Content uploaded by Itziar Navarro-amezketa
Author content
All content in this area was uploaded by Itziar Navarro-amezketa on Jan 07, 2017
Content may be subject to copyright.
SSS10 Proceedings of the 10th International Space Syntax Symposium
054
Plotting urban growth:
Fishing towns in southern Portugal, 1970-2014
Abstract
This paper examines the evolution of eight fishing towns in Algarve, in order to identify the spatial
typologies that characterise changes in the urban fabric. The star model (Hillier et al., 2012) and the
mean and maximum variables of normalised integration and choice (NAIN and NACH) were employed
to compare the spatial configuration’s changes in two different periods: 1970 and 2014. The results
show an increase in segregation from 1970 to 2014 as well as a more fragmented and less cohesive
urban tissue in most of the towns.
Keywords
Space syntax, NAIN and NACH variables, star model, idealised geometric diagram, urban space
evolution, southern Portugal towns.
1. Introduction
From fisheries to tourism: Urban changes in the coastal fishing towns of Algarve
Since the mid 1970’s, the coastal of Algarve in southern Portugal has experienced an intense urban
evolution mainly due to tourism-induced development, attracting thousands of tourists every
summer. Domestic tourism, which until then had merely supplemented foreign tourism, experienced
great development being converted into a strategic sector (Almeida, 2012).
Based on space syntax concepts and models this paper attempts to build a model representing
morphological changes that occurred in the urban configuration of a set of eight coastal fishing
towns, which were confronted with tourism pressure. “Fishing towns” refers to urban settlements
that before the tourism boom had an economy mainly based on fishing, counting on a fishing port or
a small size commercial fishing marina. The purpose is to provide a description of the morphology of
theses fishing towns, placed in a diachronic context and thus offer an understanding of the processes
by which they are being transformed.
Itziar
Navarro-Amezketa
Architecture Department of the Higher Technical School of Architecture of Donostia
-San Sebastián,
University of the Basque Country, UPV/EHU
i.navarro.amezketa@gmail.com
Mafalda
Batista Pacheco
Instituto Superior Técnico/ Universidade de Lisboa, IST UL
mbatistapacheco@gmail.com
Teresa
Heitor
Instituto Superior Técnico/ Universidade de Lisboa, IST UL
teresa.heitor@tecnico.ulisboa.pt
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:1
SSS10 Proceedings of the 10th International Space Syntax Symposium
Although these fishing towns have undergone significant morphological changes, research
concerning these changes remains limited. Besides, these towns present key geo-morphological
features, which could serve as a basis for explaining tourism-induced changes in coastal areas. This
will provide a useful framework for future research and its operationalization will help to gain insight
for morphological evolution of Algarve coastal area.
The morphological changes in this set of fishing towns under analysis are characterized by the
expansion of the urban tissue together with the fragmentation or the dispersion of the historical
core, following an expansion pattern where new developments are often discontinuous, low dense
and extensive. The main research question is focused on how far theses changes are related with
intrinsic spatial factors. It is argued, the main hypothesis, that although the foreground structure
strengthened after the urban expansion, the background became segregated and even more the
expansion areas, forming isolated cores.
Space syntax techniques and description tools are used to analyze these urban settlements in a
comprehensive manner and to capture the relationships within their functional structure. Geometric
and topological properties of the urban network on a global scale are also explored, such as
integration and choice variables, measures based on both the axial map and the angular segment
maps (Hillier and Iida, 2005). These variables identify structures that overcome the local area and
connect semi-connected areas through the global network at different scales. Advances in angular
analysis allow the comparison of systems of different sizes introducing normalisation of integration
and choice variables (Hillier et al., 2012). The choice variable becomes a necessary condition to
minimise the cost of segregation in the urban area, defined as the principle of cost-benefit, as
introduced by Tao Yang (ibid.)). Cost-benefit approach postulates that people primarily move in a
way that enables them to conserve effort, time and expense. In that sense, the cost-benefit measure
refers to the outcomes of the urban configuration with regard to optimal accessibility patterns
measured in terms of travel cost (distance, time and energy).
