Measuring and assessing urban sprawl: A proposed indicator system for the city of Thessaloniki, Greece

Author’s Accepted Manuscript Measuring and Assessing Urban Sprawl: A proposed indicator system for the city of Thessaloniki, Greece Georgia Pozoukidou, Ioannis Ntriankos www.elsevier.com/locate/rsase

PII: DOI: Reference:

S2352-9385(16)30116-1 http://dx.doi.org/10.1016/j.rsase.2017.07.005 RSASE71

To appear in: Remote Sensing Applications: Society and Environment Received date: 3 October 2016 Revised date: 25 June 2017 Accepted date: 13 July 2017 Cite this article as: Georgia Pozoukidou and Ioannis Ntriankos, Measuring and Assessing Urban Sprawl: A proposed indicator system for the city of Thessaloniki, Greece, Remote Sensing Applications: Society and Environment, http://dx.doi.org/10.1016/j.rsase.2017.07.005 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Measuring and Assessing Urban Sprawl: A proposed indicator system for the city of Thessaloniki, Greece.

Georgia Pozoukidoua*, Ioannis Ntriankosb a Assistant Professor, School of Spatial Planning and Development, Faculty of Engineering (1st Floor), Aristotle University of Thessaloniki Campus, 54124, Greece. b City Planner [email protected] [email protected] * Corresponding author:

Abstract Urban sprawl phenomenon has been on the planning research agenda for more than five decades now. In an effort to comprehend the phenomenon there have been several quantitative approaches to measure it and identify its spatial patterns. This paper attempts to develop a refined indicator system for measuring and assessing the spatial characteristics of urban sprawl in Greek cities. To do so it suggests a system of indicators adapted to the specificities of Greek cities that tend to be quite unique in terms of their urban expansion practices. At the same time, the study deals with the critical issue of data availability and for that it proposes a minimum, but sufficient subset of indicators based on free satellite imagery that could be applicable in any Greek city. Furthermore, a composite analysis using population data is attempted. The proposed indicator system is applied to the Greater Area of Thessaloniki, the second largest city in Greece that was in constant transition till the recent economic crisis. Application of methodology suggests that the proposed system of indicators can contribute to the measurement and assessment of sprawl, since the results of the indicators validate theoretical findings and recordings for urban sprawl in Thessaloniki.

Keywords: urban sprawl, suburbanization, urban form, Thessaloniki

1. Introduction Urban sprawl phenomenon has been on the planning research agenda for more than five decades now, mainly through the debate in regard to the ideal form of cities and metropolitan areas. In the relevant bibliography, two main models of urban development patterns have been recorded: "compact" and "sprawled". Their difference lies mainly in building density, land use mix, and structure of the transport network. Furthermore, it seems that their environmental impact varies substantially, with urban sprawl model considered to be an extremely unsustainable way for our cities to grow (Jaret et al., 2009). To this end there are numerous studies presenting the impacts of urban sprawl based purely on quantitative approaches and several studies attempting a qualitative and maybe more objective approach (Hamidi et al., 2015). The quantitative approach is usually based on measurements of certain urban sprawl characteristics utilizing a single or a set of indicators.

This study examines urban sprawl as the spatial imprint of the sprawled city model and attempts to develop a refined indicator system for measuring and assessing the spatial characteristics of urban sprawl in Greek cities. Urban sprawl is perceived as the result of a complex mechanism where sociological, political, legal, economic or environmental mechanisms shape city outskirts. Nevertheless the paper focuses mainly on the spatial manifestation of urban sprawl and tries to identify sprawl patterns by analyzing the spatial imprint of the built environment. The proposed indicator system is based on definitions and procedures acquired from several studies that elaborate a quantitative approach in measuring urban sprawl’s spatial characteristics. Indicators are adjusted to the peculiarities of the Greek urban development tradition and planning legislation system. Data availability issues which usually arise in these processes are given high consideration. New data sources that became available over the last decade played a decisive role in the formation of the set of indicators. Hence the necessary data to compute urban sprawl indices were exclusively based on free of cost remote sensing images. Under this notion the proposed set of indicators can be easily applied to any Greek city at no cost by planning practitioners who wish to identify and assess urban sprawl. The system of indicators is validated through its application for the city of Thessaloniki.

2. The Urban sprawl phenomenon 2.1. Conceptual definitions The term “urban sprawl” has been widely used in spatial planning discipline. Despite the widespread adoption of the term and spatial identification of the phenomenon, there has not been a commonly accepted definition (Johnson, 2001, Torrens, 2008). Initial efforts to determine “urban sprawl” date back to the 1970s, and till now numerous definitions have been attempted by several research institutions, universities and organizations. According to Hess the term “urban sprawl” was initially used mainly as an adjective to describe a type of urban development, as a verb to describe the process of this development and as a noun to describe a certain type of urban form (Hess, et al., 2001). In the late 20th and early 21st centuries, as the phenomenon of sprawl became the prominent urban development pattern, there were various definitions stemming mainly from the European and North American literature. However, despite the recognition of the phenomenon and its importance for sustainable urban development still there has not been a universally accepted definition (Johnson, 2001, Jaret, et al., 2009, Terzi and Bolen, 2009, Jaeger, et al., 2010, Hammidi, et al., 2015). Furthermore, when there is a definition it is related to its spatial characteristics, the physiognomy of the study area and the available data, rather than a solid description that would apply to any case of urban sprawl. This peculiarity stems from the fact that each city is a separate entity in which the phenomenon of "urban sprawl" is manifested in different ways mainly due to market, geographic and policy factors; therefore it is difficult to reach general conclusions and to establish a common definition (Bruegman, 2005, Paulsen, 2014). It is worth mentioning that differences in the way urban sprawl is defined can be identified even between formal European agencies such as the European Environmental Agency, the Council of Europe Conference of Ministers Responsible for Spatial/Regional Planning (CEMAT) and the Architects Council of Europe. The review of 16 different definitions of urban

sprawl derived from research papers from the USA and Europe confirms the assertion, which also Johnson (2001) supports, that there is not a common definition for urban sprawl, and each definition is influenced by the particular characteristics of the phenomenon in the area under study, and by the specific choices of scholars or agency that conducts the study. An indicative example would be that in the USA most of the times factors like car usage and road infrastructure are incorporated into the definition of sprawl, in contrast to European studies where this factor is rarely used. Conclusively and for the purposes of this study there was a conscious decision to select an existing definition of urban sprawl that will not rely on the characteristics of the phenomenon per se but will describe the phenomenon with the minimum use of its characteristics, such as the definition proposed by Ermer, Mohrmann and Sukopp stated in 1994, as the process of ‘‘…the spilling-over of settlement areas and of excessive use of the open landscape by unsystematic, mostly weakly condensed extensions of settlement areas in the fringes of urban agglomerations” (Jaeger et al., 2010, p.399).

