NOTE: This website has moved to a new address at urbandemographics.org. This content has not been updated since April 2020.
After 1530 blog posts and 471 thousand pageviews over a period of 9 years and 10 months, we are moving to Urban Demographics 2.0, at urbandemographics.org. I will keep a frozen version of this blogspot website online for the record, but I won't be updating any content.
Blogger has been a good companion for a long time, but I felt it was time to move on to a more flexible and versatile platform. Urban Demographics 2.0 is entirely written from within R using blogdown and Hugo. You can find more info in the 1st post of the new website :)
If you follow UD updates via RSS, I am trying move our feed to the new website so you don't have to do anything. I hope this won't break things, but this is our new feed if you want to make sure not to miss any new posts. The Twitter and Facebook channels remain unchanged.
A talented team at the OECD and the European Union have developed a consistent method to delineate metropolitan areas – or functional urban areas (FUAs) – in the entire world. They recently published an open access paper where the explain the method and use it to analyze the population growth of metropolitan areas in the world between 2000 and 2015. You can find more info about the paper below and a geopackage data set of FUAs can downloaded from here.
This paper presents a novel method to delineate metropolitan areas – or functional urban areas (FUAs) – in the entire world and assesses their population trends. According to the definition developed by the OECD and the European Union, FUAs are composed of high-density urban centres with at least 50 thousand people plus their surrounding commuting zones. The latter represent the urban centres’ areas of influence in terms of labour market flows. The proposed method combines a functional and a morphological approach to overcome the dependency on travel-to-work data to define commuting zones and allow a global delineation. It relies on a probabilistic approach and the use of population and travel impedance gridded data across the globe. Results show that around 3.9 billion people, making up 53% of the world population, live in 8,790 FUAs, out of which 17% live in their commuting zones. Between 2000 and 2015, population growth was higher in larger FUAs, highlighting a general trend toward higher concentration of the metropolitan population. Commuting zones grew faster than urban centres, though with heterogeneous patterns across world regions, income levels and metropolitan size.
obs. This is a by-product of the Access to Opportunities Project. I'm grateful for an amazing team of co-authors who helped me put this piece together in such a short time.
Summary:
The Covid-19 epidemic crisis is causing a rapid growth in the number of hospitalizations for severe acute respiratory syndrome (SARS) in Brazil. According to recent studies, this could soon overload the country's public health system (SUS). As of this writing, most of the confirmed cases of Covid-19 are concentrated in the country's largest cities, where the spread of the disease is at a rapid pace and affecting a growing number of people in disadvantaged communities.
In this policy report, we analyze accessibility to healthcare services in Brazil's 20 largest cities. The research focuses on how easily patients with suspected and severe cases of COVID-19 could reach public health facilities. The study has two purposes:
In the first half of the report we estimate how many vulnerable people (low-income above 50 years old) live in areas with poor access to healthcare facilities that could either screen suspected cases of Covid-19 or provide hospitalization of severe cases with the support of ICU beds and mechanical ventilators.
In the second half, we estimate the ratio between the number of ICU beds and mechanical ventilators available at each hospital and the population living withing its catchment area.
These two analyses combined provide actionable information to local authorities. The study puts disadvantaged communities with poor access to health services on the map, indicating in which neighborhoods local authorities could build makeshift hospitals or develop strategies via pre-hospital care with mobile units or through the work of health community agents. This research also helps local authorities identify which hospitals could more likely struggle with the rising demand for hospitalizations, and hence would need investments to expand capacity.
I must say I've been feeling saturated and underwhelmed by the number of data analysts who suddenly became public health and epidemiology experts creating so many data dashboards on COVID-19. There I said it.
Having said that, there many interesting and intelligent people working to understand how the COVID-19 epidemic affects and is affected by demography and human mobility patterns, and how these relationships intersect with and reveal our socioeconomic inequities. I've selected some of the best pieces on the coronavirus crisis I've come across so far.
Our World in Data: possibly one the best websites to get updated data and interactive visualizations, by Max Roser and an amazing team at Oxford.
My all favorite. Article on the NYT illustrate how quarantine is a class and racial privilege. Smartphone location data reveals how many lower-income workers continue to move around in cities across the U.S., while wealthier people 'can afford' to stay home and limit their exposure to the coronavirus. Brilliant piece by Jennifer Valentino-DeVries, Denise Lu and Gabriel Dance.
Creating an animated gif of public transport networks using GTFS data and the gtfs2gps package
In this gist we show with a reproducible example how to create an animation of public transport networks using GTFS data in R. We use a few packages to do this. One of the core packages here is the new gtfs2gps. The gtfs2gps package converts public transport data in GTFS format to GPS-like records in a data.frame/data.table, which we will be using to create a .gif with the gganimate package.
Step 1 - Convert GTFS to GPS-like records
The first step is to process a GTFS.zip file. The function gtfs2gps{gtfs2gps} interpolates the space-time position of each vehicle in each trip considering the network distance and average speed between stops. The output is a data.table where each row represents the timestamp of each vehicle at a given spatial resolution. In this example, we use a sample of the public transport network of Sao Paulo (Brazil) mapped every 30 meters.
