Manhattan: A Tale of Two Population Extremes

Measuring just 13 miles long by 2 miles wide, New York City's most-famous borough has a well-earned reputation for shoebox-sized apartments, shoulder-to-shoulder subway commutes, and cubicle-farm high-rises. Though the US Census provides a cozy estimate of 1.6 million people calling Manhattan 'home', the island's true population swells considerably, like clock-work, every Monday through Friday. Obtaining upper bounds on the workday population of Manhattan is no easy task, and the NYU Wagner School has ballparked a peak estimate of about 4 million people, representing an influx of roughly the population of Houston every workday. To put this in perspective, if Manhattan were a state, it would sit just above Oregon at 27th on a list of U.S. states ranked by population every afternoon.

Though the upper and lower bound for Manhattan's dynamic population provides an interesting point of conversation, the question of how Manhattan's population is distributed hour-by-hour, neighborhood-by-neighborhood is of importance to urban planners, public safety managers, and armchair geographers alike. By no means does Manhattan's population spread out evenly across the length of island- one can imagine the Empire State Building being filled to the brim with inter-state workers by late morning, while at the same time much of the working-age population of Inwood has migrated away from its sleepy uptown hamlet. Finding appropriate sources for estimating the dynamic population was part of a research project undertaken by Justin Fung while studying at Columbia University under the direction of former NYC Transportation Commissioner Lucius J. Riccio, and ultimately, a transit-based solution was found right under their feet. Considering the 'vehicle-of-choice' for most Manhattanites is one of the MTA's 6,500 stainless-steel subway cars, and that the nearly 150 stations serving Manhattan provide reasonably-uniform coverage across the island, it stood to reason that subway entrance and exit data would be a pretty good proxy for population distribution across the city.

Using the Metropolitan Transportation Authority's freely-available turnstile database and Steven Romalewski's MTA subway data in GIS format, estimates for the net flows of people in and out of Manhattan neighborhoods were made on an hour-by-hour basis both historically and for future dates. A time-series analysis of this data confirms that usage patterns are heavily dependent on time, day, and location. The visualization contained herein is the realization of a model of these flows for a hypothetical week in late Spring. As you click around Manhattan, you may uncover both obvious and not-so-obvious patterns, and perhaps even find a good time to take a stress-free bike ride around the block.

Design, Development, Data & Modelling - Justin Fung
Design, Layout inspiration - Oak Ridge National Laboratory, Urbica Design
Map engine - Mapbox GL JS
Graphing engine - D3.js
Github - @citrusvanilla

Import/Export Type:
Daytime Change:
Estimated Population:
The Story of Manhattan's Dynamic Population

The visualization you see here is a model of the dynamic population of Manhattan, block-by-block and hour-by-hour for a typical week in late Spring. The model is currently fixed to your local time. The population estimates are the result of a combination of US Census data and a geographic dispersion of calculated net inflows and outflows from subway stations, normalized to match population daytime and nighttime estimates provided by a study from NYU Wagner. You may exit the ‘Story’ mode at any time by selecting the ‘Visualization’ or ‘Statistics’ tabs in the header above. For more information, click ‘About’. To continue, click the arrows below.

1 of 15