Mapping and measuring urban places: Are we there yet? (Part 1/2)


This page in:

Source: Deuskar, C., and Stewart B.. 2016. “Measuring global urbanization using a standard definition of urban areas: Analysis of preliminary results” World Bank
This satellite image shows Sao Paolo's estimated “urban areas” based on a WorldPop gridded population layer. Areas in yellow are areas with at least 300 people per km2 and a known settlement size of 5,000 people. Red areas represent a population density threshold of at least 1,500 people per km2 and a known settlement size of 50,000 people.

There remains a surprising amount of disagreement over precisely what “urban” means despite the ubiquity of the term in our work. Are urban areas defined by a certain amount of artificial land cover such as permanent buildings and roads? Or are they more accurately described as spatially concentrated populations? The answer often depends on what country you are in, as their administrative definitions of urban areas can vary widely across and between these two dimensions.
Without a globally consistent measure of urban areas, it can be difficult to track changes in built-up areas  (land surface coverage comprised of buildings and roads) and population growth across time and space. This impacts how policymakers may understand and prioritize the challenges cities face and what investments or reforms may be needed. In a new paper, “Bright Lights, Big Cities: a Review of Research and Findings on Global Urban Expansion,” I provide a brief introduction to some of the current approaches for measuring urban expansion and review the comparative findings of some recent studies.

The UN’s World Urbanization Prospects (WUP), perhaps the most comprehensive and widely cited measure of urbanization across the world, draws from a compilation of country-level population totals based on administrative definitions. A key weakness with this set is that since each country defines “urban” differently, it is difficult to accurately compare one country’s urbanization to another, as well as to estimate the urban population of a group of countries or the world itself. Recent work has provided more sophisticated ways to measure urban growth and expansion using both satellite map data and careful application of population data.
The growing role of population data and mapping technology
The increased quality and availability of satellite map data has given us a better view of the form and extent of human settlements, for example: 
  • Algorithms are used to distinguish the colors and patterns on maps to discern between built-up areas and natural ground cover or water-covered surfaces.
  • Moderate Resolution Imaging Spectroradiometer (MODIS) provide medium resolution maps (with resolutions of about 500m) that specialists have used to map urban built-up areas across regions.
  • Higher resolution maps such as Quickbird (around 2m or less and) can be used at the city level to estimate different land use types, based on the colors, shapes, and ground cover.
  • Some recent work has even begun to use maps that distinguish different types of development for estimating natural disaster risks in urban areas.
Other maps do not even rely on daylight. Nighttime lights data using Defense Meteorological Satellite Program Optical Line Scanner (DMSP-OLS) at a higher resolution and greater electromagnetic spectrum coverage, Visible Infrared Radiometer Suite (VIIRS), allow researchers to distinguish the form and brightness of built-up areas by recording ambient light. These maps have not only been used to track urban form and expansion, but also to estimate the intensity and location of economic activity within cities.
Population-based estimates of urban areas aim to refine the application of available population census data (such as what is used for the WUP).  These approaches use known population centers and apply a grid across administrative boundaries (usually of about 1km2). They then produce an estimated distribution of the population within built-up and non-built-up areas within each grid cell, which corresponds more closely to where people are likely to reside (see the example below). Several approaches for estimating urban populations, such as Worldpop and GHSPop, build on this approach to standardize measurement by applying minimum population threshold and population density standards to define urban settlements, and then calculating the total number of people within these areas.
Source: Deichmann et al. 2001
What does this mean for future work?
These techniques provide a different picture of the global urban population from the WUP. For example, the populations of Africa and South Asia may be much more “urban” than what official estimates suggest. By the same approacch, Latin America’s population may be significantly less “urban” than otherwise thought. However, there is no consensus on the minimum settlement population and density thresholds ascribed to urban areas, and estimates can vary greatly depending on these adjustments. 
The next big step on this work is to link this population-based approach with a standard global-level map of built-up areas. This would allow for a consistent and comparable estimate of the total number of people and built-up areas across the world.
In my next post, I will offer some caveats and suggestions for researchers and practitioners on how to consider using these data and areas for future improvement.
Download draft paper "Bright Lights, Big Cities: a Review of Research and Findings on Global Urban Expansion" and feel free to provide your feedback by posting a comment. 


David Mason

Urban Development Specialist

Join the Conversation

Ignacio Arnaiz
August 03, 2017

This is the type of approach to the problem that is made from international organizations, it seeks an objective method of measuring the city based on satellite observation, do not require the collaboration of anyone, only have a satellite. It reminds me of when the English made cardboard tanks to deceive the German photo-interpreters by simulating huge military bases. Well, our point of view is that the systematization of the transformation processes of the city must be the source of data, it is as if the Germans could have put an accountant at the exit door of the factories of tanks of England and count well Those who destroy, would surely have more truthful information than looking at the stains of an aerial photograph.
Can we intervene in the processes of urban change to count? Count people, houses, parks, endowments, infrastructures ... If we can not, then we will have to opt for the satellite or ... the drones. But it is as if a bank inventoriease from time to time its balance thanks to that its employees measure with a meter the volume of tickets that are in the safe, instead of having a computer system that updates to the second its balance because all the transactions Are automated.
Our purpose is to provide cities with tools to control their processes of change and build certain and complete inventories of their content. Very low cost cloud tools ($ 100 / month approx) using collaborative methods (OSM has set the tone and demonstrated success). Although the truth is that each city addresses the problem differently, there is no political will to share the data, nor is there a national authority, much less a global one that requires it. Therefore it will be necessary to continue measuring with the satellite.
(Translated from Spanish by google)

David Mason
August 23, 2017

Dear Ignacio, thank you for your comment. I agree that data limitations, particularly in terms of consistent definitions and measures are of ongoing concern. At the city/municipality level, as you note, there are a number of mapping and data collection tools available for policy makers to better locate public services and facilities and to improve planning and investment to match areas of population growth. These are indeed important for improving city management and governance.  Apart from that, what I think the literature reveals is that as we move higher in scale (from local to national to global) it becomes increasingly difficult to measure the nexus between urban spaces and urban populations in a consistent fashion. At this level, key questions are more broad; who/how many people are living in urban areas, and how much urban area is there? From then, we can look into the type and quality of services and amenities available, and which local government may be tasked to provide them.