2.2. Evolution and spatial characteristics In contradiction to common belief, urban sprawl phenomenon has occurred since the beginning of cities’ history, when the compact city within the “city’s walls” started to expand to its periphery. This expansion was characterized by low building density and great land consumption, in relation to the accompanied population increase (Bruegmann, 2005). A substantial rise in the occurrence and intensity of urban sprawl is identified at the end of the 19th century. Actually in the USA and in major metropolitan centres in Northern Europe (London, Paris, etc.), urban sprawl was already excessive by the early 1900’s, while in the rest of Europe it became more intense after World War II. The spatial footprint of the phenomenon varies. Thus, in the USA and United Kingdom where the use of zoning planning ordinances has been extensive, there was development of vast mono-functional residential areas, along with dispersed social infrastructure, office buildings and business parks. At the same time there was an extensive development of road infrastructure and an increase of automobile dependency. On the other hand, in continental Europe four types of urban expansion can be identified, due to the differentiation in historical evolution, economic development, land use policies, geomorphology and culture of each city/country. According to Couch et al., development driven by major infrastructure projects is identified in cities that held important international events (e.g. Olympic Games, EXPO, etc.), while sprawl characteristics are indicated even in cities with population decrease. Meanwhile, a completely different sprawl pattern is revealed in post-socialistic countries, due to their particular land use policies, when in Mediterranean cities and Baltic countries the vast increase of secondary (vacation) housing created peculiar sprawl patterns away from major urban centres (Couch, et al., 2007). It is worth mentioning that European Mediterranean cities share common characteristics which clearly distinguish them from the rest of Europe. According to Newman and Thornley (1996) their planning systems belongs to the “Napoleonic family”, along with Holland and Belgium. In addition, according to EU Compendium of Spatial Planning Systems and Policies (1997), Mediterranean cities constitute a distinct category in relation to planning and its implementation, and belong to the “urbanism tradition” group, were inefficiency in controlling development is one of their main pitfalls. Giannakourou (2005), as well, acknowledges the Mediterranean differentiation, indicating that they are typical representatives of the so-called inefficiency in controlling development, which often results in a rigid, legalistic and formal

model of planning regulations. In addition Leontidou (1990) reports that although Southern European countries emerged through the same capitalistic societal formation as the rest of the “Western Countries”, they experienced diverging trajectories in urban development. Finally, according to McDonogh (2000), European Mediterranean cities are scarcely unique in the relation of society and culture in urban life and development, and as such they provide an interesting template for processual and comparative analysis. In this context, European Mediterranean cities, Thessaloniki included, share common patterns in terms of urban expansion practices and policies. Until the late 1980s, compactness had been the key development feature for several European Mediterranean cities. In that way they developed compact overcrowded centres, with vertical rather than horizontal “spatial segregation” between middle and working class. At the same time, despite the rigid planning regulations, the loose control over new development in peri-urban areas enabled the so-called “informal” housing to become the dominant way of development in the late 1980s (Salvati and Morelli, 2014, Munoz, 2003). Transition from compact to sprawl development patterns was common for several European Mediterranean cities, where two different typologies are identified based on the development course of the cities over the last two decades (Salvati and Morelli, 2014, Colantoni, 2016): A. The wide, polycentric urban regions, characterized by settlement patterns that have evolved through the differential growth of large urban centres, medium–small cities and the areas in-between. This process is identified for example in Barcelona (Duarte, et al., 2011, Bethmont, 2000) and the polycentric system of Apulia-Abruzzo-Marche in Central Southern Italy (Bethmont, 2000) B. The monocentric urban regions, dominated by a large compact centre city with an urban density that decreases inversely proportional to centre city distance. Despite the slow growth or even population decline of centre city, it still is the dominant economic core, while decentralization of the production system is part of the suburbanization process. This process is recognized for example in Athens (Salvati, et al., 2012 and 2015), Rome (Salvati, 2013), Istanbul (Terzi and Bolen, 2009), Madrid (Sisternes, et al., 2014) and Pordenone (Martellozzo and Clarke, 2011). As far as the spatial imprint of either the wide polycentric urban region or the monocentric one, three major types of urban patterns can be identified (figure 1): 1st type: Peripheral accretion, where urban fabric expansion occurs in the fringe and adjacent to existing urban development, 2nd type: Linear development along major transport axis, and 3rd type: Leap-frog development occurring in discontinuity with existing urban fabric and is characterized by the uncontrolled development of new urban cores.

2.3 Urban sprawl indicators Urban sprawl has been the prominent urban development pattern for some decades now and there have been numerous attempts to measure it over the last 30 years. However, in all measurement efforts there has not been a commonly accepted system of indicators (Jaret, et al., 2009, Jaeger, et al., 2010). This was due to the fact that sprawl has been connected to several aspects of urban life like traffic congestion, extensive commutes, environmental quality (Salon, et al., 2012, Holcombe, et al., 2012, Zolnik, et al., 2011), and most recently with physical activity, heart disease and obesity (Kostova, et al., 2011). Therefore, most scholars now agree

that sprawl is a multifaceted phenomenon that is best quantified by a combination of measures (Galster, et al., 2001; Cutsinger, et al., 2005; Frenkel, et al., 2008; Torrens, 2008; Jaeger, et al., 2010; Mubareka, et al., 2011; Hammidi, et al., 2015). In this research more than 20 studies measuring urban sprawl were examined (from European, American, Asian and African continents) and more than 70 indicators were considered (Sisternes, et al., 2014; Salvati, et al., 2013; Xu, et al., 2013; Musakwa, et al., 2013; Fang, et al., 2007; Frenkel, et al., 2008; Song, et al., 2004; Wolman, et al., 2005; Terzi, et al., 2009; Cutsinger, et al., 2005). The analysis confirms the widely accepted assertion that a single indicator could not adequately measure urban sprawl phenomenon, hence a system of indicators is required that would measure several sprawl characteristics (Ntriankos, et al., 2015). Furthermore, it seems that there is not a threshold (a certain value) which determines the existence (or absence) of sprawl development. On the contrary, most of the times the values of indicators are compared over time (for the same city) or synchronically between cities. In addition, the process of defining the most suitable indicators tends to depend more on data availability than anything else. It should be noted that differentiations are also identified in relation to the area under study. Thus, the USA case studies focus more on indicators related to the road network, intersection densities and single-family dwelling units (Ewing, et al., 2010; Holcombe, et al., 2012) while European case studies focus more on land use mix and land cover penetration (Arribas, et al., 2011). Finally, in certain studies a set of indicators is combined in a weighted index, in order to achieve a more integrated approach (e.g. Fang, et al., 2007; Sudhira, et al., 2004; Frenkel and Ashkenazi, 2008). Figure 1: Typology of urban sprawl’s spatial footprint