The star model (ibid.)) and the variables mean and maximum of normalized integration and choice
(NAIN and NACH) are employed to compare the spatial configuration’s changes in two different
periods: 1970 and 2014. The identification of the spatial configuration that characterize the urban
fabric transformation, linked with the data and analysis obtained, will contribute to the current
debate on the impact of tourism on the urban structure of the eight fishing towns under analysis.
The paper is organized in three parts: the first one introduces the study object; the second one
clarifies the materials and methods used in the analysis; the third part describes the case studies and
highlights the main urban changes occurred in eight southern coastal fishing towns in Portugal.
The urban framework
Eight fishing towns with a population between 15.000 and 65.000 inhabitants were selected to be
analysed under the scope of this study (Figure 1). The historical cores of these fishing towns are
based on an irregular street network of a radial (Quarteira, Faro and Tavira) or orthogonal form
(Portimão and Vila Real Santo António) or both (Lagos, Albufeira and Olhão), being adapted to the
topography and to the morphology of the sea front line. Their street network has a structuring value:
it establishes directions, hierarchies and functions (Teixeira and Valla, 1999). The fishing port is a
major landmark in real and symbolic terms: a convergence point linked to the fishing auction and
market and to the main commercial axis as well as the fishing related industries. The church is also
an import element, often located in a square. The built fabric is characterized by compacted blocks,
narrow buildings, facing directly the street, with small back yards. The most accessible spaces are
almost concentrated around the fishing port and limited by the sea - acting as a natural barrier. They
form a continuous structure that extends through axis of greater dimensions parallel to the sea
without penetrating deep into the interior of the system so as to articulate the surrounding areas.
Up to the mid-1960s, fishing and agriculture were the main activities of the population. After the
1970s, when the influx of tourists increased, the situation changed. Within the historical core, most
of the traditional buildings were reconverted to satisfy tourist-related activities (e.g. restaurants,
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:2
SSS10 Proceedings of the 10th International Space Syntax Symposium
coffee shops, discotheques and rent-room facilities). The street network was expanded and the
surrounded agriculture land changed use leading to new residential areas combined with other
economic activities. Hotels of varying sizes, camping grounds and touristic resorts mostly composed
of detached and semi-detached houses were located outside the town core in urban expansion
areas, which tend to create secondary and autonomous nuclei. In spite of an apparent chaotic
appearance and highly visible physical transformation, as a result of an uncontrolled development
process these towns still preserve part of their traditional character.
Figure 1: Map of southern Portugal and the eight fishing towns under study.
2. Materials and method
In order to verify the main hypothesis, this sample was divided in three secondary hypotheses: 1)
towns where the foreground was strengthened after the urban expansion; 2) towns where the
background weakened after the urban expansion; and 3) towns where the local-global relationship
became same pattern for all of them after the urban expansion.
Segment maps of the eight fishing towns were created considering their topological, metric and
angular connections with radio “n”, in two periods: before and after tourism expansion (1970’s and
2014). The first period, named Historic Network, was informed and characterised following an
archive surveys (Lagos (Padrão, 1967), Portimão (Padrão, 1969), Albufeira (Padrão, 1966), Quarteira
(Anon., n.d.), Faro (Padrão, 1970), Tavira (Padrão (b), 1970) and Vila Real de Santo António (Padrão,
1967); Olhão (Martins, 1961); Data recorded from the Municipal Master plans and Google Maps
(IGP/DGRF, 2014) platform were used to characterise the second period, called Current Network.
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:3
SSS10 Proceedings of the 10th International Space Syntax Symposium
The segment maps were designed in Autocad 2012 (Autodesk, 2012); and processed and calculated
with the UCL depthmap software (Turner and Friedrich, 2011; 2000). Statistical analysis of the data
was conducted with the R Software (Gentlement and Ihaka, 2015; 1993) and RKWard Software
(Friedrichsmeier et al., 2015; 2002).
Firstly, the Star Diagram was used to compare urban networks based on normalised variables (Hillier
et al., 2012), (Figure 2, 3, 4, 5 and 6).
The Star Diagram is composed of four vertices (ibid.)); at the top and bottom of the vertical axis are
the mean values of the variables NACH (top) and NAIN (bottom); on the horizontal axis are the
maximum values of NACH (right) and NAIN (left). All the measurements have a normalised score,
oscillating approximately 0, with negative values towards the centre and positive values at the limits
of the diagram.
