Landscape metrics in the analysis of urban land use patterns: A

Landscape and Urban Planning 99 (2011) 226–238
Contents lists available at ScienceDirect
Landscape and Urban Planning
journal homepage: www.elsevier.com/locate/landurbplan
Landscape metrics in the analysis of urban land use patterns:
A case study in a Spanish metropolitan area
Francisco Aguilera a,∗ , Luis M. Valenzuela a , André Botequilha-Leitão b
a
b
Environmental Planning Laboratory (LABPLAM), University of Granada, Campus de Fuente Nueva, Edificio Politécnico s/n, 18071 Granada, Spain
Research Centre for Landscape, Territory and Urbanism (CIPTU), University of Algarve, Faro, Portugal
a r t i c l e
i n f o
Article history:
Received 26 April 2010
Received in revised form 4 October 2010
Accepted 19 October 2010
Available online 14 December 2010
Keywords:
Spatial metrics analysis
Urban land use patterns
Future scenarios
Cellular automata
Metropolitan Area of Granada
a b s t r a c t
Urban growth patterns are characteristic of spatial changes that take place in metropolitan areas (MA).
They are particularly prominent in more recently formed MAs, such as those in certain locations in Spain,
where the structure of the traditional city has undergone sweeping changes. Given the capacity of spatial
metrics to characterize landscape structure, these metrics can be a valuable instrument to identify growth
patterns in MAs and to evaluate possible urban growth options, based on spatial characteristics.
This article focuses on a medium-sized MA (Granada, Spain), and explores the use of spatial metrics
to quantify changes in the urban growth patterns reflected in three future scenarios (2020). The scenarios were simulated with a model based on cellular automata, which reproduced three urban growth
processes (aggregation, compaction, and dispersion) and four urban growth patterns (aggregated, linear,
leapfrogging, and nodal). The scenarios were evaluated with metrics that quantified changes in the spatial
characteristics of urban processes. Thus, for example, the NP and AREA MN allowed us to characterize the
decreased aggregation of high-density residential land uses in one scenario (S1) and the linear growth
patterns in industrial land uses in another scenario (S2). In this way, spatial metrics were found to be
useful for the evaluation of urban planning.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
Urbanization is a worldwide phenomenon that has increased
significantly in the last century. According to Seto and Fragkias
(2005), currently, half of the world’s population resides in urban
areas, and this percentage will continue to rise. Urbanization is
the most dramatic form of irreversible land transformation (Luck
and Wu, 2002), affecting both landscapes as well the people who
live in and around cities. Urban sprawl is a type of urban growth
characterized by a low-density, dispersed spatial pattern with both
environmental and social impacts. It is currently one of the main
challenges for spatial planners in the 21st century (Muñiz et al.,
2008; Poelmans and Van Rompaey, 2009).
In Spain, urban growth has soared in recent years (Observatorio
de la Sostenibilidad en España, 2006). More specifically, 1995–2007
was a period of intense economic activity that witnessed the expansion and consolidation of large metropolitan areas (i.e. Madrid,
Barcelona, and Seville). During this same period, medium-sized
metropolitan areas such as Malaga, Granada, and Alicante also
emerged (Ministerio de Vivienda, 2006). This led to the appearance of new patterns of urban growth, which were considerably
∗ Corresponding author. Tel.: +34 958 24 04 47; fax: +34 958 24 89 90.
E-mail addresses: [email protected] (F. Aguilera), [email protected]
(L.M. Valenzuela), [email protected] (A. Botequilha-Leitão).
0169-2046/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.landurbplan.2010.10.004
more dispersed (Muñiz et al., 2008). These processes resulted in
the fragmentation and occupation of the rural landscape (EEA,
2006), an increase in population mobility and associated infrastructures (Newman and Kenworthy, 1999), and also the expansion
of the urban structure of these metropolitan areas (Dematteis,
1998).
A major concern in spatial planning in the Iberian Peninsula
is the study of the spatial characteristics of urban processes, such
as urban sprawl (Botequilha-Leitão, 2001; Ministerio de Vivienda,
2006; Dalda et al., 2006). This is also true in Europe (Antrop, 2000
Galster et al., 2001; EEA, 2006; Kasanko et al., 2006; Petrov et al.,
2009) as well as in the United States (Alberti, 1999; Torrens and
Alberti, 2000; Song and Knaap, 2004).
Landscape ecology focuses on the analysis of landscape structure, the spatial implications of ecological processes in these
landscapes and the changes that occur in them (Forman and
Godron, 1986; Forman, 1995; Botequilha-Leitão et al., 2006; DiBari,
2007). Because of their evident spatial dimension, landscape ecology and spatial planning converge in a common workspace (Antrop,
2001). Consequently, landscape ecology has been increasingly
employed in spatial planning (Freemark et al., 1996; Bettini et al.,
2001; Botequilha-Leitão and Ahern, 2002; Steinitz et al., 2003;
Corry and Nassauer, 2005; Botequilha-Leitão et al., 2006; Kim and
Ellis, 2009). In fact, in the last 10 years, it has been increasingly used
to study the spatial characteristics of urban processes (Herold et al.,
2003, 2005; Berling-Wolf and Wu, 2004, inter alia), namely the spa-
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
227
Fig. 1. Allocation of the study area: the region of Andalusia (A); the Province of Granada (B); the Metropolitan Area of Granada (C).
tial characteristics of urban patches, including their size, shape, and
spatial distribution.
Many spatial landscape properties can be quantified by using
a set of metrics (McGarigal et al., 2002; Li and Wu, 2004; Uuemaa
et al., 2009). Herold et al. (2003, 2005) and Seto and Fragkias (2005)
inter alia, use the term spatial metrics to more clearly differentiate
these metrics from landscape metrics. Spatial metrics characterize
urban form as such (Herold et al., 2003), whereas in ecological
landscape studies, landscape metrics are explicitly linked to
ecological functions (Gustafson, 1998; Luck and Wu, 2002; DiBari,
2007). In this context, spatial metrics can be a very valuable tool
for planners who need to better understand and more accurately
characterize urban processes and their consequences (Herold
et al., 2005; DiBari, 2007; Kim and Ellis, 2009). Relevant examples
include the following: (i) the replacement of traditional urbanization and growth patterns typical of the Mediterranean city with
other more global patterns (Dematteis, 1998); (ii) modifications
in the life style of the population and an increased automobile
dependence (Newman and Kenworthy, 1999); (ii) changes in the
periurban landscape (Antrop, 2000); (iii) the evaluation of future
planning scenarios (Van Beusekom, 2003; Franco et al., 2005); (iv)
the validation and optimization of the results of simulation models
(Berling-Wolf and Wu, 2004; Li et al., 2008).
Aguilera (2008) reviews a large body of research focusing on
the analysis of the spatial characteristics of metropolitan growth
(Geoghegan et al., 1997; Alberti and Waddell, 2000; Berling-Wolf
and Wu, 2004; Alberti and Marzluff, 2004; Herold et al., 2005;
Li et al., 2008), and highlights the value of spatial metrics in
the study of urban landscapes. According to Seto and Fragkias
(2005) most of these studies focus on cities in the USA. Thus
there is clearly a need to explore the application of spatial metrics to European cities. This is particularly true for Mediterranean
cities because of their unique morphological characteristics and
because of the changes that they are experiencing in the last years
(Dematteis, 1998).
