Receive our newsletter - data-led analysis, original reporting and insights
Community / Public space

Measuring cities by satellite means we can finally compare them

As anyone who’s ever tried it will know, cities are bloody hard to measure, define, or compare. Some city governments collect data for the entire metropolis, plus its sprawling suburbs, while others stick firmly to the city’s core. Comparing data from different cities, or even trying to accurately work out which are biggest, is nigh-on impossible. 

One organisation that spends quite a lot of its time trying to compare cities is the World Bank. And, perhaps spurred on by years of frustration, researchers working on urbanisation in East Asia have come up with a solution. It’s pretty technical, but it could represent the answer to all our geeky data prayers. 

Here’s how it works. The researchers compared satellite images of all of East Asia from 2000 and 2010 using something called “change-detection” software. Each pixel represents a 250x250m square; the software examines each in turn, and marks it as “built-up” if at least 50 per cent is covered by man-made development like roads or buildings.

Then the researchers cross-referenced these maps with population maps; if a pixel is built-up, and part of a settlement of over 100,000 people, it counts as part of an urban area.

Here’s a map produced by the researchers showing new urban growth, marked in red, in China’s Pearl River Delta region between 2000 and 2010:

Click for a larger image.

And here’s Tokyo, which expanded very little in comparison:

Click for a larger image.

Of course, this new definition comes with some of the same problems that have always beset city definitions: the question of how large the gap between built-up areas in the same urban area can be, for example. And huge developments built on the edge of Chinese cities, but lying completely empty, pose a definitional challenge: does an area count as urban area if no one lives there?

All the same, this technique comes with a unique selling point: it’s transferable. The World Bank was able to map the entire region using the same method. There’s no reason it couldn’t map every metropolis in the world in the same way (apart, perhaps, from some snow-covered ones). 

In a broader sense, the method could allow researchers to move beyond nit-picky definitions of cities and figure out exactly how many people now live in urban settlements. In East Asia, according to the World Bank report, 200 million people moved to urban areas between 2000 and 2010 – the equivalent of the entire population of Brazil. Meanwhile, the urban areas themselves expanded by an average of 2.4 per cent per year over the same period. This information seems much more useful than a long debate on whether Manila is actually as big as it claims it is. 
This article is from the CityMetric archive: some formatting and images may not be present.