total
11632
1565
186
429
78
79
47
136
86919
2047
741
13704
6885
4771
19184
236
197
163
1397]
1653
414
1136
3600
|
] Cover
ference,
change
iropean
chnical
dology
project,
"ences,
| guide
ropean
Photo-
red for
EVALUATION OF THE DYNAMIC OF URBAN MORPHOLOGICAL ZONES
USING REMOTELY SENSED DATA
M. Caetano and A. Navarro
Centro Nacional de Informagáo Geográfica (CNIG)
Edifício Inovagáo III-Sala 614, Tagus Park, 2780 Oeiras, Portugal
Tel. 351-1-4219800, Fax: 351-1-4219856; e-mail: marioGcnig.pt
Commission VII, Working Group 3
KEY WORDS: Landuse maps, Urban areas, Urban morphological zones, Land cover change,
Satellite data, Ancillary data, Contextual algorithms, Expert knowledge
ABSTRACT
A methodology for mapping Urban Morphological Zones (UMZs) based on Earth Observation
(EO) data is presented and tested for two dates (1984 and 1995) in the Área da Grande
Lisboa (1046 Km?), Portugal.
EO data (SPOT and Landsat TM) and ancillary data were used to generate landuse maps for
1984 and 1995 with the CLUSTERS nomenclature. The methodology for landuse mapping
includes spectral classification at the pixel level and the use of contextual algorithms
to identify landuse classes that cannot be identified at that level. Ancillary data were
used to improve the landuse maps generated just by using satellite data. After the
generation of the landuse map, we developed and applied an algorithm to identify the
UMZs. An UMZ is an area that includes all the man made areas that are closer than 200 m
to each other (EUROSTAT definition).
In the final landuse maps, 25 classes were identified with an overall accuracy of 80%.
Results also show that the developed methodology allows an accurate definition of UMZs.
An analysis of the 1984 and 1995 UMZs maps show an expansion of the urban area of AGL,
including new UMZs and the growth of some 1984 UMZs.
by using Earth Observation (EO) data (SPOT
1. INTRODUCTION and Landsat).
Updated maps of built-up areas are The methodology used for UMZs delimitation
important for landuse planning and decision entails the following steps: (1) generation
making in regional areas. An important of landuse maps; (2) identification of the
concept in urban planning is the Urban built-up areas and (3) delimitation of
Morphological Zone (UMZ), which reveals the contiguous built-up areas. The most complex
physical boundaries of urban areas. and difficult step is the generation of the
According to EUROSTAT, an UMZ is an area landuse maps based on EO data. For this
that includes all the man made areas that reason, most of our effort was allocated to
are closer than 200 m to each other. the production of these maps. Note that the
identification of built-up areas can be
A very important information for landuse achieved by a simple reclassification of
planning is the characterisation of the the landuse map and the identification of
growth of urban areas within a given contiguous urban areas can be easily
regional. area, i.e., a .study of the .UMZs achieved by the application of adequate
dynamic. The study reported here is about algorithms within a Geographic Information
the evaluation of the dynamic of the UMZs System (GIS) environment.
in Área da Grande Lisboa (AGL), Portugal.
The characterisation of the UMZs dynamic is For a long time that images acquired by
based on the comparison of UMZs maps for satellites have been used for landuse
two dates (1984 and 1995) that were mapping, e.g. Ormsby (1992), He et al.
generated, independently from one another, (1994), due to the low cost, periodic
acquisition and coverage of large areas.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 691