Within this context, our research hypothesis is that spatial
metrics are useful for territorial planning, particularly for the
quantification of the spatial characteristics of urban growth in
metropolitan areas in southern Europe. For this purpose, we simulated three explorative scenarios (Borjerson et al., 2006) for the
Metropolitan Area of Granada (Spain) from the year 2004. Each
scenario reflects the potential evolution of urban growth in 2020,
based on the growth in 1985–2000, analyzed in previous work. Our
study uses spatial metrics to quantify spatial processes of urban
growth for each scenario. These metrics allowed us to determine
the spatial characteristics and patterns generated by the simulated
urban processes.
Accordingly, Section 2 of this article describes the Metropolitan
Area of Granada (MAG) (2.1), the cartography of land occupation
in 2004 (2.2), the design (2.3) and simulation of scenarios (2.4),
selected spatial metrics (2.5), the spatial characteristics of urban
growth processes (2.6), and the urban occupation patterns identified with the metrics (2.7). Section 3 presents the results of the
identification of occupation patterns in the three future scenarios.
Finally, Section 4 discusses the results obtained, and Section 5 gives
the conclusions derived from the study.
228
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
Fig. 2. Map of land use categories in the Metropolitan Area of Granada (MAG).
2. Data and methods
2.2. Urban land use maps for the Metropolitan Area of Granada
2.1. Study area
For our study, we generated a map of urban land uses, which
was based on the photo-interpretation of existing aerial orthophotographs on a scale of 1:10,000 for the region of Andalusia,
published by the Cartographic Institute of Andalusia in 2004. This
interpretation process and subsequent digitization allowed us to
create a map (Fig. 2) classified in terms of the four general categories of urban land use described below (Fig. 3). Fig. 2 also includes
the main public road systems as well as the principal water courses.
The rest of the non-urban areas (farmlands, forests, etc.) are all classified in the category of “matrix”. This map was rasterized with a
cell size of 50 m × 50 m in consonance with the size of the Minimum
Mappable Unit (MMU).
The Metropolitan Area of Granada (MAG) (Fig. 1) with a surface area of 859.34 km2 has 32 municipalities (Consejería de Obras
Públicas, 1999). With a population of over 500,000 inhabitants
in 2009, it is regarded as a medium-sized metropolitan area in
comparison to other cities in Spain (such as Madrid or Barcelona)
(Feria, 2004). In the last 20 years, Granada has experienced a
series of transformations that have directly affected urban land
use (Aguilera, 2008). The rural landscape surrounding Granada
has a very high landscape value (Menor, 2000). However, it is
currently experiencing a significant transformation process that
has led to the formation of the current metropolitan area. In this
context, a metropolitan land use plan was published in 1999,
namely, the Territorial Plan for the MAG (POTAUG) (Consejería de
Obras Públicas y Transportes, 1999). However there is neither a
specific institutional body at a metropolitan level nor adequate
management instruments. As a result, the POTAUG lacks of regulation capacity with a series of consequences briefly described
below.
The transformation of the MAG is primarily the result of the
expansion of rural population nuclei and the increase of urban
sprawl. The resulting fragmentation and isolation of farmlands has
modified the traditional agricultural system. In a parallel way, this
transformation has significantly changed the mobility habits of
the population and increased automobile dependence (Consejería
de Obras Públicas y Transportes, 1999; Menor, 2000). Additionally, the POTAUG has not been able to prevent the abandonment
and loss of productive farmland, because of urban occupation.
For this reason, the agricultural sector is facing a very uncertain future. There is thus the evident need to explore potential
patterns of the evolution of urban growth and their spatial
characteristics.
2.2.1. High density residential
In these areas, there is a predominance of residential housing
structures which is typical of compact cities in southern Europe.
These buildings are mostly found in the city center though there
are new housing developments in urban nuclei in the metropolitan ring. This type of area has a density of circa 70 households per
hectare (h/ha). The city center of Granada is a good example of this
category (see Fig. 3A).
2.2.2. Low-density residential
These areas are characterized by single-family houses, which
are either traditional village houses (e.g. Santa Fé) or new housing
developments (e.g. Monte Luz, see Fig. 3B). This type of area has a
density of circa 25 (h/ha).
2.2.3. Industrial
These areas include not only industrial complexes per se, but
also logistics centers, i.e. storage warehouses and distribution facilities (e.g. industrial areas in Road N-323, see Fig. 3C). They are
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
229
Fig. 3. (A–D) General categories of urban land use in the MAG.
generally found on the outskirts of the city, and are located at main
highway intersections in the metropolitan area.
2.2.4. Commercial
These areas are where large shopping centers and recreational
areas are located (Armilla node, see Fig. 3D). They are usually found
in the vicinity of important nodes of the metropolitan road network.
2.3. Design of future scenarios
Based on the data pertaining to land occupation in the MAG
in 2004 and in line with previous research (Aguilera, 2006, 2008;
Valenzuela et al., 2008), we created three external explorative
scenarios (Dreborg, 2004; Borjerson et al., 2006), for 2020. These
scenarios were based on different shapes and patterns of urban
growth in the MAG over the last 30 years. These shapes and patterns
were combined in scenarios with a view to exploring the possible
future evolution of this metropolitan area. It should be underlined,
however, that the selected type of scenarios are not more or less
accurate predictions of the future (Berdoulay, 2009), but rather different evolution possibilities of the metropolitan area. The three
scenarios are: (S1) residential intense development; (S2) industrial
and commercial specialization; (S3) compact development.
Each scenario emphasizes one of the most important growth
tendencies for this area: dispersed residential growth, compact
growth of the traditional city, industrial and commercial developments, etc. Moreover, each shows different urban growth
intensities, compared with the growth rate in 1985–2000
(80 ha/year). This includes lower growth rates (S3), higher growth
rates (S1), and a growth rate equal to that of recent years (S2).
This calculation was based on the variation of certain urban growth
drivers for the study area, such as economic context and the regulation capacity of the territorial plan.
(S1) Residential intense development. This scenario depicts the
potential intensification of urban growth caused by increased
economic activity. The result is the development of new residential
areas stemming from a metropolitan plan that has a weak regulation capacity. It shows an increase of 30% in the annual growth rate
recorded in 1985–2000, particularly in regard to residential land
use. It is characterized by an increase of urban sprawl in residential areas. In a parallel way, industrial areas also tend to develop
in the neighborhood of existing industrial areas. Commercial land
use experiences a similar increase. Finally, we also included new
road networks in project, since they will eventually affect growth
patterns in this scenario.
(S2) Industrial and commercial specialization. This scenario
explores a stabilization of the economic activity registered between
1985 and 2000, and specifically directed at the creation of new
industrial and commercial activity that will be enhanced in
metropolitan land use plans. It is characterized by a growth rate
similar to that of previous years, a significant increase in new
technological/commercial/industrial areas in the vicinity of major
thoroughfares, and a slight increase in residential housing developments. In this scenario, technology parks and commercial areas
become new centers of attraction for high-density residential zones
that are farther away from the MAG core area. Low-density residential zones continue to become more aggregated. In the same
ways as S1, we also included new road network projects in this
scenario.
(S3) Compact development. This scenario presupposes a lesser
favorable economic context but favors the development of new
industrial and commercial areas. It also contemplates a more
restrictive metropolitan land use plan, especially in relation to residential growth. This reflects a decrease in urban land use growth
rate (−30%), particularly affecting low-density residential areas.
It also means that commercial and industrial land uses tend to
flourish in the areas surrounding road network nodes that provide
greater accessibility. New road network projects were not included
because in our opinion, they are not relevant to this type of urban
development.