http://www.geocases.co.uk/sample/urban1.htm

3. Methodology The proposed urban sprawl measurement and assessment methodology is illustrated in figure 2. The process starts with several conceptual clarifications that include definition of the phenomenon of sprawl and determination of its characteristics. Subsequently, local factors contributing to urban development are identified. These factors are usually related to the historical context and facts that affected urban development patterns and to the legal framework of peri-urban and ex-urban development (where sprawl usually occurs). Thereafter, the system of indicators is established by completing three sub-steps: identifying the existence and degree of

urban expansion (step 3a), determining if the identified, in step 3a, urban expansion presents sprawl characteristics (step 3b) and finally indicating the spatial pattern and characteristics of sprawl (step 3c). Last but not least, an evaluation of the proposed indicators is performed (multivariate analysis), in order to establish their significance and contribution in explaining urban sprawl phenomenon. Following is a detailed description of the proposed indicators, which corresponds to step three of the methodology process. As far as step one, it has already been presented in section 2.1 and 2.2 and step two will be presented later in this paper through the case study of Thessaloniki. 3.1 The system of indicators The proposed system of indictors aims to identify, measure and assess the spatial imprint of sprawl. It is based on definitions and procedures acquired from several studies that elaborated a quantitative approach in measuring urban sprawl’s spatial characteristics. Therefore, this paper does not propose new indicators but refines existing ones to fit the Greek context. From the 70 indicators considered in this study, a set of 7 was chosen to form the proposed system. The choice of indicators was made based on two criteria: the easiness of their computation and the sprawl characteristics they attempt to measure. In terms of computability all indicators can be calculated using exclusively free of cost remote sensing images via Landsat USGS Explorer (http://earthexplorer.usgs.gov/). In terms of the sprawl characteristics they attempt to measure, they can be classified into 2 categories. The first category, land cover, consists of indicators which measure land cover penetration and degree of urban expansion. The second category, geometry, is composed of indicators originating in fractal geometry and attempt to identify urban fabric’s geometry and specificities of its spatial imprint. Figure 2: Urban sprawl assessment and measurement methodology steps

•1a. Define the phenomenon of sprawl •1b.Determine the characteristics of sprawl

•2a. Historical Context •2b. Legal Framework

•3a. Identify the existence and degree of urban expansion •3b. Determine if the identified, in step 3a, urban expansion has sprawl characteristics •3c. Indicate the spatial pattern and characteristics of sprawl •3d. Evaluate the proposed indicators

As mentioned earlier, indicators proposed here concern mainly Greek cities, and more specifically they are “calibrated” to fit the peculiarities of the Greek legislation system for peri-

urban areas. Table 1 summarizes all indicators, the spatial entity they were calculated in, and the characteristics they attempt to measure. 1. Urban expansion rate This indicator was established by Xu and Min (2013) and calculates the percentage of change of built-up areas based on formula (1) , where UA is the built-up area in two time points (i and n+i) and n is the time interval of the calculation period. This index is calculated for each time period and corresponds to step 3a of the methodology. UAni  UAi 1 R  100 (1) UAi n 2. Land cover penetration Land cover penetration (step 3b) is computed using two different indicators. The first one originated by Chrisoulakis, et al., (2004), is computed by defining a “window” of 3x3 cells around each cell of the study area. For each adjacent cell with different land cover than the central one, the value of “1” is assigned. Otherwise the value is “0”. Penetration for each cell is calculated based on the values of the “window” and the following formula: 8 x  x 1, if x0  xi 0 i mix( X 0 )   , δ (2) 8 i 1 0, if x0  xi The second index was issued by Musakwa and Niekerk (2013), using a 2x2 km canvas. The indicator is computed using the formula: LUM={-Σ[(pi)ln(pi)]}/lnk, where pi is the proportion of each land use class and k the number of different classes in a 2x2 km “neighbourhood”. The results of the indices depend on the number of the different land cover classes of the dataset. Thus, a more detailed dataset leads to more accurate measurements. Likewise, in case land use data are available, the same indicators can be used to compute land use mix. In this study land cover penetration index is calculated using a 540x540m canvas due to land fragmentation patterns and peri-urban development laws in Greece. 3. Urban fabric continuity This index was issued by Galster, et al., (2001), and computes the degree to which newly developed land has been built in an “unbroken” fashion (step 3b). Study area is divided in ½ mile by ½ mile canvas and each cell is characterized as “built-up” or not. Although, Galster, et al., considers a cell as “built-up” when more than 50 employees or 10 housing complexes are located in ½ mile canvas, in this study, a cell is considered as “built-up” when more than half of the maximum allowed construction has taken place. The maximum allowed construction is calculated based on the peri-urban development legislation (in Greece) and for a 540m canvas the area is 7,236m2. 4. Urban fabric fragmentation Land use or land cover fragmentation (step 3b) is computed using fractal geometry indices. The most common fragmentation index is the mean patch area, where patches are polygons of different urban land uses. This index is used by the European Environmental Agency to compute fragmentation of natural and semi-natural areas (EEA, 2007). However, the normalization of the index with the total built-up area is required to make accurate comparisons. This index was computed using 60x60m cells. The “search” distance is defined as the greatest “minimum distance” between two built-up patches, and for the Greater Area of

Thessaloniki this was 5.86km (calculated for 1977-2011 study period). However, due to the fact that distances are calculated between cells, an inner cell coefficient must be added to each distance (Jaeger, et al., 2010). This coefficient is the mean weighted distance between two points in a 60x60 grid area and was computed, using Monte Carlo method, to be 6.72. Table 1: Urban sprawl indicators Indicator

Spatial entity

Sprawl characteristic measured

Methodology Step

1

Urban expansion rate

Total city

Urban expansion

3a

2a

Land cover penetration (3x3 cells canvas)

Cells, 60x60m grid Existence of sprawl

3b

2b

Land cover penetration (540x540m canvas)

Cells, 540x540m grid

3

Urban fabric continuity

Cells, 540x540m grid

Land use mix

3b

4 5

Urban fabric fragmentation Linear development

Cells, 540x540m grid Cells, 60x60m grid

3b 3c

6

Leap-frog development

Cells, 60x60 m grid

7

Peripheral Accretion

Cells, 60x6 m grid 0

Existence of sprawl Identifying linear development Identifying leap-frog development Identifying development in the periphery of the builtup area

3c 3c

5. Linear development This indicator is computed by the mean “minimum distance” of new built-up areas to major roads axis. It corresponds to the 2nd type of sprawl’s spatial imprint (step 3c) and was used by Fang, et al., (2007) and Bhatta, et al., (2010) resulting in sufficient measurements. 6. Leap-frog development Leap frog development is calculated by the mean “minimum distance” of new built-up areas to pre-existing ones. It corresponds to 3rd type of sprawl’s spatial imprint (step 3c) and was used by Fang, et al., (2007) and Bhatta, et al., (2010) resulting in sufficient measurements. 7. Peripheral Accretion Peripheral accretion corresponds to the 1st type of sprawl’s spatial imprint (step 3c). This index is computed as the mean weighted distance between two points: the first point is within the existing urban fabric in lag time (i.e. 1977) and the second point is within the urban fabric occurring in subsequent time (i.e. 1984). The system looks for new development within a certain distance defined by the researcher. According to Jaeger, et al., (2010) it is required to take into account a weighted function1 to compute the distance. The weighted function is computed via the

