A pluralistic approach to defining and measuring urban sprawl - Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment - page 161

meet. Next, we present variables that have been used to measure
sprawl in empirical studies. We conclude the section by ranking
the variables according to the criteria we set at the outset of
the section.

time are not, on their own, sprawl measures. They are, however,
the most important variables to measure because most sprawl
measures that follow are dependent on them. Urban land cover
is referred to analogously as built space or impermeable surface
cover, although each has slightly different implications for how
much land will ultimately be quantified as urban (Orenstein
et al ., 2010).
Estimations of values for urban land cover and changes in
land cover over time are also the most important contributions
of remote sensing experts to studying sprawl processes. The sheer
amount of published literature on urban remote sensing (this
topic included) testifies to its importance as well as to the rapidly
advancing state of the art (Ward, Phin and Murray, 2000; Ste-
fanov, Ramsey and Christensen, 2001; Zhang et al ., 2002; Sutton,
2003; Rogan and Chen, 2004; Xian and Crane, 2005; Martinuzzi,
Gould and Ramos Gonzalez 2007; Jat, Garg and Khare, 2008; Pu
et al ., 2008; Bhatta, Saraswati and Bandyopadhyay, 2010). Aside
from estimating generic urban land cover, rapid improvements in
the quality of data and interpretive methodologies make it possi-
ble to differentiate between types of urban land cover (Foresman,
Pickett and Zipperer, 1997; McCauley and Goetz, 2004). Differ-
entiating growth in residential area (as contrasted with industrial,
business and commercial areas) and in low-density residential
area is particularly important, as they are two sub-variables
commonly used for characterizing sprawl (McCauley and Goetz,
2004; Irwin and Bockstael, 2008). Computing the amount of and
change in availability of developable land, assessed in conjunction
with ancillary data like statutory land use plans, also provides
important data for sprawl characterization.
Sprawl measures suggested in the literature can be divided
into five major groups (Table 12.1):

12.4.1 Criteria for a good sprawl
measurement variable

In order to minimize discord between various studies on sprawl,
spatial variables used to measure sprawl should be held up to
certain criteria. These criteria include:
1 Objectivity . The variable must be quantifiable and repro-
ducible (Ewing, Pendall and Chen, 2002; Lopez and Hynes,
2003; Torrens, 2008). Since sprawl is a subjective term,
researchers should provide all measured values and the values
at which they consider sprawl to be occurring, thereby allow-
ing users to decide for themselves if the values suggest sprawl
or not (Wilson et al ., 2003).
2 Applicability to a large number of places . The variable must
be generalizable to a wide range of study sites and times
and not be specific only to the study site of the current
examination (Lopez and Hynes, 2003; Wilson et al ., 2003;
Irwin and Bockstael, 2008; Torrens, 2008). If it is applicable
in only particular situations, the researcher should be explicit
regarding the limitations of the variable's application.
3 Appropriateness for multiple spatial scales of investigation .
Sprawl may occur at a variety of spatial scales (e.g. hous-
ing unit, neighborhood, town. region, metropolis, state or
country). A good variable is robust enough to apply to mul-
tiple scales of investigation, while others may be appropriate
to only a certain scale.
4 Meaningfulness, usefulness, and simplicity . The variable must
capture one of the descriptive elements of sprawl (Ewing,
Pendall and Chen, 2002; Lopez and Hynes, 2003; Wilson
et al ., 2003; Torrens, 2008). The data emerging from sprawl
studies must be relevant to stakeholders, and therefore it is
crucial that the variables are easily explained, understood and
relevant to them (Lopez and Hynes, 2003; Wilson et al ., 2003).
5 Ease of application . An additional quality of a good sprawl
indicator is one that is not overly dependent on complex
calculations, software that requires a highly specialized skill
set, or inaccessible data such that other researchers or prac-
titioners would not be able to employ the measures in their
research. Some variables are good in theory, but the data
maybeinaccessibleornotavailableatthescaleofresolution
or historical period desired for research. On the other hand,
some methodologies for data preparation demand a high level
of computational or spatial analysis skills, an advanced under-
standing of spatial metrics, or access to computer hardware
and software that may make the variables less desirable for
the intended end user.

● density (building and population);
● relative population growth rates;
● spatial geometry of built and open space;
● accessibility between residential, commercial and business
areas;

aesthetic measures.

●

Due to the nature of the current volume with its emphasis on
remote sensing, we focus on those variables whose values can
be derived through remote sensing data and analysis. We briefly
mention other variables as well, but those are generally quantified
using other, non-remote sensed data, such as census and survey
data. As such, aesthetic measures as a category are not included
in Table 12.1, but see below).

12.4.2.1 Density
There are various types of densities, as well as many ways and
scales at which to measure them (Churchman, 1999; Burton,
2000; Chin, 2002; Tsai, 2005). Density can be defined as the
ratio between the amount of a certain urban activity and the
area on which it exists, for instance population size (Lopez
and Hynes, 2003) or housing units per unit area (Razin and
Rosentraub, 2000). Population density is considered a key theme
in sprawl literature (Galster et al ., 2001) and while some argue
that it is the most important measure (Fulton et al ., 2001; Maret,
2002; Lopez and Hynes, 2003), they are careful to specify that,
while important, it is not the only measure of sprawl. As a
sprawl measure, population density fails to take into account

12.4.2 What shall wemeasure?

Prior to the calculation of sprawl measures, total built area must
be measured. Urban spatial growth and its rate of change over