230
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
residential, commercial or industrial) in that specific cell. This transition potential was obtained by combining four parameters, which
were all transformed on a scale from 0 to 1.
Neighborhood (N): This parameter estimates the probability of
change for each cell in the raster input layer, depending on existing
neighboring urban land uses and the distance to the central cell.
Based on the review by Santé et al. (2010), we defined a neighborhood formed by a square window of 11 × 11 grid cells of a total of
121, i.e. 550 m by 550 m. Each of the 121 cells in the neighborhood
exerts over the central cell an effect or attraction or repulsion for
different urban uses. This effect depends on the type of urban land
use existent in each of the neighboring cells. For example, industrial uses generally repel residential uses. Additionally, the closer
to the central cell they are, the higher the repulsion effect. These
attraction or repulsion values should be calculated for the study
area by trial and error (White et al., 1997). A cell with the highest
value is assigned with 1, and those with the lowest are assigned
with 0. The other intermediate values are linearly transformed to a
value between 0 a 1.
Suitability (S) is composed of two raster layers: slope map and
urbanizable areas. The slope map was derived from the Digital
Elevation Model for the area, and was transformed by a linear transformation in values ranging from 0 to 1 with maximum values for
the zones of minimum slope. The map of urbanizable areas was
obtained from the areas classified as urbanizable in the metropolitan land use plan (POTAUG). Urbanizable areas received a value of
1 and non-urbanizable areas, a value of 0.
Accessibility (A): This parameter is defined as the nearest Euclidian distance to the road network. Areas contiguous to the road
network have a value of 1, and those farther away have a value
of 0. For the rest, a linear transformation function is used.
Stochastic (V): The objective of this parameter is to incorporate
a randomness component that is typical of urban spatial processes
(Batty and Xie, 1997). It is obtained with Eq. (1):
V = 1 + (− ln(rand))
˛
(1)
where rand is a random number between 0 and 1, and ˛ is a
parameter that permits an adjustment of the degree of perturbation (dispersion). The value of ˛ (0.3) was computed, based on the
radial dimension (Barredo et al., 2003) and validated to better simulate the urban growth patterns identified for the MAG between
1985 and 2000.
The transition potential for each urban use (Pj ) was finally
obtained by combining these parameters with Eq. (2):
Pj = Nj × Sj × Aj × Vj
Fig. 4. Simulations (S1–S3) generated for future scenarios in the year 2020 to be
evaluated with spatial metrics (Aguilera, 2008).
2.4. Simulation of future scenarios with a model based on cellular
automata
Based on the rules described in Section 2.3, we simulated the
three types of scenario with a model based on cellular automata
(CA) (Fig. 4 ). This model was developed and implemented with
IDRISI Andes software (Clark Labs, 2003). It is based on theoretical
premises proposed by White et al. (1997), which are the basis for
other simulation models (Barredo et al., 2003; Petrov et al., 2009).
To calculate the localization of urban growth areas (the quantity of simulated urban growth should be established by the user),
the model employs an urban land use map (Fig. 2 in raster format
as input data). For each cell in the input layer, the model obtains a
transition potential (P) that represents the possibility of the appearance of a new urban land use (high-density residential, low-density
(2)
This model has been previously used for the simulation of the
dynamics of past urban growth, more specifically for 1985–2000
(Valenzuela et al., 2008). This necessarily entailed a calibration process, especially of the attraction–repulsion values used to calculate
the neighborhood. The results of the simulations were compared to
2000 urban land use map using a cross tabulation which originated
an overall kappa of 0.74 (Aguilera, 2006; Valenzuela et al., 2008).
However, for the simulation of the three scenarios, the parameters of the model had to be modified, especially the neighborhood
parameter and its rules of attraction and repulsion. However,
as shown in Table 1, other parameters were also modified. The
objective was to simulate the spatial processes described for each
parameter in Section 2.3. The knowledge extracted from the simulation of past dynamics (1985–2000) described above was critical
to simulate the future scenarios.
2.5. Set of spatial metrics
For the evaluation of the changes in land use patterns in the three
simulated scenarios, a small set of spatial metrics was selected.
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
231
Table 1
Implementation of rules for scenarios in the parameters of the CA-based model. The suitability parameter remained constant in all three scenarios.
Scenarios
Adaptation of the rules for scenarios in the parameters of the CA-based model
Residential intense development (S1)
Neighborhood: The attraction strength of residential land uses for more residential land uses diminishes. Beyond a
certain distance, this force is transformed into repulsion, which favors urban dispersion.
Accessibility: High-capacity public highway projects are included.
Stochastic parameter: The weight of the randomness factor (˛) is increased for residential land uses in order to reflect
new dispersed urban growth in areas at a distance from existing urban zones.
Industrial and commercial
specialization (S2)
Neighborhood: There is an increase in the attraction strength of public roadways on commercial and industrial land
uses. The attraction strength of residential land uses on themselves also increases, which favors aggregation.
Accessibility: High-capacity public highway projects are included. Their weight increases for industrial and
commercial land uses.
Stochastic parameter: The stochastic parameter here is similar to urban growth simulations of the past.
Compact development (S3)
Neighborhood: The attraction strength of residential land uses for more residential land uses increases, which favors
aggregation. The attraction strength of the nodes increases industrial and commercial land uses.
Accessibility: High-capacity public highway projects are not included.
Stochastic parameter: The stochastic parameter is reduced to show a less dispersed growth, especially in residential
land uses.
Since our objective was to estimate the spatial characteristics of the
different urban land uses as a whole, the metrics were computed at
the class level (McGarigal et al., 2002), such that a single value was
obtained for each of the various classes of urban land uses. It was
rather difficult to select a reduced set of metrics since there is a wide
range of metrics to choose from (O’Neill et al., 1988; BotequilhaLeitão and Ahern, 2002; McGarigal et al., 2002; Botequilha-Leitão
et al., 2006; DiBari, 2007). In fact, many of them were interrelated
since they are derived from the same primary measurements of
the patches (area, patch number, edges, etc.) (Riitters et al., 1995).
Additionally, the selection of metrics invariably depends on the
objectives of the research.
For this purpose, we chose six metrics (Table 2), based on
the core set of 10 metrics proposed by Botequilha-Leitão et al.
(2006) (for a full description of these metrics, see McGarigal et al.,
2002; for a description of their application in spatial planning,
see Botequilha-Leitão et al., 2006). This core set was specifically
selected to support spatial planning processes.
Thus, it was found appropriate to use the subset of those metrics
in our research. It is our assertion that these metrics can successfully
characterize various spatial processes affecting urban land uses (see
following sections).
2.6. Urban spatial processes
The three urban spatial processes identified in the MAG are
described below.
2.6.1. Aggregation
Aggregation corresponds to the clustering of patches to form
patches of a larger size. This evidently reduces the total number
of patches (NP), thus producing an increase in their mean area
(AREA MN). The aggregation process was estimated by means of
a joint interpretation of the evolution of the NP and AREA MN values, along with the PLAND, all of which have been used in a wide
range of research studies (e.g. Mateucci and Silva, 2005).
2.6.2. Compaction
Compaction involves the formation of rounded patches in a circular shape that makes them more compact. The inverse process is
Table 2
Selected set of metrics for the analysis of urban growth patterns. The selected set of metrics is in consonance with the definition of McGarigal et al. (2002). See also
Botequilha-Leitão et al. (2006) for a discussion on metrics applications and planning.
Spatial metric
Abrrev.