1

The weighted function f(a), where a is the distance between two points, must have the following 3 properties (Jaeger, et al., 2010):f(0)=0, due to the fact that when the distance is 0, the weighted distance must also be 0, f(0)=1,

differential equation (df(a))/da=γ (f(a))/a, with solution f(a)=c*a^γ, where c is a constant. In the case of 3 points, representing 3 built-up area patches (figure 3), an increase of the value of x, should result in an increase of the index, so γ must be within (0,1). The most simple of these functions is when γ equals 0.5 and results in f(a)= (-1)+√(1+2a). The mean of the distances is computed by the integral of this function (Jaeger, et al., 2010). Figure 3: Explanation of the weighted function of Peripheral Accretion index

3.2 The population factor In addition to the above indicators that are related to the spatial characteristics of sprawl, another type of indicator associated with population data is taken into account. In general, population change is considered to be a significant sprawl inducing factor (Barnes, et al., 2001). Therefore it could provide us with useful information in regard to urban sprawl spatial patterns, if a composite analysis is performed. Population change is not included in the proposed indicators system but was used in this paper as a way to better understand and interpret indices’ results. 3.3 Data sources and data processing Acquiring appropriate data has always been a major bottleneck in quantitative analysis in the planning field. Recently technology has enabled us to acquire, store and process data faster and in large quantities. Nevertheless there are cases where finding appropriate data is still difficult, if not an impossible, task. Therefore it was imperative to use data sources that are widely available in order to make the proposed indicators system easily applicable. To this end, there was a deliberate decision to use remote sensing images with the highest resolution obtainable that are widely available and at no cost. More specifically the data taken into account to compute the indices were exclusively based on remote sensing images. A total of 5 Landsat satellite images was used corresponding to years 1977, 1984, 1990, 2001 and 2011. All satellite images had 30m accuracy, except 1977 which was 60m. This was the best data available that could be acquired for free, covering the whole study area (Greater Area of Thessaloniki, GATH). In order to calculate the indices, images were classified. Three different land cover categories were identified: built up areas, rural areas and natural areas. The classification method used was supervised classification and the accuracy of each classification is indicated in table 22. As far as population data, census data for 1971, 1981, 1991, 2001 and 2011 were used, provided by the Hellenic Statistical Authority. The lowest spatial level for which this data were

due to the fact that when the distance is 0 the slope must be 1 and (Δf(a))/(f(a)=γ *(Δa)/a, due to the fact that the weighted distance change must be proportional to the relevant distance change. 2

ERDAS Imagine Accuracy Assessment was used for the computations.

available is the municipality entity3 or “municipality/local community” (MLC). The study area, GATH, is comprised of 69 MLCs and this was the spatial entity used to calculate population changes. Table 2: Remote sensing data accuracy Image 1 Image 2 Date Satellite Blue band Green band Red band Near Infrared Short-wave Infrared Pixel size Total accuracy

Image 3

Image 4

Image 5

22/08/1977 24/06/1984 11/7/1990 30/05/2001 19/06/2011 Landsat 2 Landsat 5 Landsat 5 Landsat 7 Landsat 5 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 2 1 1 1 1 2 2 2 2 60 87.50%

30 85%

30 90%

30 82.50%

30 82.50%

The fact that population change was calculated in a different spatial entity than sprawl indices created the need for certain data manipulation in order to acquire composite results. More specifically, sprawl indicators were calculated in cells (60m x 60m) or group of cells (neighbourhoods of 540m x540m), while population indicators were calculated in census units (MLCs). In order to deal with the different spatial entities a cross examination process for the two types of data was conducted. Following is a description of how population data was integrated in the sprawl pattern analysis using a cross examination process: 1. The first step was to identify, in census units, significant changes in urban development patterns based on the calculated urban sprawl indices. Determining the significance of indicators’ results in a census unit (MLC) is not as simple as in a single variable distribution. There are 3 variables to be considered in the evaluation process within each census unit: (a) the number of cells in which indicators’ values are assigned, (b) the mean value of the indices in these cells, and (c) the variance of indicators’ values. Due to crucial differences between indicators a uniform test is not recommended (e.g. the number of cells in which the “peripheral accretion” index is computed remains stable in all computations, while in “urban fabric continuity” and “linear development” indices constantly changes). Furthermore the significance of change in a census unit, based on the proportion of cells with certain index value, differentiates on indicator basis. For example, linear development in a large unit crossed by a major road (e.g. in a rural area) could be significant even though the proportion of cells with index values is very low. Thus, the creation of a generalized function could be achieved through either the use of weight matrix for census units and indicators or the use of step functions and splines, but this is out of the scope of this paper. 3

MLC entity is the smallest administration entity in Greece and the smallest entity that data were available for all study periods. MLCs correspond to the municipality level as defined before the 2011 administrative reform.

2. The second step involves the assignment of a binary variable to each indicator per study period, determining the existence (or absence) of significant sprawl change in each MLC. 3. Next, another categorical variable is assigned for each MLC for population data per time interval, recording 3-5 different types of population trends: a. Percentage change above mean + standard deviation b. Percentage change below mean – standard deviation c. If mean-standard deviation is not always below 0, a new interval could be added at 0. In addition, if either mean standard deviation is too low or mean + standard deviation is too high, new intervals could be created. For example, in case of mean + standard deviation being always above 50%, a new interval should be created to identify changes between 20% and mean + standard deviation d. No significant population change (not included in the above categories) 4. Finally, a binary variable is assigned to state the existence of significant development for each MLC. Due to the fact that continuous urban fabric is considered to be a built-up area that has no more than 50m between buildings4, significant built – up area (SBA) is considered to take place when more than 1/(5+2*sqrt(2))5 of the MLC’s cells are classified as built-up area. Summing up, the first two steps determine which MLCs experience significant change in their sprawl patterns, while the third step determines significant changes in population and the fourth adds a variable to determine the significance of the built-up area to the total area of each MLC. Thus, urban sprawl indicators and population indicators are analyzed alongside and in comparison, to result in five MLC profiles: 1. Significant population increase along with significant sprawl 2. Significant population increase along with absence or small sprawl patterns 3. Significant population decrease without significant sprawl patterns 4. Significant population decrease along with sprawl patterns 5. Significant sprawl patterns without significant population change 3.4 Limitations and prospects of the proposed system Urban sprawl is undoubtedly a complex phenomenon and constitutes the synthetic result of various sociological, political, legal, economic or environmental mechanisms. This study proposes a set of indices in order to identify, assess and measure sprawl through the spatial characteristics of the built-up environment. The focus on identifying and measuring sprawl solely by its spatial imprint entails several advantages and disadvantages. The proposed system can certainly locate the occurrence of urban expansion and identify its characteristics. More specifically it can identify the type of urban development and its spatial form (linear, leap-frog, peripheral). On the other hand, it does not provide any information in 4 5

The value of 50m results from the relative Greek legislation of what constitutes a continuous built-up area.

