ABSTRACT
Aerial photography has long been employed as a tool in urban analysis (Jensen 1983, Garry 1992). Indeed, this form of remote sensing is still ex ten ively used today and can now benefit from digital image-processing technique , provided of course that the photographs are digitized first (Futz 1996). For reasons of their widespread availability, frequency of update and cost, however, the focus of urban remote sensing research has shifted more towards the use of digital, multi spectral images, particularly those acquired by earth-orbiting satelli te sensors. This trend was initiated with the advent of what might be described as 'first generation' satellite sensors, notably the Landsat MSS (Multispectral Scanning System), and was given further impetus by a number of second generation devices, such as Landsat TM (Thematic Mapper) and SPOT HRV (High Resolution Visible). Data from the former were initially used to analyze regional urban systems and for exploratory investigations of some of the larger cities in North America (For ter 1980, Jackson et al. 1980, Jensen 1981, Jensen 1983). The availability of still higher spatial resolution (20m/ lOrn) images from the latter enabled more detailed studies of the older, more compact urban areas characteristic of Europe (Welch 1982, Forster 1983, Baudot 1997). The advent, over the last few years, of a third generation of very high spatial resolution ( <5m) satellite en ors is likely to stimulate the development of urban remote sensing still further (National Remote Sensing Centre 1996, Aplin et al. 1997, Fritz 1999). The data they produce should facilitate improved discrimination of the dense and heterogeneous milieu of the old urban cores that are characteristic of European cities (Ridley et al. 1997), and will also help to disentangle the urban fabric in the rapidly expanding agglomerations and 'edge cities ' of many developing countries.
