Human Cartograms

Definition

Human cartograms depict a geographical area in units other than land area.

Maps usually represent land area. When one thinks of the world, a picture resembling a land area map comes immediately to mind… Cartography can depict territory in units other than square kilometres… In these maps, the size of each country has been changed to represent the variable that is being mapped, rather than the land area for that country. [p. 234][1]

[Cartograms are] maps in which the sizes of geographic regions such as countries or provinces appear in proportion to their population‚Ķ [p. 7499][2]

Human cartography in the last decade (since 2004) has used an algorithm developed by Newman and Gastner to create the cartograms:

Changing the size of countries on a map is not a new idea. What is original… is the algorithm used to create these. This algorithm… allows countries to be expanded and shrunk while preserving their boundary shapes and relative positions as far as is possible… The other noteworthy aspects of this algorithm are the quality of the cartograms created and the speed with which these are produced.[p. 234][1]

Previous algorithms have been computationally intensive, as well as produce “highly distorted maps that are difficult to read or projections that are badly behaved under some circumstances, with overlapping regions or strong dependence on coordinate axes.”[p. 7499][2]

Advantages

Traditional mapping can plot either counts or proportions. Counts will naturally highlight areas with larger populations (e.g. cities). See: [XKCD Heatmap]. Rates overcome this problem, but do not take in to account the background population at all. For example, two areas might have the same incidence rate of 1 per 1,000. In a city this means a lot of people (thousands!), but in rural areas this might mean one or two cases.

One case in a thousand means something entirely different in Sydney from what it means in Siberia.[p. 7499][2]

Cartograms produce maps that illustrate both the number of cases and the proportion of the population affected.

Limitations

  • Data must exist for all regions included in the cartogram. Where there is missing data, it must be estimated.
  • As with all things, the quality of the cartogram depends on the quality of the data. Garbage in, garbage out.

Such mapping obviously requires good quality data from every country. Most of these have been sources from United Nations agencies. However, wherever data are missing, estimates have been used. These estimates have some limitations. A pertinent example is that of data about refugees. During a war or natural disaster – major causes of displaced populations – disruption of the government often makes accurate recording of numbers difficult. [p. 234][1]

Counts, not Proportions

Human cartograms must use counts, not proportions, rates or percentages. Counts can be summed across all regions considered and provide a meaningful total. Proportions, etc., across regions cannot be summed and provide any useful or meaningful information.

Existing Human Cartograms

The [World Mapper] project has over 300 examples of human cartograms depicting the world geography by theme.

Creating a Human Cartogram with Scapetoad

[Scapetoad] accepts standard ArcMap shapefiles and creates a human cartogram. The shape file loaded needs to contain the outline of the geography in question (for example ward, LSOA, etc.), as well as the data relating to the geography in question. Typically load ArcMap and add a shapefile with a geographic outline, then join data in question, then export to a new shape file to input in to Scapetoad.

Don’t forget to use counts, not proportions!

References

1. Barford, A and Dorling, D (2006) A new approach to mapping the world: Visualizing facets of international health in The National Medical Journal of India, vol. 19, no. 4, pp. 234-235.

2. Gastner, Michael T. and Newman, M. E. J. (2004) Diffusion-based method for producing density equalising maps in Proceedings of the National Academy of Sciences of the United States of America, vol. 101, no. 20, pp. 7499-7504.

Advertisements