ANALELE ŞTIINTIFICE ALE SCIENTIFIC ANNALS OF

UNIVERSITĂŢII „AL. I. CUZA” IAŞI “Al. I. CUZA” UNIVERSITY OF IAŞI

Tom LX, nr. 2, s. II – c, Geografie 2014 Volume LX, no. 2., s. II – c, Geography series 2014
ISSN 1223-5334 (printed version) (online version) 2284-6379 e-ISSN
http://dx.doi.org/10.15551/scigeo.v60i2.346
© Author(s) 2012. CC Attribution 3.0 License

ASSESSING THE SPATIAL DIFFERENCES AMONG SOME URBAN

EXPANSION DRIVING FORCES IN CONSTANȚA METROPOLITAN
AREA (ROMANIA)

¹Ema CORODESCU, ²Cătălin CÎMPIANU

Faculty of Geography and Geology, Alexandru Ioan Cuza University of Iași

¹ema.corodescu@gmail.com,² catalin.cimpianu@yahoo.com

Abstract: The urban expansion process has become a major challenge for the scientists and urban
planners, mainly due to its various spatial and functional expressions. The present study aims to
identify the spatial patterns of several driving forces of this phenomenon in Constanta
metropolitan area. In order to reach our purpose, multi-temporal remote sensing data was
employed; by image processing, we obtained the built-up area for 2001 and 2013.
Subsequently, a Geographically Weighted Regression model was employed to explore the
relationship between the newly-built-up area and four driving forces related to accessibility:
distance to the shoreline, distance to the existing built-up area, distance to the road network
and distance to Constanta city center. By studying the spatial distribution of the regression
coefficients for the explanatory variables we mapped the relevance of each driving force,
leading to the general conclusion that the strongest driving forces are the distance to the built-
up area and the distance to the road network. At the same time, the spatial distribution of the
correlation coefficient and of the residuals revealed that the model best fits the axial and
suburban areas, while the internal, consolidated areas record highly deviated residuals.

Keywords: urban expansion, supervised classification, built-up area, driving

forces, Geographically Weighted Regression

I. INTRODUCTION
The urbanization phenomena have been suffering a severe shift from a slow,
compact and population growth-oriented development, to a much more complex
process, driven by various forces, which became “the most powerful and visible
anthropogenic force that has brought about fundamental changes in land use and
landscape pattern across the globe”. Consequently, a large amount of research
focused on defining and analysing the spatial impacts of various growth types(Shi et
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

al, 2012): compact growth (Jenks et al., 1996), (Holden and Norland, 2005) linear
development, infill development (Muller et al., 2010), urban sprawl (Bhatta, 2010),
leap-frog expansion (Ewing, 2008) etc. Various driving forces stay behind these
growth types, creating different patterns across the globe; however, the most evident
type of extension is urban sprawl, which is more and more observed not only in
North American cities – where it was firstly and widely refferred – but also in various
European fragmented, functionally diversified and low-density city fringes (Besussi
et al., 2010). At the same time, the urban sprawl gradually becomes an universal
process, affecting small settelments and even remote villages, besides the large urban
centers (Antrop, 2004); the complexity of the urban expansion phenomenon is also
highlighted by the mismatch between population decline and increase in the built-up
area for some European cities (Kasanko, et al., 2006).
Most of the researches conducted on the urban expansion theme rely on
advanced remote sensing techniques. The availability of multi-temporal remote
sensing data, including long term satellite data is of great importance for
environmental monitoring (Gomarasca, 2009). Nowadays, earth observation satellite
imagery serves as a promising source to extract the land use/land cover classes.
Remote sensing is cost effective and technologically sound, so is increasingly used
for the analysis of urban sprawl (Sudhira, et al., 2004; Yang and Liu, 2005; Haack
and Rafter, 2006). Mapping using temporal satellite data has been successfully used
in the process of monitoring urban land dynamics for large metropolitan areas. These
databases are very useful in recognizing urban growth patterns, and understanding
its profound changes on the landscape (Kamini et. al., 2006). Managing urban
growth by conventional methods is a slow process, and by the time information
becomes available to planners, it is already outdated and redundant since the damage
has already been done. Satellite remote sensing data and application of GIS
technologies provide an alternative means of rapidly assessing the dynamics and
development of sprawl for optimal spatial actions (Farooqand Ahmad, 2008).
Various studies (Taubenböck et. al., 2012, Karolien et. al., 2012, Inostroza et. al.,
2013) have used Landsat satellite data for mapping urban extent and its changes,
using either high resolution images (Brigante and Radicioni, 2014) or medium
resolution images (Haque and Bandyopadhyay, 2012), (Cîmpianu and Corodescu,
2013); hence, remotely sensed data and techniques are increasingly used in projects
aiming the operational monitoring of the effects of urbanization.
The efficient monitoring of these effects mostly rely on a thorough
understanding and quantitative analysis of the driving forces shaping the urban
expansion. In this context, many studies focused on exploring these driving forces
of urban expansion, leading to a general classification into four types of factors (Li
et al., 2013): physical, socioeconomic, neighborhood and urban planning factors.