Description
Formulae
Percent of landscape
PLAND
Measures the percent of the landscape
PLAND = Pi =
Number of Patches
NP
Identifies the number of patches in
each of the urban land uses
NP = ni
Mean patch size
AREA MN
Calculates the average mean surface of
patches
AREA MN =
Mean radius of gyration
GYRATE MN
Mean of the GYRATE. GYRATE equals
the mean distance (m) between each
cell in the patch and the patch centroid
MGYRATE =
n
aij
j=1
A
n
× 100
Pi = proportion of the landscape
occupied by patch type (class) i;
aij = area (m2 ) of patch ij; A = total
landscape area (m2 )
ni = number of patches in the
landscape of patch type (class) i
hij
j=1
ni
z
h
ijr
hijr = distance (m) between cell ijr
[located within patch ij] and the
centroid of patch ij (the average
location), based on cell center-to-cell
center distance; z = number of cells in
patch ij
z
r=1
n
Mean shape index
SHAPE MN
Measures the ratio between the
perimeter of a patch and the perimeter
of the simplest patch in the same area
SHAPE MN =
Mean Euclidean distance
neighbor
ENN MN
Measures average distance between
two patches in a landscape
ENN MN =
j=1
pij / min pij
ni
n
j=1
ni
hij
pij = perimeter of patch ij in terms of
number of cell surfaces;
Min pij = minimum perimeter of patch
ij in terms of number of cell surfaces
hij = distance (m) from patch ij to
nearest neighboring patch of the same
type (class), based on patch
edge-to-edge distance, computed from
cell center to cell center
232
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
Fig. 5. Land use patterns in the industrial and commercial specialization scenario (S2) as compared to the situation in 2004 for (A) industrial; (B) high-density residential;
(C) commercial. Arrows 1 and 2 show industrial uses in 2004; arrows 3 and 4 depict the linear growth process in the simulated scenario; arrows 5 and 6 show high density
residential for 2004; and arrows 7, 8 and 9 show the simulated growths. Finally, arrows 10 and 11 show new commercial areas around the road network nodes in the
simulated scenario.
elongation, in which the shape of the patches becomes more elongated. This process is measured with SHAPE MN and GYRATE MN.
Lower metric values are indicative of compaction, whereas higher
metric values indicate elongation.
2.6.3. Dispersion/isolation
This process involves an increase in the distance that separates
patches of the same land use type. It is measured by ENN MN,
such that an increase reflects greater isolation. However, in contrast to classic landscape ecology studies, in the analysis of urban
processes, greater isolation indicates greater dispersion. The reason for this lies in the fact that when new patches appear within
existent patches, e.g. due to urban sprawl, the distance separating
them decreases (see Fig. 5C, Section 3.2, for examples of the study
area).
Nevertheless, other spatial metrics could have been selected
to measure the three previously described spatial processes. For
example, the Mean Proximity Index could have been used to estimate the dispersion process. However, its dependence on the patch
size and the need to establish a neighborhood radius was the reason
why we decided against using it. Another possibility would have
been to use only one metric for the characterization of the compaction process. However, the lower sensitivity of the SHAPE MN
and the dependence on patch size of the GYRATE MN were factors
that led to our decision to use both metrics instead of only one.
2.7. Urban growth patterns
We established relationships between these spatial characteristics and processes described above and distinct urban growth
patterns. These patterns were characterized in terms of form as
well as type of land use (Galster et al., 2001; Song and Knaap, 2004).
Based on the characteristics of urban land use found in European
metropolitan areas (Font, 2004; Indovina, 2005), we defined four
urban patterns.
Aggregated pattern: This pattern corresponds to the conventional type of urban growth in Mediterranean cities: new urban
areas are added onto an already consolidated city (Dematteis, 1998;
Indovina, 2005). It is characterized by increased aggregation and
generally by reduced dispersion. This urban growth pattern tends
either to remain constant or to decrease landscape fragmentation.
Linear pattern: This pattern refers to urban growth around road
networks, preferably industrial or mixed industrial land use (Font,
2004). In this type of pattern we were able to identify two main
processes: (i) increased or stable aggregation; (ii) decreased compaction.
Leapfrogging: This pattern reflects the appearance of urban
patches with a principally residential function. It is characterized
by a predominance of low-density dispersed single-family houses,
and is dominated by the following processes: (i) decreased aggregation; (ii) decreased elongation because of the formation of more
or less rounded patches; (iii) increased dispersion.
Nodal pattern: This pattern largely reflects existing industrial
and commercial urban growth near the main transportation nodes.
It behaves in much the same way as the urbanization pattern
(decreased aggregation along with increased dispersion and compaction). However, it mainly affects industrial and commercial uses.
Table 3 shows the relation between the four patterns identified
and the spatial processes determined by the metrics (aggregation,
compaction, and dispersion). For example, urban growth in the
aggregated pattern shows greater aggregation and compaction. In
contrast, the leapfrogging growth pattern reflects less aggregation
and a greater dispersion of urban uses, whereas linear patterns
show lower compaction values. PLAND values are not included in
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
233
Table 3
Interpretation of the three spatial processes and change in the spatial metrics values in relation to four different patterns of urban growth identified for the MAG. Spatial
processes: Aggregation (Aggre), Compaction (Comp) and Dispersion/Isolation (Disp): “+” Increase, “−” Decrease. Metric values – “↑” Increase, “↓” Decrease, ( ) Possible
increase/decrease.
Spatial metrics
Four urban patterns and three spatial processes associated
Aggregative
Aggre
−
NP
AREA MN
GYRATE MN
SHAPE MN
ENN MN
Linear
Comp
=/−
Disp
=/−
↓
↑
Aggre
=/−
Leapfrogging
Comp
−−
Disp
=/−
Aggre
−
Comp
−
Disp
+
↑
(↓)
↑
↑
↑
=/↓
=/↓
Nodal
=/↓
this table because they are not directly related to any pattern, but
rather are jointly interpreted with the rest of the metrics.
3. Results
After calculating the metric values for the three future scenarios,
we were able to trace the evolution of the metrics for each urban
land use category with respect to their value in 2004. The changes
in the metrics reflected the modification in each land use pattern in
terms of its aggregation, compaction and dispersion. Table 4 shows
the changes in each land use in terms of the metrics values and
the related processes that were identified. The following sections
provide an interpretation of the results for each scenario.
3.1. Residential intense development scenario (S1)
The S1 scenario reflects relevant changes in patterns of urban
land use (Table 4). Regarding industrial and commercial uses, there
was significant growth (PLAND + 54% and 90%, respectively). These
land uses have a more aggregative pattern, resulting from a stronger
aggregation process (−11% NP and +74 AREA MN for industrial
areas; stability in NP and +91% in AREA MN for commercial areas),
due to the clustering of patches in their neighborhood.
Residential land uses with moderate or significant growth (+21%
in PLAND for low-density and +65% for high-density) exhibit
a totally different behavior. High-density residential uses show
increased dispersion and less aggregation as a consequence of the
appearance of new patches. This is reflected in an increase in NP
(+139%), and a moderate decrease in patch size (−35% AREA MN)
and in mean distance to the nearest patch (−52% ENN MN). In contrast, low-density residential seems to show a more aggregative
pattern, which corresponds to a larger patch size (+18% AREA MN).
3.2. Industrial and commercial specialization scenario (S2)
This scenario reflects an important change in the spatial metrics, which suggests modifications in land use patterns, especially
for industrial land use. The overall behavior of the metrics seems to
indicate that its growth (+65% PLAND) is reflected in more elongated urban patches, which develop as follows. Some of these
patches come together to form narrow ribbons along the main
transportation networks. This is indicative of a tendency to linear
patterns, as reflected in decreased compaction (+5% in SHAPE MN
and +15% in GYRATE MN), combined with an increase in AREA MN
(+26%) as a consequence of the coalescence of multiple patches
forming longer ribbons (see Fig. 5A).
