Where the 7 Billion Are and How Pictures of Nighttime Lights Can Be Useful

There are seven billion people in the world. Where is everybody, really?

Discovering Maps

The Population Reference Bureau (PRB) has a great interactive population density map that incorporates information on standards of living, such as life expectancy and income.

PRB World Population

This PRB map aggregates populations at the country level using data from official census results and multilateral institutions. If countries are your unit of analysis, this map is the ticket. Either way, play with it – it’s fun!

But I want to see a global view of population distribution that isn’t limited by political boundaries. People aren’t actually evenly distributed within national borders, so this PRB map is not really fit for my purpose.

I then came across National Geographic’s very cool map of the world that shows population density at a more granular level (it only displays national boundaries for reference). The map also colour-codes populations into four categories of national income (as measured by GDP per capita – low, lower middle, upper middle and high income). There are great infographics comparing living standards among each of these categories, from life expectancy at birth to carbon dioxide emissions. Browse through this one, too.

To produce “The World of Seven Billion” map, Nat Geo worked with the PRB, United Nations, World Bank , Carl Haub (of the PRB) and Hans Rosling. It used data from the LandScan Global Population Database for 2009 – a dataset that incorporates satellite images of lights at night.

As I followed the LandScan thread, I found (to my great delight) a cornucopia of granular maps and datasets. The FAO has a good review of existing georeferenced population datasets (from 2005). And yes – GRUMP is a real, legitimate acronym (GRUMP is a high-resolution dataset that also incorporates nighttime light images).

Uses for Nighttime Light Images

A picture of nighttime lights from space from the famous DMSP-OLS dataset** Photo from www.earthzine.org.

See the difference. A sample picture of nighttime lights from the famous DMSP-OLS dataset** Photo from http://www.earthzine.org.

Nighttime images can help scientists understand how humans interact with the environment, especially when it comes to energy use and urbanization. The introduction to this 2002 CIESIN paper gives a good overview, though the piece itself is very technical. Earthzine has a good explanation of how this works when assessing human impact on the environment (2008). Experts have proposed using light emission data to find refugees in times of conflict and predict epidemic outbreaks.

And they can be useful in assessing economic activity, particularly where data is missing or misleading. For instance, in 2012, a group of Japanese researchers used the distribution of nighttime lights to examine Myanmar’s economy. There are good commentaries on this work by Banyan in The Economist and by Martin Lewis in GeoCurrents.

Richard Florida very nicely explains the general concept in CityLab (from The Atlantic) and points to two recent studies. Unfortunately for my unfunded curiosity, you have to pay to read the Nordhaus and Chen piece. However, I did get to have a closer look at the other study Mr. Florida showcases, and it’s pretty cool.

A team of scientists looked at Sweden* to see whether nighttime lights can really tell us something about economic activity. In their well-written and relatively layman-friendly 2013 paper, Mellander et. al. find “the link between light and economic activity, especially estimated by wages, to be slightly overestimated in large urban areas and underestimated in rural areas”. Nevertheless, the correlation between nighttime light and economic activity is strong enough for light to be a good proxy for population density.

My favourite paper is the 2013 work by Pestalozzi and colleagues at ETH in Zurich on how nighttime lights are distributed in the world and how this has changed over time. Some of their findings:**

  • The planet’s mean center of light has shifted east since 1992 – just as economic liberalization, privatization and globalization began to roar in our age. This matches up neatly with 2012 McKinsey Global Institute study that found the global economic center of gravity has moved towards Asia.
  • When it comes to more densely urbanized areas, light is unequally distributed within countries. Surprisingly, the degree of inequality is very similar across all countries, whether they be advanced or developing economies. So, globally, countries seem to be equally… unequal… when it comes to light.
  • Light distribution nationally AND globally is getting more unequal over time – light is centralizing.

It’s worth taking a moment to make special mention of the DMSP-OLS, or the Defense Meteorological Satellite Program – Operational Linescan System dataset and satellite images. It’s legendary. The US government collects the data for one thing (to monitor clouds globally) and, because the US government is sharing it, scientists can use the data for a lot of other important things (everything I mentioned above, and more). Mellander et. al. and Pestalozzi et. al. use it.

This July 2014 study by Huang et. al. on the DMSP-OLS offers the most recent, focused and comprehensive meta-analysis and systematic lit review that I know of. It also points out some things to look out for when it comes to next-generation nighttime light detectors (such as the VIIRS – Visible Infrared Imaging Radiometer Suite).

Bright Lights, Big City

I return to the ETH paper. What fascinates me most is the analysis of agglomerations, or continuous areas, of bright lights – city clusters, basically. They ignore national boundaries and look at how these agglomerations have changed over time. Cities that were in the top 20 in 2009 but not in 1992 include Shenzhen/Shanghai, Seoul, Cairo and Sao Paulo.

(Note that brighter doesn’t necessarily mean bigger or denser populations. Also, city regions are named after the cities with the biggest populations, so for instance, the New York city region includes Philly, Bridgeport and Hartford.)

POP QUIZ QUESTION

Can you guess which city region had the biggest continuous area of light in 2009?

Hint: It was the same in 1992 and I highly suspect that it’s the same today (in both years, the second biggest agglomeration of lights was less than half the size of this one).

ANSWER

Look in the Pestalozzi paper linked above or see below***.

There’s a lot of cool science behind it. At the end of the day, if (for whatever reason) I wanted to find people and didn’t have a clue where to start, I would let bright lights, big city go to my head.****

THE STARS

* Why Sweden? Their micro data is the best in the world! The researchers could compare Sweden’s reliable, granular data with nighttime lights data. And Sweden’s energy use is not that different from other advanced industrial economies.

The authors explain the science and technology behind nighttime lights, as well as Sweden’s geo-tagged socio-economic data very clearly – I recommend it.

** Thought-provoking analysis and discussion on what all of this might mean makes the full Pestalozzi et. al. paper well worth the read. The introduction of the paper also explains the history of the DMSP-OLS beautifully.

*** The pop quiz answer according to Frank or Alicia and Jay. Both are karaoke classics. Do look at the paper for the league tables.

**** Jimmy Reed and a mean harmonica interlude.

Advertisements

One response to “Where the 7 Billion Are and How Pictures of Nighttime Lights Can Be Useful

  1. Pingback: The Population of New York City in 2081 | Science | Fiction | Technology·

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s