78
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

Each type include a wide range of sub-factors, approached by different authors;

nevertheless, this list of factors can be permanently updated, as new driving forces
are discovered. At the same time, urban expansion represents a complex synergistic
process driven by these factors (Pătroescu, et al., 2012), which may face a strong
temporal variability (Reilly et al., 2009; Li et al., 2013). Consequently, it is difficult
to identify all the factors influencing the urban expansion process, as the availability
of data is limited as well as the local features are very different across the world.
Nevertheless, there are several factors took into consideration by most of the studies
considered in review articles: distance to roads, distance to urban centers, slope and
even zonning or environemntal factors (Santé et al., 2010). There are also studies
considering the distance to built-up space as a key driving force and including
different restrictions related to land suitability (Santé et al., 2010; Vermeiren et al.,
2012).The quantitative assessment of the role played by each potential factor in the
urban expansion process was approached through a variety of methods (Li et al.,
2013), among which the widest used are simple regression (Ma and Xu, 2010),(Ye
et al., 2011), multiple regression (Dewan and Yamaguchi, 2009) and logistic
regression (Dubovyk et al., 2011; Vermeiren et al., 2012).
The present study tries to present a different approach, focusing on the
spatial variability of the driving forces across the study area. Consequently, a
Geographically Weighted Regression (GWR) model is employed to analyze four
accessibility-related driving forces of the urban expansion: distance to road network,
distance to the city center, distance to the built-up area and distance to the shoreline.
The first three factors are frequently used in the above mentioned studies; the fourth
factor – distance to the shoreline – was considered very important as the present
study approaches a seaside city. At the same time, even if it is widely used, the
topographic setting – slope, elevation etc. – was excluded, as Constanta Metropolitan
Area overlaps a lowland area, recording insignificant differences in topography. The
detailed objectives of the study are: (1) mapping the urban expansion in Constanta
Metropolitan Area between 2001 and 2013 through remote sensing data; (2)
analyzing the spatial pattern of four driving forces of urban expansion through GWR;
(3) creating a possible general GWR model of urban expansion for the area.
Our hypothesis is that Constanta Metropolitan Area faced a particular urban
expansion pattern, specific to coastal urban areas, where the harbor and the satellite
tourist resorts played a structuring role in the urban expansion process. Hence, both
Constanta City and Constanta Metropolitan Area faced a polycentric development.
The present study attempts to explore four accessibility related driving forces of
urban expansion; the spatial distribution of the GWR coefficients accompanying
these driving forces is aimed at verifying our hypothesis of a particular polycentric
urban expansion pattern.

79
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

II. METHODOLOGICAL FRAMEWORK

II.1. Study area
According to the Romanian geographical literature which defines
metropolitan areas as “spaces under the influence of urban centers that have macro-
regional functions and whose population exceeds 1 million people” (Erdeli et al.
2001), only one metropolitan area (Bucharest) falls into this category. Nevertheless,
the metropolitan development in Romania was encouraged by some provisions
introduced in the legislation, according to which a metropolitan area could be
established based on the joint character of administrative-territorial structures
through association, by voluntary partnership between the main urban centres (the
capital city of Romania and the first-rank municipalities) and adjoining the urban
and rural settlements situated at distances up to 30 km, that established cooperation
relations at different levels (Law no. 351/ 2001, National Territory Management
Plan, Section IV – Settlements, Law no. 350/2001, Ordinance no. 53/2002 and Law
no. 286/2006) (Grigorescu et al., 2012).
Established in 2007 and with a total population of 425.916 inhabitants,
Constanţa metropolitan area developed an urban system based on such legislative
context which supports metropolitan development based on the joint character of the
administrative units under the influence of a city. It includes the municipality of
Constanţa, the towns of Năvodari, Ovidiu, Eforie, Murfatlar, Techirghiol and 8
communes: Mihail Kogălniceanu, Cumpăna, Valul lui Traian, Lumina, Tuzla,
Agigea, Corbu and Poarta Albă, consisting of 16% of Constanţa County area (Fig.
1).
Over the last years, the Romanian cities have registered significant
dynamics, facing a built-up area expansion of up to 200% (e.g. Arad 60%, Iaşi
73,7%, Suceava 76%, Mihaileşti 106,3%, Bragadiru 114,6%, Buftea 106% etc.) or
even more (e.g. Măgurele 872,4% in the Bucharest Metropolitan Area) due to their
position in the proximity of important urban centers (Suditu et al., 2010). The city of
Constanta and its surroundings make no exception, and its important harbor-trade
and touristic functions enhanced the urban expansion process.