In contrast, residential land uses also experienced slight or
moderate growth (+20% in PLAND for low density residential and
+10% in PLAND for high density), but their behavior was precisely
the opposite. On the one hand, high-density residential land use
showed a substantially lower level of aggregation (+144% in NP
and −51% in AREA MN) and higher dispersion (−46% in ENN MN).
Comp
−
Disp
+
↑
(↓)
(↓)
(↓)
↓
Aggre
+
(↓)
(↓)
↑
↑
Despite this higher level of dispersion, it is worthwhile noting that
their growth occurred adjacent to patches of other land uses (see
Fig. 5B). On the other hand, low-density residential land use became
more aggregated (+14% in AREA MN and a −4% NP).
Finally the commercial land use increased the most (PLAND
+332%). It showed lesser aggregation, as can be observed in the
pronounced rise in the NP (+332%). There was also greater dispersion, given the reduction in the ENN MN (−43%), which reflects a
much more distributed (and thus more dispersed) commercial land
use throughout the territory.
3.3. Compact development scenario (S3)
The S3 scenario showed the least change in land use shapes or
patterns. Regarding both high- and low-density residential zones,
its evident stability is reflected in the slight changes in the metrics (see Table 4). In contrast, commercial and industrial land uses
showed significant growth (+113% and 26% PLAND, respectively)
with a change in pattern, as can be deduced from the decreases
for both land uses in ENN MN (−17% and −8%, respectively) and
from the increases in NP (+117% and 7%, respectively). In the case
of industrial land use, these changes were accompanied by an
increase in AREA MN (+18%). Overall, these changes can be interpreted as a tendency to greater dispersion, possibly associated with
an increased nodality. Fig. 6 shows how these changes originate as
a consequence of the increase in the nodality of both land uses.
4. Discussion
Based on the interpretation of spatial processes such as aggregation, compaction, and dispersion, we were able to use spatial
metrics to identify changes in land use patterns in the three future
scenarios simulated. In order to accomplish this, we based our work
on the correspondence between the changes in the metrics and
the spatial processes in the landscape structure (Table 3). We were
thus able to satisfactorily estimate aggregation by using the NP,
AREA MN and PLAND (Seto and Fragkias, 2005; Botequilha-Leitão
et al., 2006), even though the applicability of other metrics to urban
areas had been previously explored, such as the ENN MN (DiBari,
2007), PLAND (Alberti and Marzluff, 2004; Herold et al., 2005),
and SHAPE MN Index (Mateucci and Silva, 2005). On the basis of
this correspondence, we identified changes in land use patterns,
supported by the spatial processes identified by the metrics (see
Table 3).
The scale of analysis is an aspect of particular relevance in issues
pertaining to landscape ecology (Turner, 2005). In this study, the
scale used was the entire metropolitan area. This identification of
patterns was a simple, global way of enriching conventional analyses of land use changes (Herzog and Lausch, 2001) on a metropolitan scale. Furthermore, it provides urban planners with another
instrument that they can use to evaluate different proposals
234
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
Table 4
Interpretation of pattern changes in the different urban land uses as reflected by changes in spatial metrics values.
Land use
Low-density residential
High-density
residential
Residential intense
development (S1)
Industrial
Commercial
Low-density residential
High-density
residential
Industrial and
commercial
specialization (S2)
Industrial
Commercial
Low-density residential
High-density
residential
Compact development
(S3)
Industrial
Commercial
Metric
% Change
Spatial changes
PLAND
NP 516
AREA MN 4.51
GYRATE MN 60.39
SHAPE MN 1.25
ENN MN 200
PLAND
NP 323
AREA MN 4.88
GYRATE MN 60.53
SHAPE MN 1.27
ENN MN 152
PLAND
NP 354
AREA MN 4.74
GYRATE MN 58.08
SHAPE MN 1.17
ENN MN
PLAND
NP 18
AREA MN 4.26
GYRATE MN 64.87
SHAPE MN 1.17
ENN MN 879
+21%
+3%
+18%
+6%
+1%
−5%
+65%
+139%
−35%
22%
5%
−52%
+54%
−11%
+74%
+12%
Stable
+4%
+90%
Stable
+91%
+22%
+2%
−2%
Moderate increase in the land use with larger
size patches (AREA MN) and a slight decrease
in the ENN MN. There is a slight aggregation, as
reflected in the increase in the AREA MN.
PLAND
NP 483
AREA MN 4.36
GYRATE MN 57.53
SHAPE MN 1.26
ENN MN
PLAND
NP 329
AREA MN 3.65
GYRATE MN 52.17
SHAPE MN 1.23
ENN MN
PLAND
NP 526
AREA MN 3.42
GYRATE MN 59.45
SHAPE MN 1.23
ENN MN
PLAND
NP 63
AREA MN 2.76
GYRATE MN 52.61
SHAPE MN 1.13
ENN MN 514
+20%
−4%
+14%
−1%
+1%
+1%
+10%
+144%
−51%
−33%
−9%
−46%
+66%
+32%
+26%
+15%
+5%
−21%
+332%
+250%
+26%
−1%
−1%
−43%
A moderate growth of this land use with a
slight increase in aggregation, as revealed by
the slight decrease in NP and the increase in
AREA MN. Small low-density urban patches
cluster together.
PLAND
NP
AREA MN 3.62
GYRATE MN 54.99
SHAPE MN 1.23
ENN MN 204
+3%
+8%
−5%
−3%
−1%
−3%
For a small growth in residential land use
(PLAND = 3%), there is a slight decrease in
aggregation, which is reflected in the increase
in the number of patches (+8% NP) and a slight
decrease in mean patch size (−5% AREA MN).
PLAND
NP 136
AREA MN 7.46
GYRATE MN 77.52
SHAPE MN 1.34
ENN MN
PLAND
NP 428
AREA MN 3,21
GYRATE MN 55.52
SHAPE MN 1.19
ENN MN 225
PLAND
NP 39
AREA MN 2.20
GYRATE MN 52.11
SHAPE MN 1.18
ENN MN
Stable
+1%
−0.5%
+1%
0%
−2%
+26%
+7%
+18%
+7%
+2%
−8%
+113%
+117%
−1.5%
−2%
+3%
−17%
Changes in metrics are very small (2% at the
most) with only a very slight increase in this
land use (PLAND). For this reason, the spatial
characteristics of this use remain the same.
Large increase in this land use with less
aggregation, which is evidenced in the striking
increase in NP and decrease in AREA MN. There
is greater dispersion with the decrease in the
ENN MN since it is a land use that was
extremely aggregative.
Important growth in industrial land use based
on greater aggregation, as reflected in the
increase of the AREA MN and a drop in the NP.
ENN MN increases. Larger size patches are
generated because of the aggregation of other
existing patches.
Significant growth based on greater
aggregation as reflected in the increase in
mean patch size while the other spatial metrics
values remain stable.
Slight growth of this land use with a very
important growth in dispersion (very high
increase in NP and significant decrease in
AREA MN). Moreover, there is a drop in the
ENN MN, which suggest greater dispersion.
Large increase in industrial land use and all the
metrics except for SHAPE MN (only a slight
increase) and a decrease in the MEEN. There is
greater dispersion of the land use with more
elongated patch shapes.