Secondly, the Idealised Geometric Diagram (ibid.) was used to compare the structures of different
towns by normalising the variable choice. A particular value for choice was specified, and values of
the urban structures above and below that value were analysed. Segments with a value of 1.5 or
higher identify the global structure of the system, foreground; those with 1.4 represent how the
global structure is connected with local organisation, background; and those with 1.6 identify the
NACH central core of the system (ibid.)).
Considering a global structure and how town sectors are accessible to each other and to outside
sectors, the analysis suggests three structural features: 1) radial structure connecting the town
centre to the outside; 2) lateral structure, connecting the town sectors independent of the centre;
and 3) ring structure in specific scales to make local areas accessible to each other and the radial and
lateral structures. These three structural features can be visualised in the Idealised Geometric
Diagram for the urban system (Figure 2), which is an approximation of a proper urban network
procedure (Hillier et al., 2012). The segments with background values of 1.3 (orange) and 1.4 (thin
red), foreground values of 1.5 (medium-thick red) and central cores of 1.6 (thick red) on each
segment map of the eight fishing towns were marked, forming structures according to the radial,
lateral and ring categories (Figures 3, 4, 5 and 6).
Figure 2: Star Diagram and Idealised Geometric Diagram (ibid.): 1) radial, 2) lateral and 3) ring.
Thirdly, descriptive characteristics of the study data were analysed through exploratory data analysis
using mean, standard deviation and a correlation analysis between the measured variables. The
contrasting hypothesis method is parametric, formulating the null and alternative hypotheses with
an error margin of 0.05.
3. Result and analysis
The Star Diagram and Idealised Geometric Diagram of the fishing towns.
Lagos. The Star Diagram of the current Lagos presents decreasing values of mnNAinteg, from 0.87 to
0.63, and of mnNAchoice, from 0.9 to 0.8 (Figure 3.1a). However, mxNAinteg and mxNAchoice
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:4
SSS10 Proceedings of the 10th International Space Syntax Symposium
remain approximately 1.1 and 1.4, respectively. The decrease in both background variables’ mean
values signifies low accessibility of residential areas, creating more holes in these urban networks.
Regarding the urban structure before the expansion, the main streets were distributed radially
through the centre and extended through the lateral with a NACH value of approximately 1.3 (Figure
3.1b). Even having higher movement compared with other areas, then, is not expected to result in
much difference. In today’s Lagos (Figure 3.1c), the streets with maximum NACH values of 1.4 cause
two main radials that cross along the lateral sides of the urban network without entering the
background network.
Portimão. Comparing the variables of mxNAinteg, mnNAinteg and mnNAchoice for the Historic
Network and Current Network, the variables values are reduced (mxNAinteg from 1.5 to 1.3,
mnNAinteg from 1.1 to 0.8 and mnNAchoice 1.0 to 0.8). In contrast, the variable mxNAchoice, which
represents the main urban network, rises from 1.4 to 1.6 (Figure 3.2a). These values indicate that the
urban expansion lowered their integration and additionally increased the holes in the background
network. Adding the fact that the urban foreground structure increased, the network remains far
from cohesive. The foreground consolidation increases the segregation of the background, i.e., the
expansion areas.
Analysing NACH geometry for the Portimão Historic Network (Figure 3.2b), the streets with main
circulation surrounded the urban network, and the difficulty of accessing the centre and the
residential area presents higher NACH values (1.4 and 1.3). However, the Current Network is
structured through a main radial (with a NACH value of 1.5 in the centre), which passes through all
the urban area and then subdivides into radials with NACH values of 1.4 and 1.3 (Figure 3.2c).
Albufeira. The Star Diagram shows that the variables mxNAinteg, mnNAinteg and mxNAchoice
increased with the urban expansion (from 0.7 to 1.1, 0.5 to 0.6 and 1.4 to 1.6, respectively; however,
the mnNAchoice value remains approximately 0.8.), which means that global accessibility improved
and urban holes in the background were reduced (Figure 4.3a). Regarding NACH axial geometry of
the Albufeira Historic Network, the maximum value is approximately 1.3, with low movement
variation among streets (Figure 4.3b). Instead, in the Current Network, a main radial crosses the
entire urban network, obtaining the highest value of 1.5 at the centre, also having some radial
subdivisions (Figure 4.3c). This finding indicates that this radial accepts the main movement of the
urban network, differing from the background, i.e., mostly residential areas.
