1 Spatial Modelling of the Terrestrial Environment - Georeferencial
1 Spatial Modelling of the Terrestrial Environment - Georeferencial
1 Spatial Modelling of the Terrestrial Environment - Georeferencial
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
10<br />
Characterizing Land Use in Urban<br />
Systems via Built-Form<br />
Connectivity Models<br />
Stuart Barr and Mike Barnsley<br />
10.1 Introduction<br />
It has been argued that information on urban land use cannot be directly inferred from<br />
automated analyses <strong>of</strong> multispectral, remotely sensed images, when performed at <strong>the</strong> level<br />
<strong>of</strong> individual pixels (Sadler et al., 1991; Barr and Barnsley, 1995, 1999). While <strong>the</strong> spectral<br />
response <strong>of</strong> <strong>the</strong> latter may be functionally related to land cover (i.e., <strong>the</strong> physical materials,<br />
such as grass, tarmac, concrete and water, at <strong>the</strong> Earth surface), land use is an altoge<strong>the</strong>r<br />
more complex construct – one that is defined principally in terms <strong>of</strong> socio-economic activity<br />
and that is typically expressed spatially across multi-pixel regions <strong>of</strong> <strong>the</strong> urban scene<br />
(Barnsley and Barr, 2001; Barnsley et al., 2001). Human photo-interpreters, on <strong>the</strong> o<strong>the</strong>r<br />
hand, are generally able to recognize and classify a range <strong>of</strong> different urban land-use<br />
categories in airborne- and satellite-sensor images. One might posit that <strong>the</strong>y do so, whe<strong>the</strong>r<br />
consciously or subconsciously, by identifying key land-cover objects/entities within <strong>the</strong><br />
urban scene (e.g. buildings, roads and various types <strong>of</strong> open space), by considering <strong>the</strong> size<br />
and shape <strong>of</strong> <strong>the</strong>se objects, and by examining <strong>the</strong> spatial and semantic relations between<br />
<strong>the</strong>m. It seems likely, <strong>the</strong>refore, that automated analyses <strong>of</strong> urban land use in remotely<br />
sensed images can best be achieved using a similar multi-stage approach: in o<strong>the</strong>r words,<br />
by inferring land use from an analysis <strong>of</strong> previously identified land-cover types (Wharton,<br />
1982a; Moller-Jenson, 1990; Barnsley and Barr, 1992, Johnsson, 1994, 1996, 1999;).<br />
<strong>Spatial</strong> <strong>Modelling</strong> <strong>of</strong> <strong>the</strong> <strong>Terrestrial</strong> <strong>Environment</strong>. Edited by R. Kelly, N. Drake, S. Barr.<br />
C○ 2004 John Wiley & Sons, Ltd. ISBN: 0-470-84348-9.