A huge increase in commercial land use (332%
PLAND), combined with increases in NP and
AREA MN, and a wider distribution of the land
use in the territory (decrease in ENN MN). This
can be interpreted as the multiple appearance
of new urban patches in this land use.
For a moderate growth of industrial land use
(PLAND), there is a slightly higher (NP) of
larger size (AREA MN). In addition, there is a
wider distribution in the area of study (slight
decrease in ENN MN).
A significant increase in land use occupancy
(PLAND), which also implies an important
increase in NP and a moderate decrease in
ENN MN. It reflects a greater dispersion and
distribution of this land use.
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
235
Fig. 6. Nodal pattern for industrial and commercial land uses in the compact development scenario (S3). Both residential land uses patterns are not changed (arrows 3 and 4);
arrows 1 and 2 show new industrial patches around the road network nodes; arrow 5 shows new commercial patches also around nodes; finally arrow 6 shows commercial
patches existent in 2004.
and alternatives for future developments and scenarios (Prato,
2007) in both existing and future metropolitan land use plans.
Nevertheless, it should be highlighted that this general analysis could be complemented by other more detailed studies, such as
those that use spatial metrics along transects or concentric rings
(Luck and Wu, 2002) to better localize changes in land occupation
patterns. This would make it possible to discover and locate pattern changes in the vicinity of large highway networks or in the
fringe of principal urban areas. Such analyses would also permit the
use of more detailed scales which could better identify small-scale
dispersion processes, such as the urban growth of isolated houses,
which are not detected by the adopted cell-size (50 m × 50 m). They
would thus be a valid approximation that would complement a
more general analysis, such as the one described in this study.
Regarding the use of metrics to identify pattern changes, in certain cases, we were obliged to visually identify the maps obtained
in the simulations in order to be able to classify more complex spatial behaviors. For instance, the leapfrogging pattern and the nodal
pattern are not easily distinguished from each other by metrics
alone. Both are characterized by relatively small patches, which
are widely separated from each other. However, the nodal pattern
emerges around public thoroughfare nodes, whereas the urbanization pattern can occur anywhere. Since the metrics do not have
the power to discriminate these situations, visual interpretation is
required. The behavior of residential and commercial land uses in
the S1scenario is a clear example of this. We thus evaluated their
location in the context of thoroughfare nodes in order to classify
them as nodal.
Equally problematic was the growth of high-density residential
land uses in the S2 and S1 scenarios. Both cases showed an increase
in NP, a decrease in AREA MN, and a decrease in ENN MN. A large
portion of high-density residential growth was adjacent to certain
urban nuclei of low-density residential (Figs. 5B and 7). Consequently, this resulted in a combined growth pattern that aggregated
high-density and low-density residential zones. This made certain
urban nuclei considerably denser though the high-density land use
class in itself is more dispersed and fragmented (see Fig. 7).
Nevertheless, other situations offered fewer doubts. This is the
case of scenario S3, especially for residential uses (Fig. 8). In this scenario, the stability of the metrics corresponds to the permanence of
the spatial characteristics analyzed, and thus, to the preservation
of the existing land use pattern. Consequently, by combining the
evaluation of metrics with a visual interpretation of spatial characteristics, we were able to characterize the changes in the urban
growth patterns in future scenarios on a metropolitan scale.
Regarding the results of the evaluation of pattern changes in the
simulated scenarios, in scenario S3, we found that urban occupation patterns were stable. They did not produce a significant impact
since urban dispersion did not increase. In contrast, there was a
more pronounced transformation of urban growth patterns in the
other two scenarios. In the S1 scenario, there was a more dramatic
transformation of residential land uses, which are the predominant urban uses in the MAG. The S2 scenario showed a greater
transformation of commercial and industrial land uses. Despite the
fact that the S1 and S2 scenarios reflect a greater degree of dispersion than the S3 scenario, the S1 scenario has a more dispersed
growth, due to the greater intensity of urban growth and the predominance of leapfrogging patterns, especially in residential areas.
This higher dispersion can be interpreted regarding many environmental issues. For example the appearance of new urban nuclei
within the rural landscape matrix results in an increase of the
interface between those land uses. Potential disturbances
236
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
Fig. 7. New high-density residential patches in residential intense development scenario (S1). New patches produce lesser fragmentation in this land use, but which
correspond to aggregative growth with pre-existing residential zones. Arrow 1 shows existent high-density residential patches in 2004. Arrows 2 and 3 show new patches
in the simulated scenario.
Fig. 8. Stability of residential growth patterns in the compact development scenario (S3). This figure aims at depicting the stability of residential growth patterns. Arrows 1
and 2 show how the high-density residential patterns remain constant as observed in low-density residential (arrows 3 and 4); inversely arrow 5 shows unoccupied nodes
in 2004 and arrow 6 the same nodes occupied by industrial uses in the simulated scenario.
originates in urban areas are water pollution from source into the
hydrological system, noise pollution from cars, construction and
small industrial activities, habitat loss, etc. As a consequence, the
disturbance introduced by S1 scenario can be assumed as a higher
impact across the MAG landscape matrix.
However, despite the high level of dispersion in the S1 scenario, the linear growth patterns along the main roadways that
were identified in the S2 scenario (see Fig. 5A) can produce as
well an extremely negative impact, sharing with S1 most of the
described effects, with a special reference to the barrier effect
(Forman et al., 2003; Marull and Mallarach, 2005). The result of the
latter is the fragmentation of the rural landscape in the metropolitan area (Aguilera and Talavera, 2009).
5. Conclusions
Spatial metrics have been used in many research studies as a
valuable tool for the analysis, monitoring, and tracking of changes in
land use shapes and patterns. The approaches used range from the
mere description of urban landscape dynamics (Herold et al., 2003,
2005; Seto and Fragkias, 2005; Li et al., 2008) to their application
as an instrument for the comparison of scenarios of urban growth
(Alberti and Waddell, 2000; Berling-Wolf and Wu, 2004).
The application of the spatial metrics to a metropolitan area,
subject to an intense process of transformation and growth, such
as in Granada (Spain) has demonstrated its usefulness for the
quantification and interpretation of spatial growth characteristics
and patterns in urban environments. In our study, we simulated
three future scenarios that showed different spatial processes of
urban growth. These processes were quantified by spatial metrics,
which allowed us to measure the changes in urban occupation patterns associated with each scenario. Moreover, this quantification
provided us with a simple way of evaluating the scenarios by identifying those with a greater urban dispersion (S1) or fragmentation
of the adjacent rural landscape (S2), in contrast to those with a
greater compaction and stability in growth patterns (S3).
The results obtained illustrate the usefulness of spatial metrics for metropolitan land use planning. Spatial metrics can thus
be applied to monitor changes in urban growth patterns. Furthermore, they can also be used to evaluate the spatial consequences of
urban planning policies and future scenarios (Van Beusekom, 2003;
Franco et al., 2005; Aguilera, 2008), based on the characterization
of spatial processes, such as urban dispersion, aggregation, linear
growth, and their principal environmental consequences (Galster
et al., 2001; Muñiz et al., 2008). These consequences include the following: (i) automobile dependence associated with more dispersed
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
growth patterns (Newman and Kenworthy, 1999); (ii) a greater
consumption of land and energy associated with more dispersed
growth patterns in metropolitan areas (Muñiz et al., 2008); (iii)
the isolation of periurban rural spaces resulting from the decrease
in the connectivity of the landscape (Marull and Mallarach, 2005;
Aguilera and Talavera, 2009); (iv) impermeabilization processes in
aquifer recharge zones (CWP, 1998), etc.