Assuming that we consider 60m as a limit, due to the fact that this is the cell size of our data, then 8 cells are the adjacent cells to each cell and 4+2*sqrt(2) are the ones in less than 60m distance. Thus in each collection of 9 cells, 1/(5+2*sqrt(2) should be classified as built-up area for the total of MLC to be categorized as “SBA”.

regard to the causes of the phenomenon and a cause-effect relation cannot be established, needed to fully comprehend the phenomenon. Nevertheless, the proposed system constitutes the first necessary step towards this direction since it can provide us with information to monitor urban sprawl phenomenon and initiate relevant planning policies. Furthermore, the study attempts to comprehend the phenomenon by performing a composite analysis. It proposes a method to deal with different data scales, where for instance population data are used along with remote sensing images. Composite analysis using several types of census data (e.g. economic, social etc.) along with remote sensing data could lead to better understanding the phenomenon and maybe even its causes. Finally a multivariate synthesis of indices to provide one composite indicator that determines the existence and characteristics of sprawl is not discussed in this paper due to time and resource limitations, but it could be part of the future development of this research. Finally, the simplicity of proposed methodology and system of indicators enhances its applicability to planning practice, since it is available to practitioners who wish to study urban sprawl, understand its spatial manifestations, and check the success of their planning policies in containing sprawl.

4. The case study of the Greater Area of Thessaloniki, Greece The proposed indicator system was applied to the Greater area of Thessaloniki (GATh). GATh consists of 11 municipalities, extends over an area of 1,455 km2 and, according to the most recent census data (2011), its population reaches approximately a total of 1,000,000 inhabitants. The city of Thessaloniki is the main urban core of GATh, while Thessaloniki’s Urban Agglomeration (TUA) constitutes the continuous and compact urban fabric within GATh (Figure 4). Salvati considers GATh a typical “type B” sprawled Mediterranean city due to its moderate size and the fact that it has been an area of crucial spatial transformations several times throughout its modern history (Salvati, 2014). 4.1. Urban development history of Thessaloniki Thessaloniki presents an intriguing urban development course throughout its modern history, with a significant turning point set by the conflagration of 1917 (Yerolympou, 2013). A major part of the city was totally destroyed, while a plan to re-build the city was proposed and partly implemented. Until 1917 Thessaloniki was a multi-cultural compact city with a dense urban centre inside the city’s Roman walls. The city was spatially segregated in certain ethnicreligious groups (Greeks, Turks, Jewish), while residential areas outside the walls started to arise. The peculiar to Greece planning system of “antiparohi”6, the basis for city’s urban development ever since, has its origins in conflagration, along with city’s planning and land segregation tradition (Yiannakou, 2008).

6

Specifically, “Antiparohi system”, which is still valid today, is a housing development scheme where the land owner provides land for the developer and gets in return a percentage of the developed property (Yiannakou, 1993).

Figure 4: Thessaloniki: Urban Agglomeration and Greater Area

Thessaloniki’s Urban Agglomeration (TUA) Greater Area of Thessaloniki (GATh) Built Areas (Corine 2006)

Subsequently, the migration of thousands of refugees (117,000) due to the Asia Minor Catastrophe in 1922, significantly transformed the social structure and spatial form of the city (Hastaoglou - Martinidis, 1977). The ethnic-religious spatial segregation was replaced by a socio-economic spatial separation, while the refugees drastically altered population composition. Refugees settled mainly in new districts and small villages in the city’s immediate perimeter or in nearby rural villages, establishing the roots of Thessaloniki’s current polycentric entity. Actually the newly established districts were Thessaloniki’s compact urban development cores until the 1970s, while the small villages in the city’s immediate perimeter became the suburbanization cores over the past 3 decades. World War II and the Civil War that followed altered, as well, the city’s social structure. Thus Jews, a significant part of the population, vital for the city’s commercial activities, left the city or were captured and sent to concentration camps, while in-migration and out-migration completely changed the city’s structure and functionality. During the next 3 decades and especially in the 1960s and 1970s, the city was built-up intensively, using the “antiparohi” model. Thus, urban fabric became denser, any remained open space was built-up, old buildings were replaced by new ones (especially in the 1950s and 1960s due to the legislation allowing higher buildings), while the profitable exploitation of the housing market was at its peak (Kolonas, 2012). During the late 1980s and particularly in the 1990s and 2000s the monocentric compact city of Thessaloniki altered substantially. This is when the actual process of suburbanization began where residents from dense parts of the city moved towards suburbs that had become more accessible. It should be noted that in the mid-80s the ring road of Thessaloniki was completed, increasing accessibility to adjacent east and west undeveloped or semi-developed areas. Hence areas of former second housing along Thermaikos Gulf became highly desirable areas to live, mainly for middle income households that were looking to fulfil the “Greek dream” of owning a single-family semidetached house. However, voluminous and non-luxurious blocks of flats were