Quarteira. In contrast with the previous urban network, all the variables analysed in Quarteira
decreased, as seen in the Star Diagram: mxNAinteg decreased from 1.4 to 1.0, mnNAinteg from 0.95
to 0.7, mnNAchoice from 0.9 to 0.8 and mxNAchoice from 1.5 to 1.4 (Figure 4.4a). These findings
indicate that urban accessibility and structuration worsened with the urban expansion. The Historic
Network of Quarteira is characterised by a main radial that crosses the settlement with NACH values
of 1.5 in the centre (Figure 4.4b), unobserved in other examples of the Historic Network. Instead, in
the current network, the streets with superior movement fluxes are almost at the lateral, only
crossing the network at the centre (Figure 4.4c). Furthermore, the foreground value decreases,
reaching values far from the 1.5 of the Historic Network.
Faro. As the Star Diagram shows, all the variables increased their values with the urban expansion –
mxNAinteg improved from 0.7 to 1.4, mnNAinteg from 0.5 to 0.9, mnNAchoice from 0.8 to 0.9 and
mxNAchoice from 0.4 to 1.5 (Figure 5.5a). Faro is the only fishing town studied that improved in all
the variables measured. The global network, foreground, of the Historic Network was articulated by
a central radial, with an mxNAchoice value of 1.4 in the centre and 1.3 on the lateral sides (Figure
5.5b). Currently, radials, rings and lateral sides abound in the urban network. The global structure is
strengthened in the centre, with an mxNAchoice value of 1.5, while the relationship with a local scale
occurs in a constant way (Figure 5.5c).
Olhão. The mxNAinteg and mnNAinteg values increased, from 1.4 to 1.6 and 0.9 to 1.0, respectively
(Figure 5.6a). Furthermore, the mnNAchoice variable decreased from 0.9 to 0.8, indicating that some
holes were created in the local network configuration. The mxNAchoice values increased from 1.5 to
1.6, structuring the global scale regarding local. The axial geometry of the Olhão Historic Network
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:5
SSS10 Proceedings of the 10th International Space Syntax Symposium
indicates that the main streets are mostly on lateral, with a global scale mxNAchoice value of 1.5,
very different from the rest of the network (Figure 5.6b). In the Current Network, streets with higher
movement create a radial structure, with some subdivisions and rings (Figure 5.6c).
Tavira. Comparing the Historic Network with the Current Network, the variable values (mxNAinteg,
mnNAchoice and mxNAchoice) remain the same, except for the increase in mnNAinteg from 0.6 to
0.7 (Figure 6.7a), indicating that the Historic Network urban characteristics were preserved in the
urban expansion (Figure 6.7c). The axial geometry of the Historic Network reveals a single radial
street that articulates the entire nucleus with a maximum value of 1.5 of mxNAchoice in the centre
and gradually decreases to the limits (Figure 6.7b). The current Network presents a radial that
crosses the settlement, with small subdivisions to the laterals, without high values in structuration,
i.e., in mxNAchoice (Figure 6.7c).
Vila Real de Santo António (VRSA). Comparing the Star Diagram values of the Historic Network with
the Current Network, the variable values of global level accessibility (mxNAinteg from 2.1 to 1.9)
decreased as local level accessibility (mnNAinteg from 1.8 to 1.2) and also reduced mnNAchoice
(from 1.1 to 0.9), creating more holes. Only global structure increased significantly (mxNAchoice
from 1.4 to 1.6) (Figure 6.8a). This finding indicates that in the Current Network, the foreground
segregates the background even further, which considerably worsens the global-local relationship.
The axial geometry of the Historic Network of Vila Real de Santo António shows a hierarchy on the
grid, without differences among streets, that does not constitute a strong structure (Figure 6.8b).
Instead, the Current Network leans in a radial that crosses the entire network, reaching mxNAchoice
values of 1.6, articulating main streets’ fluxes in this radial. The background structure remains
perpendicular to this global structuration (Figure 6.8c).