80
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

Fig.1.The study area location

II.2. Data sources and overall methodology
The methodology used for the present study included three main steps: data
collecting, image processing and statistical analysis using GIS. For the first step,
open-source data was employed (Table 1): the land cover for 2001 and 2013 resulted
from satellite images processing, provided by USGS, while the vector data used in
the statistical analysis was collected from Geo-spatial portal and Open Street Map.

Table 1. The data sources

Data type Description Data source
Satellite images Landsat ETM+ (row 29, path 181; 25-05- USGS
2001) Landsat 8 (row 29, path 181; 02-05- http://www.usgs.gov/
2013)
Polygon Territorial Administrative Units - NUTS 5 http://earth.unibuc.ro/
Shapefile level
Point Shapefile Rural and urban entities http://earth.unibuc.ro
Line Shapefile Open Street Map, excepting paths and http://www.geofabrik
pedestrian .de/
81
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

The next step consisted in image processing, in order to obtain the two land
covers (for 2001 and 2013), subsequently used in assessing the urban expansion
during this period. Finally, four driving forces were introduced in a GWR model, as
an attempt to explain the urban expansion in Constanta Metropolitan Area.

II.3. Image processing

With the opening of Landsat archive by USGS in 2009, a huge amount of
images have become available (Woodcock et al., 2008), causing an increasing need
for automatic image preprocessing algorithms (Huang et al., 2009). Landsat ETM+
(row 29, path 181) (25-05-2001) and Landsat 8 (row 29, path 181) (02-05-2013)
images were used to create build-up cover maps for the Constanța metropolitan area,
covering a 45 km × 70 km area.
The Landsat ETM+ image consisted of 8 spectral bands with spatial
resolution of 30 m for bands 1 –5, 7; 60 meters for band 6 and 15 meters for band 8.
The Landsat 8 image consisted of 11 spectral bands with spatial resolution of 30
meters for bands 1 – 7, 9; 15 meters for band 8 and 100 meters for bands 10 – 11.
Only the reflective bands 1 – 7 and 8 were used for this study. Image processing
required three steps: pre-processing, supervised classification and accuracy
assessment (Fig. 2).
Pre-processing of the images consisted of two steps radiometric and
atmospheric corrections using the dark object subtraction method. (Chavez, 1996;
Song et al., 2001). Radiometric Calibration is a multi-step process that involves the
use of standard equations to convert 8-bit satellite digital numbers (DN) to at-satellite
reflectance. Landsat 7 and 8 images were first converted to at-satellite radiance using
Eq. (1).
Lmaxλ−Lminλ
Lλ = × (QCAL − QCALmin) + Lminλ(1)
QCALmax−QCALmin

Where Lλ – spectral Radiance at the sensor's aperture; QCAL – the

quantized calibrated pixel value; Lminλ – the spectral radiance that is scaled to
QCALmin; Lmaxλ – the spectral radiance that is scaled to QCALmax; QCALmin –
the minimum quantized calibrated pixel value (= 0 for 1989 image and =1 form 2004
image); QCALmax – the maximum quantized calibrated pixel value (=255)
(Landsat-7 Science Data Users Handbook). After conversion to at-satellite radiance,
each image was converted to at-satellite reflectance (assuming a uniform Lambertian
surface under cloudless conditions) using Eq. (2).

π×Lλ×d2
Pλ = ESunλ× cos θ
(2)

82
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

Where Pλ – Unitless planetary reflectance; Lλ – spectral radiance (Equation

1); d – Earth-Sun distance; Esunλ – mean solar exoatmospheric irradiances; θ –
solar zenith angle.

Fig.2.The workflow performed for image processing

In different applications (e.g. determination of ground values of reflectance,
albedo and temperature; extraction of biophysical parameters; some applications of
change detection), it is recommended to correct the images from the atmospheric
effects (Gomarasca, 2009). Dark Object Subtraction (DOS) method assumes that
within a satellite image there exist features that have near-zero percent reflectance
(i.e., water, dense forest, shadow), such that the signal recorded by the sensor from

