Census data tells you where people sleep; it says almost nothing about where they are at 10 a.m. on a Tuesday or during a mass-casualty event. The gap between residential population and ambient population drives chronic errors in transport planning, emergency response pre-positioning, flood and earthquake casualty estimation, and pandemic contact-tracing. No survey programme updates fast enough to catch a new industrial zone pulling 80,000 commuters daily from three provinces, or a resort city that quintuples in population every weekend.
A purpose-built constellation closes that gap by stacking three complementary signals. Nighttime lights from a low-noise panchromatic sensor resolve lit floor area as a proxy for sleeping population; daytime multispectral imagery reads parking-lot occupancy, market-stall density, and construction footprint expansion as proxies for working population; thermal infrared detects heat load from building clusters, separating occupied from vacant structures. Crossing all three signals with mobile-network activity data—licensed from the national operator rather than a foreign aggregator—yields ambient population grids updated every few hours.
The operational payoff is concrete. Emergency management agencies get pre-event baseline grids so that, when an earthquake strikes at 7 a.m., the casualty model reflects who is on the road, not who is listed as a resident. Transport ministries rebalance bus and rail capacity against real commuter flows rather than five-year-old projections. Tax authorities correlate declared commercial activity against detected daytime footfall to identify underreporting. A sovereign constellation running this service continuously makes population intelligence a routine utility, not an emergency procurement after the disaster has already started.