Figure 3: Star Diagrams (a) and Idealised Geometric Diagrams of the fishing towns of Lagos and Portimão
Historic Network (b) and their Current Networks (c).
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:6
SSS10 Proceedings of the 10th International Space Syntax Symposium
Figure 4: Star Diagrams (a) and Idealised Geometric Diagrams of the fishing towns of Albufeira and Quarteira
Historic Network (b) and their Current Networks (c).
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:7
SSS10 Proceedings of the 10th International Space Syntax Symposium
Figure 5: Star Diagrams (a) and Idealised Geometric Diagrams of the fishing towns of Faro and Olhão Historic
Network (b) and their Current Networks (c).
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:8
SSS10 Proceedings of the 10th International Space Syntax Symposium
Figure 6: Star Diagrams (a) and Idealised Geometric Diagrams of the fishing towns of Tavira and Vila Real de
Santo António Historic Network (b) and their Current Networks (c).
Table 1 shows how the mean values of NAIN and NACH decrease in the background (from 0.9 to 0.8
in both mnNAinteg and mnNAchoice) and increase in the foreground (from 1.27 to 1.3 in mxNAinteg
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:9
SSS10 Proceedings of the 10th International Space Syntax Symposium
and 1.4 to 1.5 in mnNAchoice) in the current network after the tourism expansion, weakening the
background network and strengthening the foreground.
Historic Network
Current Network
Mean
Stand. Dev.
Mean
Stand. Dev.
mxNAinteg
1.2752
0.4239
1.3245
0.2913
mnNAinteg
0.9312
0.3969
0.817
0.1929
mxNAchoice
1.467
0.0506
1.5387
0.0515
mnNAchoice
0.9294
0.0993
0.8557
0.0422
Table1: Statistical mean of four variables of Historic Network and Current Network after tourist expansion.
Global structure, foreground (mxNAchoice and mxNAinteg variables)
Figure 7a shows the differences between the two periods under analysis. The Historic Network, are
more dispersed and the slope of the regression line is small, while in the Current Network, the
regression line and the slope is steeper.
In the group Historic Network, the regression coefficient, b, is 0.041 and the coefficient of
determination R2 is 0.1178, i.e., having little explanatory power. In contrast, in the group Current
Network, the regression coefficient is 0.1512, which is expected to change mxNAchoice for each
increment mxNAinteg. The coefficient of determination is 0.730, and considering the values ranging
from 0 to 1 (0< R2<1), we can state that the regression line has a good capacity to fit; therefore,
there is a direct relationship between variables (Figure 7a).
Local structure, background (mnNAchoice and mnNAinteg variables)
The scatterplot 7b (Figure 7) shows that the two regression lines of the Historic Network and Current
Network groups are parallel to each other and have a similar slope.
The group Historic Network has a regression coefficient of 0.23 and a coefficient of determination of
0.91, indicating that the data fit adequately to the model. The Current Network group has a
regression coefficient of 0.2 and coefficient of determination of 0.84, lower than those of group 1
(Figure 7b).
Figure 7: 7a) Global Structure, foreground. Scatterplot of the independent variable mxNAinteg and the
dependent variable mxNAchoice of the Historic Network and Current Network; 7b) Local Structure, background.
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:10
SSS10 Proceedings of the 10th International Space Syntax Symposium
Scatterplot of the independent variable mnNAinteg and the dependent variable mnNAchoice of the Historic
Network and Current Network.
Local-global relationship (Global.Foreground and Local.Background variables)
In the scatterplot of Figure 8, the two regression lines have equal slopes. The values of Current
Network are approximately 1, as in the Historic Network. In the group Historic Network, the
regression coefficient is 0.93 and the coefficient of determination R2 is 0.98, which has a higher
capacity to fit. In the Current Network, the regression coefficient is 0.91 and the coefficient of
determination R2 is 0.98, as in the group Historic Network.
In the Historic Networks of towns with high integration and low hierarchy of their networks, such as
Lagos, Portimão, Quarteira and Vila Real de Santo António, in the Current Network, both the
hierarchy of the foreground and segregation increased (Figure 8). While Albufeira, Faro, Olhão and
Tavira had the lowest values in the local-global relationship of the Historic Network, axial and ring
geometries were created in the expansion area, improving the integration value.