With a view to exploring the connection between aspects
related to spatial processes and their environmental and territorial consequences, analyses such as the one described in this
article could be integrated in metropolitan observatories (Keiner
and Arley, 2007). This would make it possible to study the relation between the processes and spatial patterns identified by the
metrics as well as the changes in these aspects measured by available indicators (e.g. energy consumption, automobile dependence,
use of public transportation, alteration of environmental processes,
etc.). This could become an important future research line, in conjunction with studies on the selection of metrics that provide more
information concerning spatial characteristics, such as the principal
components analysis (PCA). We can therefore conclude that metrics are highly applicable to the study of urban landscape dynamics
and processes. This especially true for Southern European cities,
given the acceleration of their urban growth processes (Ministerio
de Vivienda, 2006) and the pressing need to characterize them.
References
Aguilera, F., 2006. Predicción del crecimiento urbano mediante SIG y modelos basados en AC (Predicting urban growth using GIS and CA-based models). Geofocus
6, 86–112 (in Spanish: English abstract).
Aguilera, F., 2008. Análisis espacial para la ordenación eco-paisajística de la Aglomeración Urbana de Granada (Spatial analysis for the ecological landscape
planning in the Metropolitan Area of Granada). PhD Thesis. University of
Granada, Granada (in Spanish: English abstract).
Aguilera, F., Talavera, R., 2009. Valoración de escenarios futuros a través de
la conectividad del Paisaje (Future scenarios assessment through landscape
connectivity). Observatorio Medioambiental 12, 17–36 (in Spanish, English
abstract).
Alberti, M., 1999. Urban patterns and environmental performance: what do we
know? J. Plann. Educ. Res. 1, 151–163.
Alberti, M., Marzluff, J., 2004. Ecological resilience in urban ecosystems: linking
urban patterns to human and ecological functions. Urban Ecosyst. 7, 241–265.
Alberti, M., Waddell, P., 2000. An integrated urban development and ecological
simulation model. Int. Assess. 1, 215–227.
Antrop, M., 2000. Changing patterns in the urbanized countryside of Western
Europe. Landsc. Ecol. 15, 257–270.
Antrop, M., 2001. The language of landscape ecologists and planners. A comparative
content analysis of concepts used in landscape ecology. Landsc. Urban Plann. 55,
163–173.
Barredo, J.I., Kasanko, M., Mccormick, N., Lavalle, C., 2003. Modelling dynamic spatial processes: simulation of urban future scenarios through cellular automata.
Landsc. Urban Plann. 64, 145–160.
Batty, M., Xie, Y., 1997. Possible urban automata. In: Besussi, E., Cecchini, A. (Eds.),
Artificial Worlds and Urban Studies. DAEST, Venezia, pp. 153–164.
Berdoulay, V., 2009. La historia de la geografía en el desafío de la prospectiva ([[nl]The
history of geography and the challenge of strategic foresight). Bol. Asoc. Geo. Esp.
51, 9–23 (in Spanish: English abstract).
Berling-Wolf, S., Wu, J., 2004. Modelling urban landscape dynamics: a case study in
Phoenix, USA. Urban Ecosyst. 7, 215–240.
Bettini, V., Gianoni, P., Di Noto, F., Stevanin, M., Zannin, D., 2001. Landscape ecology
e la teoria della percolazione in ecologia urbana: un’applicazione alla revisiones
di un PRG: il caso Pozzallo, Ragusa (Landscape ecology and percolation theory in urban ecology: an application in the Masterplan of Pozzallo, Italy). Arpa
Ecosistemi Urbani, Bologna (in Italian).
Borjerson, L., Hojer, M., Dreborg, K.H., Ekvall, T., Finnvenden, G., 2006. Scenario types
and techniques: towards a user’s guide. Futures 38, 723–739.
Botequilha-Leitão, A., 2001. Sustainable land planning: towards a planning framework. Exploring the role of spatial statistics as a planning tool. PhD Thesis.
Technical University of Lisbon, Lisbon.
Botequilha-Leitão, A., Ahern, J., 2002. Applying landscape concepts and metrics in
sustainable landscape planning. Landsc. Urban Plann. 59, 65–93.
Botequilha-Leitão, A., Miller, J., Ahern, J., Mcgarigal, K., 2006. Measuring Landscapes.
A Planner’s Handbook. Island Press, Washington, DC.
Center for Watershed Protection, 1998. Rapid Watershed Planning Handbook. A
Comprehensive Guide For Managing Urbanizing Watersheds. Prepared by the
CWP for the Office of Wetlands, Oceans and Watersheds and Region V. U.S. EPA,
Maryland, USA.
Clark Labs, 2003. Idrisi GIS Software Version Andes. Clark University, MA, USA.
237
Consejeríade Obras Públicas y Transportes, 1999. Plan de ordenación del territorio
de la aglomeración urbana de Granada (Metropolitan plan of the Urban Area of
Granada). Junta de Andalucía, Seville (in Spanish).
Corry, R.C., Nassauer, J., 2005. Limitations of using landscape pattern indices to evaluate the ecological consequences of alternative plans and designs. Landsc. Urban
Plann. 72, 265–280.
Dalda, J.A., Docampo, M.G., Harguindey, J.G., 2006. La ciudad difusa en Galicia (The
dispersed city in Galicia, Spain). Conselleria de Política territorial, Obras públicas
y transporte, Santiago de Compostela (in Galician).
Dematteis, G., 1998. Suburbanización y periurbanización. Ciudades anglosajonas y
ciudades latinas. In: Monclús, J. (Ed.), La ciudad dispersa. Suburbanización y
periferias. Centre de Cultura Contemporánea de Barcelona, Barcelona, pp. 17–33
(in Spanish).
DiBari, J., 2007. Evaluation of five landscape-level metrics for measuring the effects
of urbanization on landscape structure: the case of Tucson, Arizona, USA. Landsc.
Urban Plann. 79, 308–313.
Dreborg, K.H., 2004. Scenarios and structural uncertainty: explorations in the field
of sustainable transport. PhD Thesis. Royal Institute of Technology, Stockholm.
European Environment Agency (EEA), 2006. Urban Sprawl in Europe. Report 10.
European Environmental Agency, 60 pp.
Feria, J.M., 2004. Problemas de definición de las áreas metropolitanas en España
(Definition problems in metropolitan areas in Spain). B. Asoc. Geogr. Esp. 38,
85–99 (in Spanish: English abstract).
Font, A. (Ed.), 2004. L’explosió de la ciutat. COAC i Fòrum Universal de les Cultures
de Barcelona, Barcelona (in Catalonian).
Forman, R.T.T., 1995. Land Mosaics: The Ecology of Landscapes and Regions. Cambridge University Press, New York.
Forman, R.T.T., Godron, M., 1986. Landscape Ecology. John Wiley, New York.
Forman, R.T.T., Spelling, et al., 2003. Road Ecology: Science and Solutions. Island
Press, Washington, DC.
Franco, D., Bombonato, A., Mannino, I., Ghetti, P.F., Zanetto, G., 2005. The evaluation of a planning tool through the landscape ecology concepts and methods.
Manage. Environ. Qual. 16, 55–70.
Freemark, K., Hummon, C., White, D., Hulse, D., 1996. Modeling risks to biodiversity in past, present and future landscapes. Technical Report No. 268. Canadian
Wildlife Service, Ottawa, 60 pp.