83
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

those features is solely a result of atmospheric scattering (path radiance), which must
be removed (Chavez, 1996).
After the pre-processing steps, in order to improve the image resolution a
pan-sharpening method was performed. (Pohl and Van Gendeeren, 1998). Over the
years, various image fusion processes have been developed for improving spatial and
spectral resolutions of remote sensing data (Choi, 2006; Karathanassi et al., 2007).
For this study the Gram-Schmidt method was used (Laben et al., 2000). The Gram-
Schmidt method is used to sharpen multispectral data using high spatial resolution
panchromatic data. The Gram-Schmidt spectral sharpening method enhances the
spatial resolution of the MS image by merging the high resolution PAN image with
the low spatial resolution. The high resolution PAN is simulated from the MS having
lower spatial resolution (Yuhendra et al., 2011). Then, the Gram-Schmidt
transformation is performed on the simulated PAN and MS, where the simulated
PAN is employed as the first band. The high spatial resolution PAN is replaced with
the first GS band. Finally, the inverse Gram-Schmidt transformation is applied to
obtain the pan-sharpened spectral bands (Laben et al., 2000).
Image classification in the field of remote sensing, is the process of assigning
pixels or the basic units of an image to classes (Palaniswami et. al., 2006). Remotely
sensed images from aerial and space borne sensors are a powerful tool to quickly
identify the land use/land cover classes by utilizing some classification algorithms.
Advanced classification algorithms have been developed in the last decade for
enhancing the efficiency and accuracy of the classification process (Varshney et. al.,
2013). The statistical method employed for the present study is the Maximum
Likelihood Classifier, a supervised procedure that highly depends on the cognition
and skills of the image specialist and the ability to recognize conventional classes in
a scene from prior knowledge, such as, personal experience with the region, by
experience with thematic maps, or by on-site visits. Maximum Likelihood
Classification uses the training data (data selected by the image specialist for every
class) by means of estimating means and variances of the classes, which are used to
estimate probabilities and also consider the variability of brightness values in each
class; therefore a statistical processing in which every pixel is compared with the
various signatures and assigned to the class whose signature comes closest (Perumal
et. al., 2010).

II.4. Statistical analysis

The final step consisted in an advanced GIS analysis, using the data obtained
through the first two steps. The entire workflow is presented in Fig. 3.
First of all, the vector representing the built-up area in 2001 was clipped by
the one corresponding to 2013, in order to get the newly-built-up area during this
84
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

period. The new vector was spatially-joined to a rectangular grid having a cell of 10
ha, so that the newly-built-up area was assigned to every grid cell. Subsequently,
four Euclidean distances were calculated from each grid-cell to: the shoreline, the
existing built-up area, the road network and city center of Constanta.

Fig.3.The ArcGIS workflow performed for the statistical analysis

The necessary data being prepared, we started the statistical analysis, by
testing different regression models for our study, through Exploratory Regression
Tool in ArcGIS 10.2. The results of this test (Table 2) lead to the conclusion that the
best model would include all the four variables taken into consideration, as it
recorded the highest correlation coefficient: r=0.4225, for 541 observations.

Table 2. Comparison between 9 exploratory regression models

Model No. of No. of
no. observations r AICc variables
1 541 0.1976 31630.1462 1
2 541 0.1919 31632.8533 1
3 541 0.2391 31608.1071 1
4 541 0.3437 31530.5697 1
5 541 0.2869 31577.7223 2
6 541 0.3391 31535.6962 2
7 541 0.4183 31454.2156 2
8 541 0.3429 31533.2576 3
9 541 0.4225 31451.3024 4

85
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

In order to assess the spatial pattern of these driving forces, we run a 50

neighbors GWR model, including all the four driving forces. The GWR model was
chosen in order to meet the main issue of the paper i.e. assessing the spatial
differences in urban expansion driving forces. The GWR coefficients vary across the
study area, thus highlighting these differences.

III. RESULTS AND DISCUSSIONS

III.1. Urban expansion mapping
The first result of the present study consists in two datasets, representing the
built-up space in 2001 and 2013 (Fig. 4), which subsequently permitted to calculate
and spatially model the newly-built-up space. The urban expansion process was very
active – 4075 ha of newly-built-up space, namely approximately 48% of the built-
up area in 2001. Nevertheless, this average value hides important spatial differences,
as the urban expansion was mainly compact (infill development and adjacently to
the existing built-up space) or linear.
The most visible expansion of the built-up space followed a linear pattern
and affected the western part of the metropolitan area (Fig. 4), forming a half-ring
around the city of Constanța (the A4 highway, representing Constanța ring-road) and
a transversal segment (belonging to A2 – Bucharest-Constanța highway). These
massive road infrastructures created extended built-up areas, not only by themselves,
but also by producing ancillary areas, especially at different stops.
Compact development – including new patches reaching the edge of the
existing ones – was the also important, affecting some bigger, high density
settlements, such as Constanta (on west, south-west and north-west sides), Ovidiu,
Lumina, Năvodari; this development type was mainly related to residential or
commercial functions.
The infill expansion – consisting in internal densification of the built-up
space – was specific to lower density settlements, particularly to several sea resorts
which encountered an increase in their attractiveness throughout the studied period:
Corbu, Mamaia, Agigea, Eforie Nord, Eforie Sud.

