I spent the last few days at the Applied Urban Modelling conference in Cambridge. While much of the work I do has a technical or engineering focus, the bulk of this conference had an economic flavour and looked primarily at land use and transportation modelling. I certainly learned a lot and met several great researchers.
One of the most interesting tidbits was Kiril Stanliov and Paolo Masucci‘s paper on the history of London’s street network. Having painstakingly created a hand-digitized series of maps showing the growth of London’s streets from 1786 to the present, they presented a series of analyses based on this spatial database.
The dataset is particularly helpful for answering a common question in urban geography: what is the boundary of a city? Administrative boundaries may be convenient but they often fail to capture what’s actually going on, unlike a more functional definition such as that adopted by Columbia’s GRUMP data base or the OECD’s recent Compact Cities report. As a simple example, here is a picture of London at night with the official Greater London Authority Boundary superimposed.
Paolo’s cool trick was to develop a definition of an urban boundary based on the density of street intersections. Using the Jenks natural break algorithm, the maps can then be divided into “urban” and “rural” areas. The results show the growth of London much more clearly than other techniques, although once the dataset hits the present GLA boundary any further expansion is obscured. The following image is from earlier, but related work, and should give a rough idea of the data set and results.
I don’t believe they’ve published these results yet, but it will be worth checking Paolo’s Arkiv page for progress.