# New Global Population Map Tracks Where People Actually Are, Hour by Hour
The Department of Energy's Oak Ridge National Laboratory released LandScan Mosaic, a population dataset that maps human movement patterns by analyzing building usage throughout the day. The tool represents a major upgrade to LandScan Global, ORNL's existing population distribution model that government agencies and researchers have relied on for decades.
Traditional population datasets show where people live based on census data and residential addresses. LandScan Mosaic goes further. It estimates where people actually spend their time during different hours, accounting for commutes to work, school attendance, shopping trips, and other daily activities. The system models building occupancy patterns to reveal these dynamic movements.
The researchers accomplished this by integrating satellite imagery, building footprint data, and machine learning algorithms. The model analyzes characteristics of structures, their sizes, locations, and typical usage patterns to infer occupancy at different times. A large office building near a city center behaves differently during business hours than a residential neighborhood. A hospital operates around the clock. Shopping districts fill during afternoons and weekends. LandScan Mosaic captures these variations.
"This level of detail transforms how we understand population distribution," explains the ORNL team. The dataset enables more precise disaster response planning. Emergency responders can now target rescue operations based on where people actually congregate rather than where they sleep. During a flood or earthquake, knowing that a stadium holds 50,000 people on Saturday afternoons but sits empty on Tuesday mornings changes evacuation strategies.
Humanitarian organizations benefit from improved targeting. Aid agencies delivering food, water, or medical supplies can reach concentrations of people more effectively. Infrastructure planners use the data to size hospitals, schools, and transit systems appropriately. Urban developers identify where utilities and transportation networks matter most.
National security agencies gain better intelligence on population movements in unstable regions. Monitoring where civilians cluster helps military and diplomatic operations minimize harm to noncombatants.
The original LandScan Global launched in the 1990s and now operates as a standard reference tool used by the United Nations, World Bank, and disaster management agencies worldwide. ORNL updates it annually with new satellite data, making it freely available to researchers and governments. LandScan Mosaic builds on this foundation with temporal resolution, adding the time dimension.
The new dataset covers the entire globe at approximately 1-kilometer resolution, meaning users can identify population concentrations at a fairly granular scale. The researchers validated their model against known occupancy patterns in test cities, confirming accuracy improved substantially over previous approaches.
LandScan Mosaic carries some limitations. The building-based model performs best in developed countries where mapping infrastructure is reliable. In regions with incomplete building surveys or poor satellite coverage, estimates carry greater uncertainty. The system also works at moderate resolution, so street-level precision remains beyond reach.
ORNL plans regular updates as satellite technology improves and researchers refine the underlying algorithms. The team made the dataset available through the LandScan portal, where registered users can access maps showing population distribution across 24-hour periods. Government agencies can integrate the data into emergency planning systems immediately.