86
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

According
to these tendencies,
we can delineate a
more general pattern
of urban expansion
(at 10 ha-grid level)
(Fig. 5); this map
shows the density of
newly-built-up space
and highlights a
strong heliotrope
expansion tendency, Fig.4.The density of the newly-built-up area
as the highest density
values (above 4988
m2/ha) are
concentrated along
the sea or the lakes
shores. High and
medium values
overlap the areas
having faced a linear
pattern of extension,
as well as central
areas of the smaller
settlements. Low
densities are specific
for most of the
western, continental
part of the
metropolitan area.
Fig.5.Built-up area in 2001 and 2013

III.2. Spatial patterns of the urban expansion driving forces

The four driving forces selected for the present paper – distance to the shore,
distance to the existing built-up area, distance to the roads and distance to the city
center – do not follow a uniform pattern. Consequently, the Geographically
Weighted Regression (GWR) coefficients (corresponding to the variables slopes for
the most suitable model, explained in the methodology) were mapped, revealing the
spatial differences related to the weight of each factor (Fig. 6).
87
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

The first driving force – the distance to the shoreline (Fig. 6A) recorded
negative values for most of the area, as this explanatory variable is inversely
correlated to the urban expansion. The strongest inverse correlation is recorded in
several hot-spots: Eforie-Techirghiol area – important sea resorts, Ovidiu and
Constanța-Murfatlar axis. The medium and low values are sparsely distributed all
over the study area, creating some groups in northern and western part of the
metropolitan area, as well as along the shoreline, between Constanța and Eforie
Nord, where the tourist activity is mixed with industrial or harbor related functions.
The direct correlation affects two categories of space: areas already having a high-
built-up density in 2001 and facing a strong real estate pressure (Constanța Nord,
Mamaia, Ovidiu) and quite remote areas, situated far away from the seashore – where
the heliotropism no longer acts.
The distance to the existing built-up area (Fig. 6B) is the second driving
force taken into consideration and it records the highest absolute values, suggesting
its major role in the urban extension process. It also represents an important evidence
of the general compact pattern of urban expansion. The correlation is inverse too, as
the coefficients are mainly negative, except several cases. The lowest values
correspond to areas where the proximity to existing built-up space was a crucial
factor of the urban extension (Constanța, Năvodari, Murfatlar etc.). Higher values
are recorded in less urbanized areas, where the low density of the existing built-up
space reduced the influence of this factor; most of these areas overlap intermediate
areas, situated between the major roads. Finally, there are several grid cells where
the correlation is positive; hence, the expansion having taken place here mainly
belonged to urban sprawl type.
The third factor – distance to roads (Fig. 6C) is also weighted by a negative
coefficient in most of the cases. The strongest relationship is concentrated along to
major axes: the first one, situated along the shore, creating a buffer zone around the
highway linking the coastline settlements and the second one, corresponding to the
Murfatlar-Constanta axis. Intermediate negative values of this coefficient are
distributed mostly in peripheral zones. The last class – recording positive values for
the coefficient has a tendency to fill the intermediate areas, situated between the main
roads.

88
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

Fig.6.The GWR coefficients (variables slopes) spatial distribution

89
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

The last factor – distance to the city center (Fig. 6D) – forms the most
heterogeneous spatial pattern, recording both positive and negative values. This
pattern may suggest either a strong spatial variability of the relationship or a
relatively low correlation between newly-built-up space and this factor. In the first
case, the spatial variability is difficult to explain, as the values do not seem to follow
a logical spatial pattern at our scale of analysis; hence, the coefficient values may be
related to other local factors or even to the randomness. The second case, assuming
that this factor is not a relevant driving force of the urban expansion is considered to
be more realistic, as it records the lowest coefficients among the 4 factors and as the
general spatial configuration of the area tends to a polycentric pattern, so
reconsidering this driving force on a different form (distance to certain facilities for
instance) may be more relevant.
As for the general model, the values recorded by the coefficients confirm a
strong relevance for two of the chosen factors: distance to the existing built-up space
and distance to roads, a good relevance for the distance to the shore and a lower
reliability for the distance to the city center. The most powerful driving forces –
distance to built-up space and distance to roads – have a very alike distribution in
their coefficients, enforcing the idea that they action as a couple of determinant
forces of urban expansion.

