Good public transit connects people to places. Ideally, this is done efficiently and sustainably, with transit routes and stations serving and connecting the most amount of people possible. But in reality, there's a lot of variation within and between cities in how effectively this is done.
To look at this, we've created maps of major rail transit lines and stations (rapid transit, regional rail, LRT) overlaid onto population density for 250 of the most populated urban regions around the globe. Click the dropdowns below to view how well transit systems serve their populations in different cities.
Each map has the same geographic scale, 100km in diameter, to be easily comparable with each other.
Using these maps, we've also computed several metrics examining characteristics of transit oriented development, and ranked how well cities perform relative to each other. Generally, the greater the density and proportion of the population that lives near major rail transit, the better.
Population data for these maps are from GlobPOP, and rail transit data are from OpenStreetMap. At the bottom of this page we describe these data sources, our methodology, and limitations in more detail.
Rail transit line and station
Population density (people / km²)
|
|
|
|
|
|
|
0
15,000
30,000+
Osaka, JPN
Buenos Aires, ARG
14.72M
Urban population
14.54M
5,100
Urban population density (people / km²)
6,000
7,800
Population density in the area 1km from all major rail transit stations
12,000
69.0%
% of the urban population within 1km of a major rail transit station
35.8%
52.7%
% of the urban area within 1km of a major rail transit station
23.1%
1.31
Concentration ratio (% urban pop near transit / % urban area near transit)
1.55
City Rankings
Select by metric:
Select by region:
US & Canada
Europe
Sub Saharan Africa
Middle East & North Africa
East Asia
South East Asia & Oceania
Latin America & Caribbean
South & Central Asia
Data & Methods
Our list of cities came from a dataset from Natural Earth. We started with a list of the 300 most populated cities, but then manually removed cases where one city was essentially the suburb of another city at our scale (e.g. Howrah was removed since it is very close to Kolkata), as well as removed cities without any rail transit.
For each city, we then defined the urban region shown on the maps as a circle with a 50km radius from the centre point noted in the Natural Earth dataset. We chose to use a standard circle size for all regions to account for idiosyncrasies in how different parts of the world define metro areas. 50km is approximately the outer range that someone would commute to/from a city centre along a major rail corridor.
We sourced the population density data from GlobPOP which provides population count and density data at a spatial resolution of 30 arc-seconds (approximately 1km at the equator) around the globe. Our urban population density metrics are computed after removing areas where population density is less than 400km², to account for how regions vary in terms of how much agricultural land and un-habitable geography they have (e.g. mountains, water, etc. 400km² is the same threshold used by Statistics Canada to define populated places.
We downloaded rail and station data from OpenStreetMap (OSM) using overpass turbo with this query. We then calculated 1km buffers around each station and then estimated the population within the buffered area via aerial interpolation. OSM is crowd-sourced data, and while the quality and comprehensiveness of OSM data is quite good in most cities, there are several cities that have missing or incorrect data. If you see any errors, please update OSM! As OSM data is edited and improved, we'll aim to update our maps and metrics in the future.
There are two main limitations with this transit data: 1) it only includes rail transit, not Bus Rapid Transit (BRT), which in many cities provides comparable service to rail. 2) it does not account for frequency (i.e. headway) of routes. While many transit agencies share their routes and schedules in GTFS format, which includes information about frequency and often technology (bus, rail, etc.), we found that the availability of GTFS at a global scale was not available everywhere, particularly outside of Europe and North America.
Now of course, where people live is just one piece; the goal of transit is ultimately to take people where they want to go (work, school, recreation, etc.). It would be great to layer on employment and activity location data onto these maps to also look at the destination side of the equation as well as analyze connectivity of networks. Something to work on in the future!
---
More information about this project, code, data, etc. are available on GitHub.