estimates floor area and roof count, and fuses those features with nightlight data, road density and survey samples to produce a probabilistic population grid at 30–90 metre resolution. The result is a living dataset rather than a five-year snapshot.

Operationally, this gives planners a tool that census bureaux alone cannot provide. A district government can instantly see whether a new informal settlement has exceeded 10,000 residents — the threshold that triggers infrastructure budget allocation. A health ministry can reroute vaccine cold-chain logistics to match actual population density. During a flood, emergency managers know within hours which grid cells contain the most people and are now inaccessible. Renting this data from a foreign commercial provider introduces both a continuity risk and a sovereignty risk: the dataset capturing every settlement in the country is too strategically sensitive to live on someone else's servers.

What matters

Census data in most developing nations is 7–15 years stale by the time policy acts on it; satellite-derived population grids can be refreshed quarterly.
Sub-100-metre gridded estimates are the minimum resolution required for school-catchment planning, vaccine-logistics routing and flood-evacuation prioritisation.
Building-footprint ML models trained on national imagery outperform global models by 15–25% accuracy in informal-settlement contexts where roof materials and lot patterns are locally specific.
A foreign provider holding a complete, high-resolution human-settlement map of your country is a geopolitical liability — it can be withdrawn, degraded or subpoenaed at any time.

Sovereignty score: 8/10 — A complete, current map of where every citizen lives is one of the most sensitive datasets a state can possess — sovereign ownership ensures it is never withheld, degraded or accessed by a foreign power.

Foreign commercial providers can terminate service, impose export controls, or be compelled by their home-country courts to hand over national settlement data, exposing population patterns, military-adjacent communities and critical infrastructure corridors.
Electoral boundary delimitation, revenue-sharing between subnational governments and humanitarian-response planning all depend on population grids; dependency on a single offshore vendor creates a single point of political failure during exactly the crises when the data matters most.
National statistical offices that anchor their intercensal estimates to a sovereign imagery pipeline retain methodological control and can certify data provenance — a legal and constitutional requirement in many jurisdictions for official statistics.

Reference architecture

Payload: Panchromatic imager, 0.7–1.0m GSD; multispectral imager (blue, green, red, NIR), 3m GSD; 12-bit radiometric depth; 20km swath per pass
Bus class: ESPA-class microsat, 120–180kg, 600W EOL solar power, 3-axis stabilised to 0.02° pointing accuracy, 512 GB onboard storage for daily collection buffering
Orbit: Sun-synchronous LEO at 500–550km, 10:30 local time descending node; 18-satellite walker constellation delivering 4–6 day national revisit for a country the size of Ethiopia or Colombia
Ground segment: 2 national ground stations (X-band downlink at 300 Mbps, S-band TT&C); cloud-based national data centre with sovereign GPU cluster (minimum 8× A100-class GPUs); SatNOGS UHF/VHF backup for housekeeping telemetry
Software & data
Latency
Cost band
Lead time

References

WorldPop Open Population Repository — Methods and Data — WorldPop documents the satellite-ancillary data fusion approach used to produce gridded population estimates at 100-metre resolution globally, and notes the accuracy limitations imposed by infrequent input imagery and outdated census anchors.
UNFPA — Population Data for Sustainable Development — UNFPA identifies timely, high-resolution population data as foundational to achieving SDG targets in health, education and urban planning, and highlights the gap between administrative data and real settlement dynamics.
ESA — Earth Observation for Urban Monitoring — ESA's urban-monitoring programme demonstrates the use of Sentinel-2 and Pléiades imagery for built-up area extraction and population proxy derivation at city and national scale.
USGS — Land Remote Sensing Program: Settlement Detection — USGS describes how Landsat and very-high-resolution commercial imagery are combined with dasymetric modelling to disaggregate census counts into fine-grained population grids.
World Bank — Open Data: Population Estimates and Projections — The World Bank flags persistent data gaps in national population statistics across sub-Saharan Africa and South Asia, underscoring the case for satellite-derived auxiliary inputs to supplement official counts.