III.3. A general model of the urban expansion for Constanța

Metropolitan Area
The general GWR model is aimed at exploring the combined explanatory
force for the four studied variables. The spatial distributions of two parameters are
very important: Bravis-Pearson correlation coefficient and the residuals standard
deviation.
Figure 7A shows the correlation coefficient, whose distribution is
asymmetrical, as most of the values are concentrated in the last class (0,415-0,797),
describing a strong correlation. These values cover urban and sub-urban areas, as
well as areas situated along the main roads. Hence, these are the areas where the
GWR model works the most effectively. Other areas record intermediate values
(0.255-0.414), suggesting that the present model is not able to explain the urban
expansion, mainly due to the presence of other local driving forces. Such areas are
situated either in the gaps between the main roads or the built-up areas or in some
sub-urban areas (Mamaia, Agigea, Ovidiu, Navodari).

90
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

Fig. 7.The spatial distribution of the local correlation coefficient (A) and of the GWR
residuals (B)
As far as the residuals are concerned, a Moran’s Index was computed,
recording a value of -0.056, with a z-score of -3.29, revealing a dispersed pattern.
Nevertheless, by examining the spatial distribution of the residuals standard
deviation (Fig. 7B), the appearance of the classified data highlights several local
clustering tendencies. The first class (represented in green) contains negative
residuals, where the predicted values are lower than the observed values with more
than one standard deviation. These are areas where the model underestimated the
intensity of urban expansion; such areas are mainly situated peripherally to the main
built-up clusters (west of Constanta, south of Agigea, north of Ovidiu and Murfatlar
etc.). In these cases, other local factors, related to the general attractiveness of certain
areas enhanced the urban expansion, besides the four considered driving forces. The
second class is not very numerous and its individuals are randomly distributed across
the study area, without creating any regular spatial pattern. The third class is the most
numerous and contains the individuals best fitting the model: areas situated along
the main roads and the proximity of the smaller settlements, where the accessibility,
the proximity to built-up space and the availability of open space favored the urban
expansion. The last two classes contain the positive residuals, where the predicted
91
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

values are higher than the actual values. These areas are almost exclusively
distributed inside the compact built-up areas, overlapping the core of some
settlements (Constanta city, Murfatlar and Navodari towns, Cumpana etc.). The
explanation of their presence is related to the unavailability of open space to build,
instead of their favorable situation concerning the four analyzed driving forces.
Hence,these areas have already faced a building saturation and cannot be subject to
further urban expansion anymore.

IV. CONCLUSIONS
Urban expansion is an important spatial tendency in Europe, including
Romanian urban areas (Suditu, et. al, 2010). This study demonstrates the use of remote
sensing and GIS to analyze the urban sprawl mapping and detect changes in urban
morphology through different years. Satellite data are found to be useful in mapping
and quantifying the extent of urban areas playing an important role in the urban
planning and in the decision making process. This project, presented in that article, is
a successful example and shows a simple but powerful solution to developing and even
developed countries which are dealing with a lack in up-to-date in geographic data.
This approach revealed an active urban expansion process in Constanta
Metropolitan Area, as the built-up area has increased by 48% in twelve years. This
evolution has been shaped by two particular features: a great part of the newly built-
up area was related to major transport infrastructures as well as to extensive, industrial,
commercial or compact residential plots developed in the peripheral urban areas and
the key role in this evolution was played by accessibility and connectivity. At the same
time, the newly-built-up area configuration reconfirmed the structure of the study area
at two different spatial scales: on the one hand, at the national level, Constanta city and
its metropolitan area represent a satellite of the Bucharest city, as a trade and touristic
hub – therefore the motorway and its ancillary developments considerably marked out
the area – and, on the other hand, at its internal level the shoreline and axial tropisms
are very noticeable. In this reticular context, the four variables analyzed – distance to
the shoreline, distance to the built-up area; distance to the roads and distance to the city
center – were all very relevant and managed to create a significant correlation
coefficient for most of the area. Nevertheless, the model can be substantially improved
by introducing other explanatory variables, leading to a more complete approach,
which may serve as a starting point for a future modeling process.
The GWR approach allowed a sensitive analysis on the spatial variation of the
relative importance of each driving force, thus offering a more detailed view over the
urban expansion process in this metropolitan area. The relevance of such an approach
– which can be substantially improved by testing and possibly introducing a higher
number of variables – is obviously related to its capacity of bringing to light the local
92
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

factors influencing the urban expansion process. Thus, the spatial distribution of the
GWR coefficients can serve as a diagnosis tool for the planners, who can subsequently
elaborate adequate, territorially-adapted policies or regulations in order to favor or to
dispirit certain types of future developments, leading to a more efficient urban
expansion management. Such measures are urgently needed and their scientific
expertise background is crucial, as the shift to the free market economy, accompanied
by the economic growth faced after 2000 lead to uncontrolled, difficult to monitor
urban developments (Pătroescu et al, 2012).