Galster, G., Hanson, R., Ratcliffe, M.R., Wolman, H., Coleman, S., Freihage, J., 2001.
Wrestling sprawl to the ground. Defining and measuring an elusive concept.
Housing Policy Debate 12, 681–717.
Geoghegan, J., Wainger, L.A., Bockstael, N.E., 1997. Spatial landscape índices in a
hedonic framework: an ecological economics analysis using GIS. Ecol. Econ. 23,
251–264.
Gustafson, E.J., 1998. Quantifying landscape spatial pattern: what is the state of the
art? Ecosystems 1, 143–156.
Herold, M., Goldstein, N.C., Clarke, K.C., 2003. The spatiotemporal form of urban
growth: measurement, analysis and modeling. Remote Sens. Environ. 86,
286–302.
Herold, M., Couclelis, H., Clarke, K.C., 2005. The role of spatial metrics in the analysis and modelling of urban land use change. Comput. Environ. Urban. 29, 369–
399.
Herzog, F., Lausch, A., 2001. Supplementing land use statistics with landscape metrics: some methodological considerations. Environ. Monit. Assess. 72, 37–50.
Indovina, F., 2005. L’esplosione della città (The explosion of the city). Ed Compositori,
Bologna (in Italian).
Kasanko, M., Barredo, J.I., Lavalle, C., McCormick, I., Demicheli, L., Sagris, V., Brezger,
A., 2006. Are European cities becoming dispersed? A comparative analysis of 15
European urban areas. Landsc. Urban Plann. 77, 111–130.
Keiner, M., Arley, K., 2007. Transnational city networks for sustainability. Eur. Plann.
Stud. 15, 1368–1395.
Kim, J., Ellis, C., 2009. Determining the effects of local development regulations
on landscape structure: comparison of the woodlands and North Houston, TX.
Landsc. Urban Plann. 92, 293–303.
Li, H., Wu, J., 2004. Use and misuse of landscape indices. Landsc. Ecol. 19, 389–399.
Li, X., Yang, Q., Liu, X., 2008. Discovering and evaluating urban signatures for simulating compact development using cellular automata. Landsc. Urban Plann. 86,
177–186.
Luck, M., Wu, J., 2002. A gradient analysis of the landscape pattern of urbanization
in the Phoenix metropolitan area of USA. Landsc. Ecol. 17, 327–339.
Marull, J., Mallarach, J.M., 2005. A GIS methodology for assessing ecological connectivity: application to the Barcelona Metropolitan Area. Landsc. Urban Plann. 71,
243–262.
Mateucci, S., Silva, M., 2005. Selección de métricas de configuración espacial para la
regionalización de un territorio antropizado (Selection of spatial configuration
metrics for the regionalization of an anthropized territory). Geofocus 5, 180–202
(in Spanish: English abstract).
McGarigal, K., S.A. Cushman, M.C. Neel, E. Ene. 2002. FRAGSTATS: Spatial Pattern
Analysis Program for Categorical Maps. Computer software program produced
by the authors at the University of Massachusetts, Amherst. Available at the
following web site: www.umass.edu/landeco/research/fragstats/fragstats.html
(last accessed 15 March 2010).
Menor, J., 2000. La Vega de Granada: transformaciones agrarias recientes en un
espacio periurbano (La Vega of Granada: recent agrarian transformations in a
peri-urban space). Ed. Universidad de Granada, Granada (in Spanish).
Ministerio de Vivienda, 2006. Atlas estadístico de las áreas urbanas en España
2004[[nk]]Statistical atlas of urban areas in Spain 2004. Ministerio de Vivienda,
Madrid (in Spanish).
238
F. Aguilera et al. / Landscape and Urban Planning 99 (2011) 226–238
Muñiz, I., García, M.A., Calatayud, D., 2008. SPRAWL. Definición, causas y efectos
(SPRAWL. Definition, causes and effects). Departament d’Economia Aplicada
UAB. Documento de treball 06.03 (in Spanish).
Newman, P., Kenworthy, J., 1999. Sustainability and Cities: Overcoming Automobile
Dependence. Ed Island Press Books, Washington, DC.
Observatorio de la Sostenibilidad en España (OSE), 2006. Informe de la sostenibilidad
en España 2006. (Sustainability in Spain: 2006 Report). Mundi Prensa, Madrid
(in Spanish).
O’Neill, R.V., Krummel, J.R., Gardner, R.H., Sugihara, G., Jackson, B.L., DeAngelis, D.L.,
Milne, B.T., Turner, M.G., Zygmunt, B., Christensen, S.W., Dale, V.H., Graham, R.L.,
1988. Indices of landscape pattern. Landsc. Ecol. 1, 153–162.
Petrov, L., Lavalle, C., Kasanko, M., 2009. Urban land use scenarios for a tourist region
in Europe: applying the MOLAND model to Algarve, Portugal. Landsc. Urban
Plann. 92, 10–23.
Poelmans, L., Van Rompaey, A., 2009. Detecting and modelling spatial patterns of
urban sprawl in highly fragmented areas: a case study in the Flanders–Brussels
region. Landsc. Urban Plann. 93, 10–19.
Prato, T., 2007. Evaluating land use plans under uncertainty. Land Use Pol. 24,
165–174.
Riitters, K.H., O’Neill, R.V., Hunsaker, C.T., Wickham, Yankee, D.H., Timmins, S.P.,
Jones, R.B., Jackson, B.L., 1995. A factor analysis of landscape pattern and structure metrics. Landsc. Ecol. 10, 23–39.
Santé, I., García, A.M., Miranda, D., Crescente, R., 2010. Cellular automata models
for the simulation of real-world urban processes: a review and analysis. Landsc.
Urban Plann. 96, 108–122.
Seto, K.C., Fragkias, M., 2005. Quantifying spatiotemporal patterns of urban land-use
change in four cities of China with time series landscape metrics. Landsc. Ecol.
20, 871–888.
Song, Y., Knaap, G.J., 2004. Measuring urban form. Is Portland winning the war on
sprawl? J. Am. Plann. 70, 210–225.
Steinitz, C., Arias, H., Bassett, S., Flaxman, M., Goode, T., Maddock III, T., Mouat, D.,
Peiser, R., Shearer, A., 2003. Alternative Futures for Changing Landscapes. The
Upper San Pedro River Basin in Arizona and Sonora. Island Press, Washington,
DC.
Torrens, P., Alberti, M., 2000. Measuring sprawl. In: CASA Working Paper Series, 27.
Turner, M.G., 2005. Landscape ecology: what is the state of the science? Annu. Rev.
Evol. Syst. 36, 319–344.
Uuemaa, E., Antrop, M., Roosaare, J., Marja, R., Mander, U., 2009. Landscape metrics
and indices: an overview of their use in landscape research. Living Rev. Landsc.
Res. 3, http://www.livingreviews.org/lrlr-2009-1 (last accessed 15 March 2010).
Valenzuela, L.M., Aguilera, F., Soria, J.A., Molero, E., 2008. Designing and assessing
of development scenarios for metropolitan patterns. In: Paegelow, M., Camacho, M.T. (Eds.), Modelling Environmental Dynamics. Springer-Verlag, Berlín,
pp. 337–362.
Van Beusekom, R., 2003. Livability and sustainability indicators for land-use models:
potential use of pattern-describing indicators for spatial planning. Msc Thesis.
University of Utrecht, Utrecht.
White, R., Engelen, G., Uljee, I., 1997. The use of constrained cellular automata for
high resolution modelling of urban land use dynamics. Environ. Plann. B 24,
323–343.