Making of the Maps
Urbanization in black-and-white:
We used Landsat 7 information to map Atlanta, Cairo, Calcutta, Istanbul, Mexico City, Milan, Paris, the cities of the Pearl River Metropolitan Region in China, the Randstad in the Netherlands, Rome and the San Francisco Bay Area. These maps resulted from multispectral analysis, a method that uses data bands 7, 4 and 5 of the eight bands sent by Landsat 7 back to Earth. The information contained in these three spectral bands permits the delineation of continuous urbanized areas. The maps follow a morphological definition of urbanized land independent of government or census boundaries.
Since May 2003 Landsat imagery is only partially readable due to a technical failure of the satellite’s scan line corrector. Therefore, for map updates and for additional maps we used feature extraction analysis. Utilizing Microsoft Bing Terraserver to view geographic features at a range of resolutions, depending on availability and legibility, we identified and mapped the extent of urbanization. The 2010 maps for Beijing, Shanghai, and Mexico City were produced in this fashion.
All maps have a fifty by fifty kilometer grid superimposed. To select a frame of fifty by fifty kilometers was of course an arbitrary decision, but when such a frame is consistently superimposed onto all maps, comparisons of scale and size become possible. The surface areas are represented graphically and are shown in the sidebar for each black and white map. This simple computation makes possible the comparisons between the maps and makes concrete the stunning differences in densities that exist among urban populations and the human tolerance for such differences.
Urbanization and Natural Systems in four-color:
The black-and-white footprint maps that emerged from multispectral analysis or from feature extraction analysis were highly abstract. They look like inkblots. Early in the process there was a desire to create a second type of map that depicts some of the natural context within which these cities are located. Rivers, shorelines, mountain ridges, and other large open space systems are shown in order to understand the context of each urban agglomeration. For this second set of maps the urbanized areas are depicted in grey, the natural systems of large open space areas in green and water bodies in blue. Depending on availability, we have mapped, in black, the extent of the historic core around 1900. For the San Francisco Bay Area historic information, we traced from geological quadrangle maps at 1:62,5000 scale, originally produced between 1897 and 1914.
We used visual imagery to verify the urban boundary delineation of the global cities. For San Francisco, we used a 2007 image created from satellite data, purchased from NPA Group, Edenbridge, UK. For Rome, Calcutta, Cairo, and Mexico City, we used Landsat 7 information. For feature extraction analysis, we read information from the Space Agency 2001 publication, “Mega Cities,” Geospace, Salzburg, Austria satellite imagery. We also use imagery from the NPA Group in Edenbridge, U.K., which we found in the 2003 Oxford Atlas of the World. For Beijing, Shanghai, and Mexico City, we use 2009 satellite imagery from Microsoft Bing Maps viewable through ArcGIS software. To delineate the historic core area in the four-color-maps, we used Melville C. Branch, 1978, Atlas of Rare City Maps (Princeton Architectural Press), to delineate the maps of Calcutta, Milan, Paris, and Rome. We found historical data for the Randstad from Polynet.org.uk/Randstad, and the San Francisco Bay Area from various historic maps in the University of California-Berkeley’s map library.
For the purpose of density calculations, we used population statistics published by the United Nations. Methods used to create population statistics vary widely. Given the morphological definition of urbanization on our maps, few sources reveal the exact geographic areas for which the numbers are gathered. For all cities, such numbers depend on government boundary delineations and not on the geographic extent of urbanized areas. The statistics also vary with regard to the quality and age of the census data. For that reason, it was not possible to use a consistent comparison of inhabitants to surface area. The population numbers used in the map collection have been updated from United Nations World Urbanization Prospects (UN), which maintains a website with population statistics for all the world’s cities with more than 100,000 inhabitants. The population statistics were supplemented by data from the ‘World Gazetteer” (WG) and Thomas Brinkhoff, “City Populations” (BCP), and other sources when noted. In a global comparison, we found the population statistics for metropolitan areas to be inconsistent: due to the incompatibility of the term “urban agglomerations,” different sources appear to base their population counts upon different geographic areas and/or different census information.