References

Antrop, M.: Landscape change and urbanization process in Europe. Landscape and Urban
Planning 67, 9-26, 2004.
Bhatta, S.: Urban sprawl measurement from remote sensing data. Applied Geography 30, 731-
740, 2010.
Besussi, E., Chin, N., Batty, M., & Longley, P.: The structure and form of urban and suburban
areas. În T. Rashed, & C. Jürgens, Remote sensing of urban and suburban areas (pg.
13-31). Dordrecht: Springer, 2010.
Brigante, R., &Radicioni, F.: Use of multispectral sensors with high spatial resolution for
territorial and environmental analysis. GeographiaTechnica 9(2), 9-20, 2014.
Chavez, P. S.: Image-based atmospheric corrections - revisited and improved, Photogrammetric
Engineering and Remote Sensing, 62, 1025-1036, 1996.
Choi, M.: A new intensity-hue-saturation fusion approach to image fusion with a tradeoff
parameter. IEEE Trans. Geosci. Remote Sens. 44, 1672–1682, 2006.
Cîmpianu, C., &Corodescu, E.: Landscape dynamics analysis in Iași Metropolitan Area
(Romania) using remote sensing data. Cinq Continents 3 (7), 18-32, 2013.
Dewan, A., & Yamaguchi, Y.: Land use and land cover change in Greater Dhaka, Bangladesch:
Using remote sensing to promote sustainable urbanization. Applied Geography 29,
390-401, 2009.
Dubovyk, O., Sliuzas, R., &Flacke, J.:Spatio-temporal modelling of informal settlement
development in Sancaktepe district, Istanbul, Turkey. ISPRS Journal of
Photogrammetry and Remote Sensing 66(2), 235-246, 2011.
Erdeli, G., Cândea, M., Braghină, C., Costachie, S., Zamfir, D.:Dicționar de geografieumană,
Edit. Corint, București, 2001.
Ewing, R.: Characteristics, Causes, and Effects of Sprawl: A Litterature Review, 2008.
Farooq S, Ahmad, S.: Urban Sprawl Development Around Aligarh City: A Study Aided by
Satellite Remote Sensing and GIS., J. Indian Soc. Remote Sens., 36: 77-88, 2008.
Gomarasca, M.: Basics of Geomatics. London: Springer Science, 2009.
Grigorescu I., Mitrică B., Mocanu I., Ticană N.: Urban sprawl and residential development in
the Romanian metropolitan areas Rev. roum. géogr./rom. journ. geogr., 56, (1), p. 43–
59, 2012.

93
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

Haack, B. N. and Rafter, A.: Urban growth analysis and modeling in the Kathmandu Valley,
Nepal, Habitat International, vol. 30, no. 4, pp. 1056–1065, 2006.
Haque, S., Bandyopadhyay, S.: Identification of metropolitan core using geo-spatial data for
Kolkata, India. Scientific Annals of Alexandru Ioan Cuza University of Iași, Vol.
LVII, s.II -c, Geography series, 185-206, 2012.
Holden, E., Norland, I.: Three Challenges for the Compact City as a Sustainable Urban Form:
Household Consumption of Energy and Transport in Eight Residential Areas in the
Grater Oslo Region. Urban Studies 42, 2145-2166, 2005.
Huang, C., Goward, S.N., Masek, J.G., Gao, F., Vermote, E.F., Thomas, N., Schleeweis, K.,
Kennedy, R.E., Zhu, Z., Eidenshink, J.C.: Development of time series stacks of
Landsat images for reconstructing forest disturbance history, International Journal of
Digital Earth 2, 195–218, 2009.
Inostrozaa, L., Baurb, R., Csaplovics, E.: Urban sprawl and fragmentation in Latin America: A
dynamic quantification andcharacterization of spatial patterns. Journal of
Environmental Management 115, 87-97, 2013.
Jenks, M., Burton, E., Williams, K.: The Compact city: a sustainable urban form? E&N FN
Spon, 1996.
Kamini, J., Jayanthi, S., Raghavswamy, V.:Spatio-temporal analysis of land use in urban
Mumbai - using multi-sensor satellite data and gis techniques, Journal of the Indian
Society of Remote Sensing, Wol. 34, No. 4, 2006.
Kasanko, M., Barredo, J. I., Lavalle, C., McCormick, N., Demicheli, L., Sagris, V.: Are
European cities becoming dispersed? A comparative analysis of 15 European urban
areas.Landscape and Urban Planning 77 (1-2), 111-130, 2006.
Karathanassi, V., Kolokousis, P., Ionnidou, S.: A comparison study on fusion methods using
evaluation indicators. Int. J. Remote Sens. 28 (10), 2309–2341. 2007.
Laben, C.A., Bernard, V., Brower, W.: Process for enhancing the spatial resolution of
multispectral imagery using pan-sharpening. US Patent 6011875 A, 2000.
Li, X., Zhou, W., Ouyang, Z.: Forty years of urban expansion in Beijing: What is the relative
importance of physical, socioeconomic, and neighborhood factors. Applied
Geography 38, 1-10, 2013.
Ma, Y., Xu, R.: Remote sensing monitoring and driving force analysis of urban expansion in
Guangzhou City, China. Habitat International 34, 228-235, 2010.
Muller, K., Steinmeier, C., Kuchler, M.: Urban growth along motorways in Switzerland.
Landscape and Urban Planning 98, 3-12, 2010.
Pătroescu, M., Iojă, C., Rozylowicz, L., Vânău, G., Niță, M., Pătroescu-Klotz, I.:Evaluarea
integrată a calității mediului în spații rezidențiale. București: Editura Academiei
Române, 2012.
Palaniswami, C., Upadhyay, A. K. and Maheswarappa, H. P.: Spectral mixture analysis for
subpixel classification of coconut, Current Science, Vol. 91, No. 12, pp. 1706 -1711,
25 December 2006.
Perumal, K., Bhaskaran, R.: Supervised classification performance of multispectral images,
Journal of computing, volume 2, issue 2, issn 2151-9617, february 2010.