# load libraries
library(gtfs2gps)
library(geobr)
library(ggplot2)
library(gganimate)
library(data.table)
library(ggthemes)
library(sf)
library(viridis)
library(sfheaders)
###### 1. process public transport data ------------------
# get local GTFS.zip file
gtfs_zip <- system.file("extdata/saopaulo.zip", package="gtfs2gps" )
# read gtfs
gtfs_dt <- gtfs2gps::read_gtfs(gtfs_zip)
# filter transport services on weekdays
gtfs_dt <- gtfs2gps::filter_week_days(gtfs_dt)
# get transport network as sf object
shapes_sf <- gtfs2gps::gtfs_shapes_as_sf(gtfs_dt)
# Convert GTFS data into a data.table with GPS-like records
gps_dt <- gtfs2gps::gtfs2gps(gtfs_dt, spatial_resolution = 30, progress = T, parallel = T)
head(gps_dt)
# subset time interval. Get only vehicle positions between 7am and 7:30am
gps_dt2 <- gps_dt[ between(departure_time, as.ITime("07:00:"), as.ITime("07:31"))]
# remove observations with 'too high speed' (in km/h)
gps_dt2 <- gps_dt2[speed < 20 , ]
# Convert "GPS" points into sf
gps_sf <- sfheaders::sf_point(gps_dt2, x = "shape_pt_lon" , y = "shape_pt_lat", keep = T)
sf::st_crs(gps_sf) <- 4326
head(gps_sf)
Step 2 - Create gif
Once you have created the GPS-like dataset of the public transport network, creating an animation gif is simple as this:
In case you'd like to add more context info to the gif, you can use rasters as spatial tiles. I've added here some background info using a mapbox tile created by Kaue Braga for the Access to Opportunities Project. Thanks Kaue.
"This Special Issue is dedicated to papers focusing on recent advances in the development of new measures and methodologies to evaluate and analyze the performance of transportation networks. These measures might include, but are not limited to, environmental costs or exposures (e.g., CO2 , noise, pollution); monetary costs (the price of access), complexity, and resilience of multimodal transportation networks; or focus on qualitative aspects of travel, where travel might be seen as a gain instead of cost (such as exposure to aesthetic or green environments). Methodologically, we welcome works using novel ways to measure transport network connectivity, performance, and accessibility, including recent advances in machine learning and AI. Special attention will be paid to papers studying transport-related questions with interdisciplinary approaches."
I am very glad to share our new paper looking at the travel time gap between private and public transport at high spatial and temporal resolutions. The study combines real-time traffic data, transit data, and travel demand estimated using Twitter data to compare this travel time gap in four cities (São Paulo, Stockholm, Sydney and Amsterdam).
Despite remarkable differences between these cities in terms of transportation networks, area, and population size, we found travel times of transit and vs. driving are surprisingly similar across cities: R < 1 for trips shorter than 3km, then increases rapidly but quickly stabilizes at 2 (figure below). Moreover, using public transport generally takes on average 1.4–2.6 times longer than driving a car. The share of area where travel time favors public transport over car use is also very small in all cities. As Giulio Mattioli noted on Twitter, these results 'would confirm that car dependence is much more than just a question of culture & attitudes'.
The paper is open access and it was written in collaboration with a great team led by Yuan Liao and Sonia Yeh at University of Chalmers, Sweden.
Cities worldwide are pursuing policies to reduce car use and prioritise public transit (PT) as a means to tackle congestion, air pollution, and greenhouse gas emissions. The increase of PT ridership is constrained by many aspects; among them, travel time and the built environment are considered the most critical factors in the choice of travel mode. We propose a data fusion framework including real-time traffic data, transit data, and travel demand estimated using Twitter data to compare the travel time by car and PT in four cities (São Paulo, Brazil; Stockholm, Sweden; Sydney, Australia; and Amsterdam, the Netherlands) at high spatial and temporal resolutions. We use real-world data to make realistic estimates of travel time by car and by PT and compare their performance by time of day and by travel distance across cities. Our results suggest that using PT takes on average 1.4–2.6 times longer than driving a car. The share of area where travel time favours PT over car use is very small: 0.62% (0.65%), 0.44% (0.48%), 1.10% (1.22%) and 1.16% (1.19%) for the daily average (and during peak hours) for São Paulo, Sydney, Stockholm, and Amsterdam, respectively. The travel time disparity, as quantified by the travel time ratio R (PT travel time divided by the car travel time), varies widely during an average weekday, by location and time of day. A systematic comparison between these two modes shows that the average travel time disparity is surprisingly similar across cities: R < 1 for travel distances less than 3 km, then increases rapidly but quickly stabilises at around 2. This study contributes to providing a more realistic performance evaluation that helps future studies further explore what city characteristics as well as urban and transport policies make public transport more attractive, and to create a more sustainable future for cities
The relationship between travel distance and travel time ratio R (PT travel time divided by the car travel time)
We are looking for driven candidates with good data analysis skills in R and interested in evaluating policy impacts on issues of transport accessibility and urban inequalities.
