World Population Density
|Data: Euro. Comm. GHSL
Design: D A Smith CASA, UCL
This exploratory map shows data from the fantastic Global Human Settlement Layer (GHSL) produced by the European Commission. Integrating huge volumes of satellite data with national census data, the GHSL has applications for a wide range of research and policy related to urban growth, development and sustainability, and is available as open data. The dataset encourages understanding of the complex hierarchy of human settlement, rather than making simple rural-urban divisions. Some introductory highlights are discussed here with links for further information.
In the early 20th century, geographers like Patrick Geddes observed how rail and road networks were allowing rapidly growing cities to fuse together into vast sprawling conurbations. A classic example is the(click on links to focus the map), a megalopolis as Gottman termed it, stretching hundreds of miles from Washington to Boston, with around 55m people.
The GHSL data shows how widespread urban megaregions have become. China has many, the largest being theregion, and the megaregion, both with around 50m population and growing significantly.
Another related form of megaregion comes from areas of dense agriculture that begin to urbanise with loser patterns of small scale industry. McGee first used the term desakota ("village-city") in relation to the incredible form of, with the densities of urban hinterlands greatly exceeding Western cities but with activity patterns remaining dispersed and linked to agriculture. Similar desakota patterns can be seen in , , and regions, and increasingly in several regions of Sub-Saharan Africa including and surrounding , though with much diversity in each case.
Urban densities are linked to cultures of living, with regions like Latin America and East Asia noted for high density urban forms. Higher population densities are also more prevalent in the Global South, as in poorer countries transport infrastructure is less developed and housing used more intensively. The highest density cities in the world are in South and South East Asia, such as, and (note this depends how density is measured- see the Analysis page).
But these cities are more prosperous than neighbouring rural areas, and high densities can also be linked to affluence., and combine extremely high densities with very high levels of prosperity. The richest large western cities, like , , and are also relatively high density and intensifying.
Another important aspect of density is its relationship with travel demand and energy use. There are many examples of huge urban regions at very low densities, most evidently in the USA with metro regions such asand . Unsurprisingly these cities have the highest rates of transport energy use and carbon emissions in the world.
Sprawl is not however limited to the USA, and similar forms at a smaller scale can be seen in countries like Australia and Canada. Thethat follows the river Rhine through western Germany to Belgium and the Netherlands, is notably dispersed and low density.
As well as exploring contemporary urbanism, many historic patterns remain engraved in the landscape of human settlement. Ancient cities first appeared by fertile rivers that could support the intensive agricultre needed to feed an urban population.spectacularly displays the contrast between the arid Sahara and the rich lands fertilised by the Nile's annual inundation. This landscape has supported urban civilisation for around 5000 years.
The oldest cities we know of were on the Tigris and Euphrates river deltas, near modern dayin Iraq. Another important ancient civilisation grew around the Indus Valley near in modern day Pakistan.
As well as rivers, some ancient transport links are visible in modern settlement patterns. The Roman roadcut across Northern Italy, through what is now Bologna and Parma. Its precise straight form is still evident 2000 years after its completion.
Analysis page- Interactive statistics on country and city density profiles, and change over time.
Euro. Comm. Global Human Settlement Layer- The dataset used to make this visualisation, which is free to download.
Citygeographics Blogpost- more information on this visualisation
The "Interactive Stats" checkbox at the top left of the map turns on density statistics for countries and cities which have been calculated from the GHSL data (1km scale). Roll your mouse over areas to see average densities and built-up areas over time. Zoom in and out to switch between country and city statistics. More discussion on the density statistics can be found on this blog post.
There are different approaches to measuring population density. The GHSL population layer shown in the map describes residents per square km, related to the underlying census data used. It is important to remember that this measure describes where people live, so areas like Central Business Districts can appear low density if they do not include residents (see for example the very centre of London or Tokyo which are dominated by office activity and have few residents).
The interactive statistics include Built-Up Area and Average Population Density of Built-Up Area. The density measures chosen focus on comparing urban areas and do not include areas of unpopulated land such as deserts and mountains. That is why for example that Saudi Arabia is recorded as having a higher population density than the Netherlands, because the built-up area of Saudia Arabia has a higher density of people despite the country including the vast Arabian Desert.
The city statistics on this website use the urban centre boundaries 2015 from GHSL. It is very clear from the GHSL data that there is diverse hierarchy of urban settlements at different scales, and setting boundaries influences statistical results. More advanced statistics with multiple boundaries could be calculated using GHSL but have not been developed here beyond the combination of national and urban centre statistics.
It should be borne in mind that the density classification and data scale used in the map influences its appearance. The classification used is non-linear (see map key top left), so the lower density classes represent changes of hundreds of people per square km, while the top classes represent changes of tens of thousands.
The scale of the underlying data is also important. The GHSL population data is available at 1km and 250m scales, and both layers are used here at different zoom levels. You can see the change in detail, with for example low density rural areas fragmenting into medium density towns and villages when switching between the 1km and 250m level of detail, particularly in countries with extensive rural populations like India.
Finally the map is influenced by the map projection used, which in this case the problematic Web Mercator projection, which shrinks the area of countries near the equator. Note the statistics are all calculated using the original data projection (World Mollweide).
The GHSL is a brilliant dataset, but does inevitably have some limitations due to the mixed quality of the underlying remotely sensed data and census data. In some very remote areas you can see large flat areas of low density, due to the lack of high quality remotely sensed data, for example Nepal and North Western China.
Variable census data also affects the density results, particularly in relation to peak density measures. This is evident when comparing major European cities. High quality census data looks to have been available for France, Spain, the UK and most Scandinavian countries, with dense city centres clearly highlighted. However densities in German, Dutch and Italian cities seem to be too low, probably linked to only lower resolution census data being available. These countries do appear to be an exception however, and good quality data has been sourced for most of the globe.
This is the first major release of the GHSL data and it will likely be further improved in future releases.