94
 EMA CORODESCU, CĂTĂLIN CÎMPIANU

Pohl, C., Van Genderen, J.L.:Multisensor image fusion in remote sensing: concepts, methods,
and applications. Int. J. Remote Sens. 19 (5), 823–854, 1998.
Reilly, M., O’Mara, M., Seto, K.: From Bangalore to the bay area:comparing transportation and
activity accessibility as drivers of urban growth. Landscape and Urban Planning 92(1),
24-33, 2009.
Santé, I., Garcia, A., Miranda, D., Crecente, R. : Cellular automata models for the simulations
of real-world urban process: A review and analysis. Landscape and Urban Planning
96, 108-122, 2010.
Shi, Y., Sun, X., Zhu, X., Li, Y., Mei, L.: Characterizing growth types and analyzing growth
density distribution in response to urban growth patterns in peri-urban areas of
Lyanyungang City. Landscape and Urban Planning 105, 425-433, 2012.
Song, C., Woodcock, C., Seto, K. C., Lenney, M. P., Macomber, S. A.: Classification and
change detection using Landsat TM Data - when and how to correct atmospheric
effects? Remote Sensing of Environment, 75, 230-244, (2001).
Sudhira, H. S., Ramachandra, T. V., Jagadish, K. S.: Urban sprawl: metrics, dynamics and
modelling using GIS, International Journal of Applied Earth Observation and
Geoinformation, vol. 5, no. 1, pp. 29–39, 2004
Suditu, B., Ginavar, A., Muică, A., Iordăchescu, C., Vârdol, A., Ghinea, B.: Urban sprawl
characteristics and typologies in Romania, Journal of Studies and Research in Human
Geography, vol. 4, issue 2, 79–87, 2010.
Taubenböck, H., Esch, T., Felbier, A., Wiesner, M., Roth, A., Dech, S.: Monitoring
urbanization in mega cities from space. Remote Sensing of Environment 117, 162–
176, 2012
Vermeiren, K., Van Rompaey, A., Loopmans, M., Serwajja, E., Mukwaya, P.: Urban growth
of Kampala, Uganda: Pattern analysis and scenario development. Landscape and
Urban Planning 106, 199–206, 2012.
Varshney, A., Rajesh E.: A Comparative Study of Built-up Index Approaches for Automated
Extraction of Built-up Regions From Remote Sensing Data. Indian Society of Remote
Sensing. DOI 10.1007/s12524-013-0333-9, 2013.
Woodcock, C.E., Allen, R., Anderson, M., Belward, A., Bindschadler, R., Cohen, W.,Gao, F.,
Goward, S.N., Helder, D., Helmer, E.: Free access to Landsatimagery. Science 320,
1011–11011, 2008.
Yang, X., Liu, Z.:Use of satellite-derived landscape imperviousness index to characterize urban
spatial growth, Computers, Environment and Urban Systems, vol. 29, no. 5, pp. 524–
540, 2005.
Ye Y., Zhang H., Liu K., Wu Q.: Research on the influence of site factors on the expansion of
construction land in the Pearl River Delta, China: by using GIS and remote sensing.
International Journal of Applied Earth Observation and Geoinformation, 2011.
Received: 30.11.2014
Revised: 19.02.2015
Accepted: 30.02.2015
Published: 20.03.2015

95