built along with more or less luxurious single-family dwellings, due to the significant and rapid economic growth especially in the banking sector (Kolonas, 2012; Yiannakou, 2008). At the same time there was substantial relocation of tertiary sector, along major transport axes and hubs outside the main core of the city, creating small employment cores in the peri-urban area (Kafkalas et all, 1999). Thessaloniki continued to develop along Thermaikos Gulf (figure 4), filling the inner area of GATH in a sprawled pattern, until the recent economic crisis. It now seems that this pattern has stopped or at least reduced, but there are no available data to indicate whether there is also relocation of households and businesses back to the dense part of the city of Thessaloniki (Yerolympou, 2013). 4.2. Greek planning legislation framework Greece, among other Mediterranean cities, belongs to the “Napoleonic family” of spatial planning systems, which is strictly hierarchal and applies a top-down approach from national, regional to local planning level (Newman and Thornley, 1996). Planning is performed through general guidelines, strict laws, rigid zoning, and coding ordinances leading to a multiplicity of laws and regulations, which in turn ends up in less effective implementation and control (European Commission, 1997). It should be noted that according to Greek planning legislation system, any peri-urban, ex- urban and rural areas are potentially developable land (residential, commercial etc.). For instance, residential development is allowed in any plot with minimum size of 4000m2 or 0.4 hectares in which 200m2 (0.02 hectares) can be built. Thus, despite the fact that the system is considered to be quite rigid, there are lots of exceptions and ways to bypass regulations, and along with state tolerance in regard to informal housing, development control remains a big unsolved issue. 4.3. Indicators and results Computation of indicators for GATH was performed based on 5 satellite images using 3 different land cover classes: built-up, rural and natural areas. “Natural areas” include all types of water bodies (rivers, lakes, wetlands etc.) and all forest types (high and low vegetation). Rural areas include all unbuilt land that does not belong to “natural areas”, such as crops, fruit trees etc. The method used for the classification of satellite images was the supervised classification, where 10 samples were utilized to create each class signature and the maximum likelihood method was employed to perform the final classification. The Erdas Imagine accuracy assessment for each satellite image was between 85 to 90%. The cell size from all products of satellite images was 60x60 m, due to the accuracy of the generic satellite images, while the neighbourhood size used to calculate indicators is mentioned analytically in the methodology section (section 3.1, table 1). It should be noted that the size of canvas (540x540m) used for the computation of the land cover penetration and urban fabric continuity index resulted from computations that were based on the Greek peri-urban legislation framework and the Greek peculiarity of small land holdings (Ntriankos, et al, 2015). As far as the road network, major road axes were identified and classified according to their date of full operation, corresponding to the 5 dates of available satellite images. In addition, population data for MLCs were used for 1971, 1981, 1991, 2001 and 2011. In order to determine diachronic changes three steps were performed. First, population change (figure 5) was classified based on the method described in the methodology section. It should be

noted that “mean +” standard deviation was in all intervals above 60%, so a new class between 20% and “mean +” standard deviation was defined, while “mean -” standard deviation was in all cases negative and without high variance, so no other class was defined. Subsequently, the significance of indicator results was determined by the researchers, considering the number of cells with non-zero values, the mean value and the variance for each indicator. Thereafter, the resulting maps (Figures 7&8) which indicates both population and sprawl alteration were examined to identify diachronic changes. The change of built-up area and the major road network are illustrated in Figure 6. The results of the composite analysis are depicted in Figures 7&8 representing the analysis for 4 periods: 1977-1984, 1984-1990, 1990-2001, 2001-2011 for sprawl indicators, and 1971-1981, 1981-1991, 1991-2001, 2001-2011 for population change. Figures 7&8 illustrate 5 binary variables for each of the above time intervals: the existence of significant built-up area (SBA), the existence of significant leap-frog development, linear development, peripheral development and population change. In total 59 (out of 69) MLCs showed significant changes in one or more sub-time periods, within the study period, regarding either population or sprawl. The diachronic changes of the 3 land cover classes indicate a significant change in builtup area during the study period (1977-2011) and especially in 1984-1990 and 2001-2011. GATH’s total population had also increased, while the most substantial increase occurred in 1984-1990 and 1991-2001. Figure 5: Population change box plot in the 4 intervals of the study

Thus, it is obvious that urban expansion did not occur only due to population increase. At the same time it is worth noting that TUA’s population had not altered greatly since 1981. In fact TUA experienced a decrease of its population, for the first time in its history, in the decade 20012011, when at the same time areas within GATH experienced great increases in their population numbers. More specifically, urban expansion rate index indicates a significant increase of built up area in 1984-1990 and 2001-2011, with a less significant increase in the interim period. Land cover

penetration index, considering both ways of computation (2a and 2b), presents similar results. The areas with the most significant increase are located mainly in the eastern part of GATH and along major transport axes. Furthermore, urban fabric continuity index implies a more sprawl development pattern since 1984 and especially during 1991 to 2001. Moreover urban fabric fragmentation index suggests a continuous increase of sprawl until 2001 and a small decrease in 2011. Therefore, the great increase of total built-up area in 2001-2011 occurred in continuity to the pre-existing urban patches, ending up in a less sprawled pattern, while smaller (compared to 2001-2011) increase of total built area until 2001, occurred in a substantially more sprawled development pattern. Figure 6: Built-up area change during the study period (1977-2011) in GATH

The sprawl pattern as documented by the above-mentioned indices took place either close to TUA or close to major road axes regardless of population changes. In 1971-1981 there was more than 20% population increase in MLCs located mainly in the eastern part of GATH, while these MLCs are considered to be sprawled only since 1990. Likewise, there were MLCs in the western part of GATH that experienced population increase rate of more than 20% when sprawled patterns were recorded only after 2001 or 2011. On the other hand, MLCs which never experienced significant population increase presented clear sprawl characteristics as early as 1977. These MLCs were mostly away from TUA and close to major road axes. Furthermore, from 1981 to 1991, 2 MLCs within TUA (one of them was the CBD) showed significant population decrease, indicating substantial suburbanization trends.

Spatial configuration of sprawl patterns was identified by linear development, leap-frog development and peripheral accretion indices. Linear development index indicates the importance of road network as a sprawl-inducing factor in GATH. During the study period most of the new built-up area was constructed within a buffer zone of less than 4 km along major road networks. Intensive linear development occurred in primary and secondary arterial axes towards the west part of the city and along the ring road (operated in 1980s). In the eastern part of the study area, significant linear development was identified along the axis leading to the airport and to neighbouring summer resort areas. Results of leap-frog development index indicate that it was not a prevalent pattern of development in GATH, with the exception of the last decade (2001-2011). At the same time, peripheral accretion index implies that most of the new built-up area was constructed adjacent to pre-existing ones, with the exception of 1990-2001 period, where the new built-up area presented more sprawled characteristics. All the above suggest that urban expansion since 1984 took place mainly outside TUA and was more intense in 1990-2001, while in the last decade (2001-2011) despite its leapfrogging pattern it occurred with more compact characteristics. Analysis of population patterns denotes similar characteristics. Most MLCs with sprawl experienced also a linear population increase, while there were not many cases where leap-frog development occurred without any population increase. Generally, significant population increase was accompanied by sprawled patterns, mainly at the eastern part of GATH. However, there were cases of substantial population increase (more than mean + standard deviation) where sprawl did not occur, particularly in the last decade (2001-2011). Additionally, there were MLCs in the fringes of GATH which slightly increased their population, with the exception of substantial decrease over the last study period (2001-2011), suggesting abandonment trends. To sum up, there was significant increase both in built-up area and population growth in GATH. Nonetheless, population and built-up changes did not occur proportionally or at the same location, particularly from1984 and afterwards. In 1977-1984, there was significant population growth both in TUA and GATH, when at the same time a little sprawl development was recorded. Subsequently, TUA’s population did not increase substantially; on the contrary, in 2001-2011 it experienced, for the first time in its history, a decrease. Sprawl development followed a different course in relation to population, even in the first interval, while its most important characteristics were peripheral accretion and, particularly, linear development. Linear development occurred due to major construction of new road networks (freeways) concentrating new development in adjacent areas. From 2001 more compact characteristics were recorded, which occurred due to: a) the enforcement of 1997 city planning law mainly after 2001, where most eastern municipalities adopted new General Development Plans that enabled them to expand in an organized and denser fashion, and b) the agglomeration effect of previously created urban functioning cores (mainly created in 1990-2000). Furthermore, it should be mentioned that all MLCs that were characterized as “built-up” in 1977 were part of TUA, and the new “built-up” areas created later were either part of it, as well, or in TUAs periphery. This implies a mononuclei model of development, even if significant patterns of sprawl were identified in its peri-urban and ex-urban area. Finally, none of the indicators was more significant than the others, as indicated by a principal component analysis performed using indicators results.