Figure 8: Scatterplot of the independent variable Local.Background and the dependent variable
Global.Foreground of the Historic Network and Current Network.
4. Discussion
As observed in the descriptive analyses of the Star Diagrams and Idealised Geometric Diagrams,
compared with the classification conducted by Hillier et al. (2012), Lagos, Portimão, Albufeira and
Olhão enhance background structure, generalising the local structure, almost without regard to the
global foreground structure, while Quarteira, Faro and Tavira exhibit a strong foreground structure
due to a weak background structure, i.e., they have a strong foreground hierarchy. In VRSA, the
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:11
SSS10 Proceedings of the 10th International Space Syntax Symposium
value of the background is maximised to reduce the differences between the background and
foreground networks.
Currently, more than half of the towns (Lagos, Portimão, Albufeira, Olhão and VRSA) decrease their
mnNAchoice, creating more holes in the network, and increase their mxNAchoice. The consolidation
of the foreground further segregates the urban network. Quarteira decrease its mnNAchoice and
mxNAchoice values, resulting in a less differentiated foreground and a more segregated background.
Tavira shows a different pattern. It increases mnNAchoice and improves the urban holes ratio while
decreasing mxNAchoice. Faro achieves a more balanced development, raising its mnNAchoice, which
now has fewer holes in its mesh, and its mxNAchoice, strengthening the structure of the foreground.
Statistical tests about the conjectures formulated above: acceptance or rejection of the null
hypothesis, H0
“The foreground was strengthened after the urban expansion” (global structure, foreground)
In the null hypothesis H0, the normalised variables of choice and integration in the foreground are
independent in both the Historic Network and the Current Network after the touristic expansion;
In the alternative hypothesis Ha, the normalised variables of choice and integration in the foreground
are interdependent, i.e., there is a relationship in both the Historic Network and the Current Network
after the touristic expansion.
In the statistics of the group Historic Network, the p-value is 0.40>0.05; as a result, H0 cannot be
rejected, i.e., the variables are independent and there is no relationship between mxNAinteg and
mxNAchoice for the Historic Network.
In the statistics of the group Current Network, the p-value is 0.0068<0.05; therefore, mxNAinteg and
mxNAchoice are dependent in the Current Network after the tourism expansion. Thus, the
foreground structures of these towns strengthened after urban expansion.
“The background weakened after the urban expansion” (local structure, background)
In the null hypothesis H0, the normalised variables of choice and integration in the background are
independent in both the Historic Network and the Current Network after the touristic expansion;
In the alternative hypothesis Ha, the normalised variables of choice and integration in the
background are interdependent, i.e., there is a relationship in both the Historic Network and the
Current Network after the touristic expansion.
In the statistics of the group Historic Network, the p-value is 0.00019<0.05; therefore, H0 is rejected,
which means that the two variables are dependent, and mnNAinteg is able to describe mnNAchoice.
Consequently, before the touristic expansion, most of the residential areas of the Historic Network
were cohesive.
In the statistics of the group Current Network, the p-value is 0.001<0.05, discarding H0, so the
variables are related. As in the Historic Network, the residential areas are related, although less
consistently.
“After the urban expansion, the local-global relationship has the same pattern in the coastal fishing
towns” (Global-Local relationship)
In the null hypothesis H0, the variables Local.Background (mnNAinteg/ mnNAchoice) and
Global.Foreground (mxNAinteg/mxNAchoice) are independent in both the “Historic Network” and the
“Current Network” after the touristic expansion;
In the alternative hypothesis Ha, the variables Local.Background (mnNAinteg/ mnNAchoice) and
Global.Foreground (mxNAinteg/mxNAchoice) are interdependent, i.e., there is a relationship in both
the Historic Network and the Current Network after the touristic expansion.
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:12
SSS10 Proceedings of the 10th International Space Syntax Symposium
In the statistics of the group Historic Network, the p-value is 1.144e-06<0.05; therefore, H0 is
rejected, which means that the two variables are dependent.
In the statistics of the group Current Network, the p-value is 1.70e-06<0.05; consequently, H0 is
discarded and the variables are related. As expected, the local-global relationship occurred before
and after the urban expansion. The most important issue is that the mean value is present in both
groups and that the data variation decreased from 0.3 to 0.18, having the same pattern in the
network after the urban expansion.
