1 Spatial Modelling of the Terrestrial Environment - Georeferencial

Characterizing Land Use in Urban Systems via Built-Form Connectivity Models 217
component, encoded in XRAG using the external relations given in equations (9) and (10),
and the union of these (i.e., G(N, r Road→Open→Built ) ∪ G(N, r Road↛Open→Built )).
10.5 Analysis of Built-Form Constellation Structure
Having analytically derived a number of discrete built-form constellations from the data
presented in Figure 10.5, we now examine whether the structural patterns that they exhibit
– in terms of their constituent built-form units (i.e., BUILT regions) – can be mapped
consistently and unambiguously onto specific categories of urban land use. We do so by
evaluating: (i) the packing and density of built-form units; (ii) the spatial and statistical
patterns of built-form area; and (iii) the spatial and statistical patterns of built-form compactness
in each of the constellations.
10.5.1 Built-Form Unit Packing and Density
Figures 10.9(a) and (b) presents information on the packing and density of built-form units
in each constellation. In this context, packing is defined as the number of BUILT nodes
(regions) per hectare, while density is defined as 100 times the total area of built-form units
100× built area
constellation area ).
in a constellation divided by the total area of that constellation (i.e.,
In terms of spatial packing, Figure 10.9a shows that there is a broadly inverse, monotonic
relationship between constellation size (area) and the number of BUILT nodes per hectare. In
other words, large constellations tend to have a proportionally smaller number of built-form
units per hectare, while small constellations have a proportionally large number. Setting
aside for the moment those constellations that fall outside the road network, and which
are therefore characterized by ROAD↛OPEN SPACE→BUILT containment relations, it is
tempting to conclude that the observed variations in built-form packing are indicative of
different types of urban land use and, more specifically, different categories of residential
land. Thus, for example, constellations with a proportionally large number of built-form
units per hectare might be taken to represent compact areas of high-density housing (small
housing units built in close proximity to one another). While this may sometimes be the
case, we also find that constellations representing areas as diverse as traditional 1920/1930s
houses and modern (1990s) residential developments have similar values of built-form
packing. Likewise, the hospital complex has a built-form packing value similar to that
of the 1980s’ residential constellation immediately to the north of it. Built-form packing
cannot, therefore, be used on its own as an indicator of urban land-use type.
Noting that Figure 10.9b is scaled linearly, relative to the maximum areal proportion
of a constellation in this scene that is occupied by buildings (i.e., 38%, for the hospital
complex), it is evident that most of the constellations have low built-form densities. This
is an interesting observation in its own right and suggests a relatively low utilization of the
available space by buildings in each of the constellations. Indeed, few of the constellations
that are known to represent areas of residential land exhibit a built-form density greater
than 20%. Moreover, there is little variability around this figure between different types
and age of housing. Given the range of residential types present in the scene, this is quite
remarkable and suggests that built-form density is also of limited value, on its own, in
terms of distinguishing different categories of urban land use. Even where information on