Visualising Space‐Time Dynamics: Graphs and Maps, Plots and Clocks
The International Symposium on Spatial‐Temporal Analysis & Data Mining at University College London, 18‐19 July 2011
Michael Batty, Martin Austwick, and Ollie O’Brien
“In many human systems, the size-distribution of the objects or events that define them reveals very few large objects and many small ones. Particularly in systems such as cities and firms, where to be large, you must once have been small, competition tends to make it increasingly unlikely that an object continually gets bigger with most objects never growing out of what has been called the ‘long tail’. These characteristics of scaling systems are often measured by power laws, the most popular of which is known as the rank size rule after Zipf. Populations of cities, revenues in firms, and incomes of individuals seem to follow such laws unerringly, with the form of the scaling remaining relatively stable from time period to time period. However when we examine how each individual object changes its size and rank at a micro level, there is considerable volatility with the half life of city sizes within the top 100 populations, for example, being something in the order of less than a century. To illustrate this volatility at the macro level in the face of strong stability at the macro, we have introduced the idea of the rank clock, where the rank of any city (or object) is plotted around the axes of a clock (where the 24 hour cycle is matched to the period over which the analysis takes place). Clocks for different system display remarkably different patterns and we thus suggest that a classification of different dynamics might be possible when we have enough comparable examples.
Rank Clocks of the Top 100 High Buildings in New York City (a) and the World (b) from 1909 until 2010
“In this talk, we will focus mainly on the visual analytics. The original analytics presents an animation of how systems change in terms of their rank and we begin by illustrating the general idea. As many of the spatial systems in which objects grow and decline in size and rank are spatial, we have linked the clock to related location maps of the objects and have ported the software to the web. This gives us much more power to examine individual cities in space and time but also lets us disseminate these ideas more effectively. We have also developed the rank clock as a kind of radar device where we have direct control over the speed and trajectories of the animation on the clock but we have also moved back to the idea of animating the rank-size space itself as well as more conventional animations of population change associated with sets of cities. One of the features of all these visualisation is that cities can be queried in the context of all others as they change in rank and size, thus providing a rich set of possibilities for the visual analysis of urban dynamics. In this talk we will illustrate these ideas using many examples: Cities in the US from 1790, recent metro areas in the US from the 1960s, skyscrapers in London, Japanese city populations, and UN data pertaining to GDP, literacy and such like.”
- View the presentation [PDF]