Although the urban expansion strengthened the foreground, it did not help structure the urban
network by integrating it but rather emphasised its segregation even further, disconnecting the
background from both itself and the foreground. Faro is an exception because its network was not
segregated and even creates rings that help integrate the background.
The importance of the segregated background, which includes the residential areas and resorts, as
opposed to the structured foreground, as a basis for the expansion model of most towns, resulted in
an urban fabric that was less integrated and the least local-globally cohesive.
5. Conclusion
The aim of this paper was to improve the understanding of morphological changes occurred in eight
fishing towns in Algarve due to tourism-induced development. The general urban expansion scenario
was addressed by comparing two different periods. The initial analytical approach has revealed that
morphological changes occurred at a deep structural level of the urban network. Since the formal
identification, characterization and quantification of these changes was inaccessible by means of
traditional analysis techniques, the star model (Hillier et al., 2012) and the variables mean and
maximum of normalized integration and choice (NAIN and NACH) were applied. This has allowed to
efficiently extract quantitative information and to gain a better understanding of the expansion
process. Nevertheless, this model should be tested to other coastal towns in order to evaluate the
degree to which it can be used as a generalized description tool and to determine how far it must be
enhanced. In particular is necessary to deep further the variables that configure the urban fabric, to
carry out an objective diagnosis of the urban space, being able to apply corrective measures and
approach urban theory and practice.
References
Almeida, F. (2012), ’La política turística en España y Portugal’. In: Cuadernos de Turismo (eds), Malaga: Univerty
of Malaga, vol. 30, pp. 9-34.
Autodesk (2012), AutoCAD 2012 [computer software]. California: Autodesk.
Friedrichsmeier, T., Ecochard, P., Roediger, S., Kapat, P., Michalke, M. (2002), Rkward (Version 2) [software
available at https://rkward.kde.org/].
Gentleman, R. and Ihaka, R. (1993). R [software available at https://www.r-project.org/].
Hillier, B. and Hanson, J. (1984). The Social Logic of Space, Cambridge: Cambridge University Press.
Hillier, B. (1996). Space Is the Machine, Cambridge: Cambridge University Press.
Hillier, B. (2002), ‘A Theory of the City as Object’. In: Urban Design International (eds.), London: University
College London, vol. 7, pp. 153-179.
Hillier, B. and Lida, S. (2005), ‘Network and Psychological Effects: A Theory of Urban Movement’. In: Cohn, A.
and Mark, D. (eds), Spatial Information Theory, Lecture Notes in Computer Science 3606. New York:
Springer Verlag, pp. 473-490.
Hillier, B., Yang, T. andTurner, A. (2012), ‘Normalising least angle choice in Depthmap- and how it opens up new
perspectives on the global and local analysis of city space’. In: JOSS. London: University College London,
vol. 3, (2), pp. 155-193.
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:13
SSS10 Proceedings of the 10th International Space Syntax Symposium
Martins, A., Castro, C. and Torres, F. (2004 (1961), ‘Zona 6 – Algarve’. In: Arquitectura Popular em Portugal.
Lisbon: Ordem dos Arquitectos, vol.2, pp.241-359.
Padrão, J. (1966; 1967; 1969; 1970), ‘Albufeira’; ‘Lagos’; ‘Vila Real de Santo António’; ‘Portimão’; ‘Faro’; ‘Tavira’.
In: Prospecção, preservação e recuperação de elementos urbanísticos e arquitectónicos notáveis, em
áreas urbanas e marginais viárias, na região do Algarve. Lisbon: Ministério das Obras Públicas de
Portugal, vols. 5, 7, 12, 19, 22, 23.
Teixeira, M. and Valla, M. (1999), O urbanismo português: séculos XII-XVIII: Portugal – Brasil. Lisbon: Livros
Horizonte.
Turner, A. and Friedrich, E. (2000-2011). UCL Depthmap (Version 10) [Software available at
http://www.spacesyntax.net/software/].
I Navarro-Amezqueta, M Batista Pacheco & T Heitor
Plotting urban growth: Fishing towns in southern Portugal, 1970-2014
54:14
