Visualisation of the World City Network
Funded by HEFCE / Loughborough University award of Study Leave (2000-01)
Researcher: P.J. Taylor in association with D. Dorling (Leeds) and P. Gersmehl (Minnesota)
To construct a set of spatial representations of the world city network with the purpose of (i) displaying its complexity in visually simple forms, (ii) exploring the visualizations as aids in understanding the behaviour of the network/system, and (iii) using the pedagogic potential of the visualizations to explore a new metageography for a new millennium.
From project 7 we have delineated several 55 x 55 matrices of connections between world cities based upon the office networks of 46 global service firms. This provides a starting point. In addition ESRC project 10 will produce a much more comprehensive data set (the "GaWC 100") to work on in 2001.
Creating a new 'cities-metageography' to rival the familiar 'states-metageography' is difficult because a quantum leap in spatial complexity in representing the former. Whereas states can be represented by a simple mosaic of some 200 spaces, the cities constitute a network so that the n=55 becomes a huge number of inter-city connections (2,970 to be exact).
Visualizations will need to be created for different types of media (Orford et al. 1998). The following would seem to be necessary:
(i) for plain text (but including colour) to be reproduced in a journal such as Scientific America and teaching texts/manuals;
(ii) for the GaWC website as a pedagogic tool;
(iii) for use off the Internet by downloading from the website to hard disk;
(iv) for CD-Rom either through downloading or direct production.
Types of solution
Thus far we have relied upon schematic diagrams to overcome the problem of clustering of world cities (especially in western Europe). Moving beyond this, our watchwords should be variety and flexibility. There are four obvious possibilities that fall into two categories.
A In representations of 'real' space
(i) Cartograms of distorted geographical-space to show a world-city space in proportion to quantity of world city formation in an area. This is a standard methodology for coping with clustering leading to overlap in conventional maps (Tobler, 1963). Relative locations between places are maintained as far as is possible to aid in interpretation (Dorling, 1994, 1996).
(ii) Animations of the cartograms to show flows moving between cities. This will almost certainly have to be done city by city i.e click on a city to see its connections moving across the screen.
B. In representations of 'service' space
(iii) New spatial constructions derived from converting connections to distances and using multidimensional scaling to create spaces of different dimensions. For visualization we will deal with 2 and 3 dimensions (possibly 4 using colour).
(iv) Movement within new space allowing interactive traveller to be a 'message' and move around the new space to explore its structures. This will use the 3d service space.
There are no doubt other representations to consider in producing good visualizations.
Which way forward?
(i) Short term : continue as we are now, i.e. experimenting with different approaches using current data.
(ii) Medium term : plan for a large multi-institutional grant(s) in order to bring in expertise from several institutions. This will require using short term work to identify the most promising ways forward.
Dorling, D (1994) Area Cartograms: their Use and Creation, CATMOG 39, Norwich: Environmental Publications
Dorling, D (1996) ‘Cartograms for visualising human geography’ in H M Hearnshaw and D J Unwin (eds.) Visualisation in Geographical Information Systems. Chichester: Wiley
Orford S, Dorling D, and Harris R (1998) Review of Visualisation in the Social Sciences: a Study of the Art Survey and report. Bristol: School of Geographical Sciences, University of Bristol
Tobler W R (1963) ‘Geographic area and map projections’, Geographical Review, 53, 59-78