Figure 7: Built-up area change in 1977& 1990

Figure 8: Built-up area change in 2001 & 2011

5. Conclusions Urban sprawl, as the spatial imprint of urban growth, has been under thorough study since the 1970s. It is considered to be one of the major issues concerning modern cities both in developed and developing countries. Thus, there is a growing and probably urgent need to find ways to comprehend its origins, evolution and evaluation. Relevant studies present the impacts of urban sprawl based purely on quantitative approaches, while several studies attempt a qualitative approach. The quantitative approach is usually based on measurements of certain urban sprawl characteristics utilizing indicators. It seems that one indicator is not sufficient to measure urban sprawl phenomenon, hence a system of indicators is required where each index measures a different characteristic of the phenomenon. Thus, in this study, a system of indicators to measure urban sprawl phenomenon is proposed. This system is addressed to Greek cities which share common characteristics, even though each one of them is a distinct case. The accuracy and efficiency of the proposed indicator system is evaluated through its implementation in GATH. Indicators refer mainly to the structural characteristics of built-up areas, while population data at a different scale were also taken into account in a composite analysis. Indicators’ results confirmed existing theoretical studies concerning urban development in GATH and revealed important morphological characteristics of sprawl. Composite analysis also concluded in significant observations regarding the area and the relation between population and urban sprawl variations. In conclusion, determination of the existence of urban sprawl and designation of its spatial characteristics can be performed by applying a four step methodology that starts with the recognition of the existence of urban expansion, the detection of the degree at which urban expansion constitutes urban sprawl, the determination of its spatial formation and finally the evaluation of the proposed indicators. These four steps could be carried out by using widely accessible satellite images, which makes current methodology easily applicable. A critical issue in the composition process of the indicators system is the identification of factors (historical, economic and legislative) contributing to peri-urban development. Furthermore, performing a composite analysis using census data even at a different spatial entity substantially enriches the understanding of urban sprawl phenomenon. Finally, the exclusive use of land cover data identifies only the morphological characteristics and spatial imprint of sprawl. Therefore it is imperative, in order to perform a comprehensive analysis of sprawl, to calculate indices regarding the functional characteristics of sprawl. However, the simplicity of proposed methodology and system of indicators enhances its applicability to planning practice, since it is readily available to practitioners who wish to study urban sprawl, understand its spatial manifestations, and check their planning policies’ success in containing sprawl.

References Arribas-Bel, D., Nijkamp, P., & Scholten, H., 2011, Multidimensional urban sprawl in Europe: A selforganizing map approach. Computers, Environment and Urban Systems, 35(4), pp. 263-275. Barnes, K. B., Morgan, J. M.III, Roberge, M. C., and Lowe S., 2001, SPRAWL DEVELOPMENT: ITS PATTERNS, CONSEQUENCES, AND MEASUREMENT, Towson University. Bethemont, J., 2000, Géographie de la Mèditerranée. Paris: Colin. Bhatta, B., Saraswati, S. and Bandyopadhyay, D., 2010, Urban sprawl measurement from remote sensing data, Applied Geography, 30(4), pp. 731-740. Bruegmann, R., 2005, Sprawl: a Compact History, London: The University of Chicago Press. Duarte, M. C., Núñez, A. C., Tresserra, M., J., Mejía, C. E. and Prieto P. C., 2011, Polycentrism in the Spanish metropolitan system: an analysis for 7 metro areas, In: ERSA (European Regional Science Association), New Challenges for European Regions and Urban Areas in a Globalised World, Barcelona 30/08/2011-02/09/2011, Available at:< https://ideas.repec.org/p/wiw/wiwrsa/ersa11p119.html >, [Accessed 03/08/2015]. Chrisoulakis, N., Karamarianakis, G., Farsari, G., Diamantakis, M., Prastakos, P., 2004, Estimating urban development formulas and sustainability indicators in GIS using satellite and socio-economic data, In: IACM (Institute of Applied & Computational Mathematics), 3rd Greek GIS Conference, Available at:, [Accessed 25/06/2014] (in Greek). Colantoni, A., Grigoriadis, E., Sateriano, A., Sarantakou, E., and Salvati, L., 2016. Back to Von Thunen: a Southern European perspective on mono-centric urban growth, economic structure and non-urban land decline. International Planning Studies, pp.1-16. Couch, C., Leontidou, L. and Petschel-Held, G., 2007, Urban Sprawl in Europe: Landscapes, Land-Use Change and Policy, Oxford ; Malden, MA : Blackwell. Cutsinger, J., Galster, G., Wolman, H., Hanson, R., & Towns, D., 2005, Verifying the multi-dimensional nature of metropolitan land use: Advancing the understanding and measurement of sprawl. Journal of Urban Affairs, 27(3), pp. 235-259. EEA, 2007, Fragmentation of natural and semi-natural areas [online], EEA, Avaliable at:< http://www.eea.europa.eu/data-and-maps/indicators/fragmentation-of-natural-and-semi#toc-1> [Accessed: 3/04/2015]. Ewing, R., and Cervero, R., 2010, Travel and the Built Environment: A Meta-analysis. Journal of the American Planning Association 76 (3), pp. 265–294. European Commission, 1997, EU compendium of spatial planning systems and policies, EU. Fang, J., Shenghe, L., Hong, Y. and Qing, Z., 2007, Measuring urban sprawl in Beijing with geo-spatial indices, Journal of Geographical Sciences,17 (4), pp. 469-478. Frenkel, Α. and Ashkenazi, M., 2008, Measuring urban sprawl: how can we deal with it?, Environment and Planning B,35, pp. 56-79. Galster, G., Hanson, R., Ratcliffe, M.R., Wolman, H., Coleman, S. and Freihage, J., 2001, Wrestling Sprawl to the Ground: Defining and measuring an elusive concept. Housing Policy Debate,12,(4), pp. 681-717. Giannakourou, G., 2005, Transforming spatial planning policy in Mediterranean countries: Europeanization and domestic change, European Planning Studies, 13(2), pp. 319-331. Hamidi, S., Ewing, R., Preuss, I., & Dodds, A. 2015, Measuring Sprawl and Its Impacts: An Update. Journal of Planning Education and Research, 35(1), pp. 35-50. Hastaoglou - Martinidis, V., 1977, A Mediterranean city in transition: Thessaloniki between the two world wars, Facta Universitatis, 1(4), Available at:< http://facta.junis.ni.ac.rs/aace/aace97/aace9704.pdf>, [Accessed 05/04/2010]. Holcombe, R. G., and Williams, D.W., 2012. Urban Sprawl and Transportation Externalities. Review of Regional Studies 40 (3), pp. 257–72.

