Association for Public Policy Analysis & Management

Our collective understanding of how people interact with cities and the built environment has historically been limited to surveys collected at low temporal and spatial frequency. The definition and estimation of urban population and population density at various spatial and temporal scales has been a persistent problem of urban science and the study of city dynamics (McDonald 1989; Sutton, Roberts, Elvidge, and Baugh 2001). Population estimates are used to estimate future demand for critical infrastructure and resources (such as energy, food, water, etc.), to efficiently provide social services and other city services, and to effectively plan for future land use changes and transportation needs. In many cases, population estimates drive policy decisions and the prioritization of public investment across regions, cities, and neighborhoods. Census data, including the decennial census and the American Community Survey, provide our most robust glimpse of spatial population at a single point in time. However, Census data are extremely limited not only by temporal and spatial constraints, but also by the time lag in the availability of updated data for analysis purposes. Relying on annual surveys for planning and policy design increases the time required for policy innovation, implementation, feedback, and evaluation.

New streams of data are being generated by a range of in-situ instrumentation, mobile sensing, and social media that can be integrated and analyzed to better understand urban activity and mobility patterns (Bettencourt 2014; Kontokosta 2016). While several studies have focused on understanding activity patterns and flows of people throughout a city (Becker et al. 2011; Jiang, Ferreira, Gonzalez 2012; Song, Qu, Blumm, and Barabasi 2010; Gonzalez, Hidalgo, and Barabasi 2008), these data can also be used to create a more spatially and temporally granular picture of local population at a specific point in time, and to forecast localized population given some exogenous environmental or physical conditions.  Effectively modeling neighborhood population dynamics could have significant implications for city operations and policy, and for the engagement and empowerment of residents in neighborhood planning processes and their collective ability to address community quality-of-life concerns.

Our research seeks to develop a real-time census of the city within the context of describing an urban phenology. Using unique access to WiFi connection data for 2015 covering more than 50,000,000 records for Lower Manhattan in New York City, combined with correlative data from the American Community Survey, the Longitudinal Employer-Household Dynamics survey, and New York City administrative records, we present a model to estimate a spatially and temporally granular population estimate of residents, workers, and visitors/tourists in a given neighborhood and localized to a specific block or geolocation. The goal of this work is to develop a real-time population estimate at the block scale that will serve as a baseline for understanding neighborhood activity levels and provide critical insight into population density dynamics to inform city operations, policy, and planning.