Population estimation and the mapping of activity spaces: process and progress in the LandScan USA project

Keywords: population, activity spaces, built environment, GIS
Session Type: Virtual Paper Abstract
Day: Saturday
Session Start / End Time: 2/26/2022 11:20 AM (Eastern Time (US & Canada)) - 2/26/2022 12:40 PM (Eastern Time (US & Canada))
Room: Virtual 63

Abstract

Mapping where people are during the day and night—and how many we can expect to find in any given location—requires knowing the locations and functions of the buildings and facilities we use every day: homes, workplaces, schools, and shops. A comprehensive database of these indoor activity spaces across the United States underlies the LandScan USA project at Oak Ridge National Laboratory, which produces daytime and nighttime population estimates across the country, gridded at a sub-100m resolution, and forming the population component of the U.S. government’s Homeland Infrastructure Foundation-Level Data (HIFLD) program, a program that provides critical data and situational awareness to emergency responders, planners, and researchers. In recent years, advances in geospatial AI have made it possible to detect and delineate the outlines of virtually every building from aerial or satellite imagery, giving us—in unprecedented accuracy and detail—a canvas on which to map the location and scale of human activity. LandScan USA employs these data in combination with land use at the level of individual parcels and tabulations of activity-specific populations (residents, workers, students, etc.). This talk will examine the strengths and limitations of this approach and outline current and future research and development threads that have the potential to increase the fidelity, temporal resolution, interpretability, and usability of these population estimates.