Jaeger, J.A.G., Bertiller, R., Schwick, C. and Kienast, F., 2010. Suitability criteria for measures of urban sprawl, Ecological Indicators,10,(2), 397-406. Jaret, C., Ghadge, R., Reid, L.W. and Adelman, R.M., 2009. The Measurement of Suburban Sprawl: An Evaluation, City & Community, 8(1), pp. 65-84. Johnson, M. P., 2001, Environmental impacts of urban sprawl: a survey of the literature and proposed research agenda, Environment and Planning A, 33(4), pp. 717-735. Karadimou-Yerolympou, A., 2013, The emergence of Modern Thessaloniki: history, people, places, Thessaloniki: UNIVERSITY STUDIO PRESS A.E. (in Greek), Kafkalas, G., Andrikopoulou, E., Yiannakou, A., 1999. Thessaloniki: Reducing mononuclearity in the urban agglomeration and the role of the tertiary sector. Thessaloniki: ZITI (in Greek). Kolonas, V., 2012, Thessaloniki 1912-2012: The architecture of a centenary, Thessaloniki: UNIVERSITY STUDIO PRESS A.E. (in Greek). Kostova, D. 2011, Can the Built Environment Reduce Obesity? The Impact of Residential Sprawl and Neighborhood Parks on Obesity and Physical Activity.” Eastern Economic Journal 37 (3), pp. 390– 402. Leontidou, L., 1990, The Mediterranean City in Transition: Social Change and Urban Development, New York : Cambridge University Press. Martellozzo, F. and Clarke, K.C., 2011, Measuring urban sprawl, coalescence, and dispersal: a case study of Pordenone, Italy, Environment and Planning B, 38, pp. 1085-1104. Mcdonogh, G., 2000, Mediterranean Cities, South European Society and Politics, 5(1), pp. 143-155. Mubareka, S., Koomen, E., Estreguil, C., and Lavalle, C., 2011, “Development of a Composite Index of Urban Compactness for Land Use Modelling Applications.” Landscape and Urban Planning 103 (3), 303–317. Munoz, F., 2003, Lock living: Urban sprawl in Mediterranean cities, Cities, 20(6), pp.381-385. Musakwa, W. and Van Niekerk, A., 2013, Implications of land use change for the sustainability of urban areas: A case study of Stellenbosch, South Africa, Cities, 32, pp. 143-156. Newman, P., and Thornley, A., 1996, Urban Planning in Europe: International competition, national systems and planning projects, London: Routledge. Ntriankos, I., Pozoukidou, G. and Karanikolas, N., 2015, Urban Sprawl measurement and indicators: The case study of the Greater Area of Thessaloniki, In: Department of Planning and Regional Development, University of Thessaly, 4th Greek Conference of Planning and Development, Volos, 24-27 September 2015, Volos: University of Thessaly (in Greek). Paulsen, K., 2014, Geography, Policy or Market? New Evidence on the Measurement and Causes of Sprawl and Infill, in US Metropolitan Regions. Urban Studies 51 (12), pp. 2629–2645. United Nations, 2014, World Urbanization Prospects 2014 Revision, [pdf] New York: United Nations, Available at: [Accessed 12/01/2015]. Salon, D., Boarnet, M. G., Handy, S., Spears, S., and Tal, G., 2012, How Do Local Actions Affect VMT? A Critical Review of the Empirical Evidence. Transportation Research Part D 17 (7), pp. 495–508. Salvati, L., Sateriano, A. and Bajocco, S., 2012, To grow or to sprawl? Land Cover Relationships in a Mediterranean City Region and implications for land use management, Cities, 30, pp. 113-121. Salvati, L., 2013, Towards a Polycentric Region? The Socio-economic Trajectory of Rome, an ‘Eternally Mediterranean’ City, Tijdschrift voor economische en sociale geografie, 105(3), pp. 268-284. Salvati, L. and Morelli, G. V., 2014, Unveiling Urban Sprawl in the Mediterranean Region: Towards a Latent Urban Transformation?, International Journal of Urban and Regional Research, 38(6), pp. 1935-1953. Salvati, L., Mavrakis, A., Serra, P. and Carlucci, M., 2015, Lost in translation, found in entropy: An exploratory data analysis of latent growth factors in a Mediterranean city (1960-2010), Applied Geography, 60, pp. 107-119.

Sisternes, S.D.P., Ayuga, F. and García, A.I., 2014, A method for detecting and describing land use transformations: An examination of Madrid’s southern urban–rural gradient between 1990 and 2006, Cities, 40(A), pp. 99-110. Song and Knaap, G.J., 2004, Measuring Urban Form: Is Portland Winning the War on Sprawl?, Journal of the American Planning Association,70,(2), 210-225. Sudhira, H.S., Ramachandra, T.V. and Jagadish, K.S., 2004,Urban sprawl: metrics, dynamics and modelling using GIS, International Journal of Applied Earth Observation and Geoinformation,5,(1), pp. 29-39. Terzi F. and Bolen F., 2009, Urban Sprawl Measurement of Istanbul, European Planning Studies, 17(10), pp. 1559-1570. Torrens, P.M., 2008. A Toolkit for Measuring Sprawl, Applied Spatial Analysis and Policy,1,(1), pp. 536. Xu, X. and Min, X., 2013, Quantifying spatiotemporal patterns of urban expansion in China using remote sensing data, Cities, 35, pp. 104-113. Yiannakou, A., 2008, City Planning of Thessaloniki: Ideology and implementation during the 20 th century, In: Kafkalas, G., Labrianidis, L., Papamichos, N., 2008, Thessaloniki on the edge – The city as a transformation process, Athens: KRITIKI Publishing S.A. (in Greek). Yiannakou, A., 1993, Residential land development and urban policy in Greece. The case of Greater Thessaloniki, Ph.D. Thesis, London School of Economics, University of London, Available at: < http://etheses.lse.ac.uk/1233/1/U056071.pdf>, [Accessed 29/07/2014]. Zolnik, E. J., 2011, “The Effect of Sprawl on Private-Vehicle Commuting Outcomes.” Environment and Planning—Part A 43 (8), pp. 1875–93.

Highlights    

A simple four step methodology for measuring and assessment of urban sprawl is proposed Data availability issues which usually arise in these processes are heavily considered It proposes a method to deal with different data scales, where population data are used along with remote sensing data. A simple and easy to use system of indicators is proposed in order to monitor urban sprawl and check success of planning policies by planning practitioners