. Istanbul 2004 
area Or “peri- 
| metropolitan 
r the purposes 
that defined by 
IMAMB. The 
on the premise 
' questions of 
rvices between 
1e association. 
, one finds the 
ral space, the 
onservation of 
ring a total of 
vision of basic 
| coastline; the 
vide range of 
increasing the 
:elona. In view 
nsibilities, the 
ults emanating 
he Universitat 
olítica de Sól i 
of specialised 
land use and 
ement, urban 
systems, the 
e city, "and 
emerging from 
pplication and 
in July/August 
odern science, 
ment, through 
ery and aerial 
by NASA in 
satellite-based 
e creation of 
ng planners a 
ban growth, as 
| management 
nparable study 
LAND project 
Use Changes) 
of which have 
Agency (EEA, 
of European 
ban dynamics, 
for a range of 
ao. These data 
ic information, 
characteristics 
related to land 
nd the changes 
ow the spatial 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004 
evolution of a group of urban areas, with the objective of 
proposing à methodology for strategic monitoring of the 
environmental impact of urban development. 
It is relevant to highlight the fact that the Report of this project 
suggests that “urban growth and sprawl is a pertinent topic for 
analysis and assessment today. The environmental impacts of 
urban sprawl and the extent of urban problems have been 
erowing in complexity and relevance, generating strong 
imbalances between the city and its hinterland. The need to 
address this complexity in assessing and monitoring urban 
planning and management processes and practices is strongly 
felt” (EEA, 2002, p.7). 
The MOLAND methodology, a more advanced version than 
MURBANDY, has created the data bases for four periods, for 
the 1950s, the 1960s, the 1980s and the 1990s, through the 
interpretation of satellite images, principally IRS images, but in 
some cases IKONOS and SPOT images (Lavalle et. alt, 2002). 
It is important to note that the nomenclature adopted for the 
“remote sensing” was a more extensive version of the CORINE 
legend. One of the fundamental differences between 
Murbandy/Moland and Corine, apart from the greater precision 
with the level of detail, is that with Murbandy/Moland it is 
possible to make the distinction between different land uses. By 
contrast, Corine is more limited related to the distinction 
between different land classifications. This methodology has 
enabled the evaluation of “urban sprawl” for the 25 case 
studies, understood as the percentage increase in the urbanised 
surface area during the period under review. 
2.2 Data 
Participation in the Spot 5 Application and Validation 
Programme provided the researchers with access to a range of 
SPOT satellite images. In view of the wish on the part of the 
UPC/MMAMB to carry out a dynamic study, satellite images 
were provided dating from 1995 permitting their comparison 
with more recent 2003 images. 
Four images were used for the research referred to in this paper: 
e Spot 5 (2003) THR, black and white, 2.5 m resolution, 
Processing level 1A 
e Spot 5 (2003) THR CNA C1-3, natural colours, 2.5 m 
resolution, Processing level 1A 
a Spot 5 (2003) THX FC C1, false colours, 2.5 m 
resolution, Processing level 1A 
e Spot 3 (1995) Infrared, 20m resolution, Processing level 
lA 
The 60km x 60km SPOT Scene extends to a land area of some 
2 . . . . . re 
2,700 km”, taking into consideration the significant component 
of the sea. 
2.3 Method 
The three 2,5 metre resolution images were orthorectified 
simultaneously, with the aid of a 2.5 metre Orthophoto, in TIFF 
format, for the entire study area. This saved time and also 
contributed to the subsequent classification process. Owing to 
the magnitude of the image, it was divided into four roughly 
equal parts, thereby making the rectifying process more 
manageable, as illustrated by the 2003 True Colour image 
below. 
The supervised classification methodology drew upon the same 
three images (true colours, false colours, and black and white, 
all with a 2.5 metre resolution) in order to maximise the number 
of distinguishable land cover classes for the study area. At the 
time of writing (April 2004) this methodology has been applied 
to the south-eastern quadrant of the study area, as illustrated in 
Figure 1, which contains the vast majority of the compact urban 
development found within Barcelona’s metropolitan area. This 
methodology has made use of maximum likelihood, binary 
encoding and parallel piped methods, prior to scattergram 
processes and the merging of all the data, drawing upon a 
subtraction process to arrive at the final classified image. It 
made use of ER Mapper and ENVI software applications. 
  
Figure 1: SPOT scene for Barcelona 
© SPOT Image Copyright 2003, CNES 
The same broad supervised classification method (parallel piped 
and maximum likelihood) was applied to the entire area 
contained within the SPOT scene, comparing the 1995 and 
2003 images, for a more reduced range of land cover classes 
than those isolated under the previous exercise. 
3. RESULTS 
3.1 Indications of urban sprawl 
It is too early to offer conclusive results of this study, on the 
basis of the work yet to be carried out. Nevertheless the work to 
date has highlighted a number of points worthy of mention. 
As can be seen from Table 2, over the eight year period between 
1995 and 2003 in the wider metropolitan territory and beyond, 
there has been an increase of just under 50% in the broad 
category of “occupied land”. At the same time, there has been a 
commensurate decrease in “green areas” and “soil”, the two 
remaining categories of this albeit broad classification, 
indicative of a wholesale outward encroachment of 
development into the countryside. These results arise following 
the application of the same supervised classification 
methodology over the entire area of the 1995 and 2003 SPOT 
Images, as illustrated by Figures 2 and 3.