titative
he first
y such
ses the
s allow
ictures,
interval
oes not
| to the
istency
cription
can be
ologies
derived
ılitative
ılitative
ange of
estions,
plied to
images.
ion and
jn. The
a set of
ions of
at have
m over
a range
yologies
whether
d, thus
yses the
vels of
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
A third work computes the landscape indexes from various
analysis techniques over five sub-areas, the same identified in
satellite images at different resolution (Landsat, Spot and
IKONOS).
2. INPUT IMAGES
The three works are based on maps of the built space obtained
through satellite image processing (Pesaresi and Bianchin,
2001). The map from Spot has been slightly corrected with
photo interpretation. Images have been co-registered in order to
allow comparison.
The following images have been used:
e Landsat 5 TM, 30m, frame 192/28, 08/20/1990
e Landsat 7 ETM+, 30m, frame 192/28, 09/08/2000
e Spot 4 Pan, 10m, frame 062/258, 03/30/2002
« IKONOS Pansharpened, 1m, 07/02/2001.
3. LANDSCAPE ECOLOGY STATISTICS
Spatial statistics of the landscape are quantitative indexes based
on geometric features of a homogeneous region called patches.
To compute them we used the sw FRAGSTATS of McGarical
and al. (2002).
The following landscape indexes have been considered:
| patch density (PD) is the ratio between number of patches and
total area. Low values of PD imply the presence of few regions,
while increases of PD mean more patches in the area.
2 edge density (ED) is the ratio between perimeter of all regions
in the area and total area. Low values can be associated with
landscapes composed of few, wide regions; high values mean
composite landscapes with several regions.
3 mean patch area (MA).
4. FIRST WORK
In the area of Veneto region (figure 1), urban studies identify
four settlement typologies: concentrated cities, diffuse city,
diffuse urbanization, corridors (Indovina et al., 1990).
Figure 1. 17 sub-areas belonging to different settlement
typologies
17 sub-areas belonging to four typologies have been drawn
(Fregolent, 2004 ).
467
e For the concentrated city: areas |, Venice, 2 Treviso,
3 Padova, 4 Vicenza.
e For the diffuse city: areas 5 Roman Centuriation,
6 Terraglio, 7 Riviera del Brenta, 8 Padova-Vicenza
axis, |.9 Noale-Scorze, 10 Treviso-Montello ^ axis,
11. Vicenza-Cittadella axis.
e For the corridors: areas 12 Treviso-Ponte della Priula
axis, 13. Cittadella-Bassano . axis, 14. Cittadella-
Castelfranco axis.
e For the diffuse urbanisation: areas 15_Bassano Montello,
16_Bacchiglione, 17_Piazzola del Brenta.
For the above sub-areas, spatial indexes have been computed
at two dates, 1990 and 2000, then compared. The comparison
shows that:
1. for a given image at date / (either for /=1990 or =2000)
values of indexes for the different typologies are not so
different as it could be expected (for example: PD is 14
for area 5, 13 for area 13, ED is 80 for area 2, 83 for area
5 and 88 for area 13, at 2000).
2. it results that the variation of indexes at two dates
characterises the four typologies represented in the various
sub-areas. The 17 sub-areas develop from 1990 to 2000
according to the typical behaviour of the typology to
which they belong independently of their localization.
This defines certain territorial uniformity.
In detail:
1. concentrated cities are qualified by a decrease of PD and
ED and an increase of MA, which means that they become
more compact (figure 2). New built spaces occur in the
voids of the core or increase existent patches.
D Areal ruere ER Nes ur o Mead
Figure 2. Concentrated cities: variation of indexes in absolute
value
2. diffuse city is qualified by a decrease of PD but an
increase of ED and MA (figure 3). New built spaces
develop contiguously to the existent ones (PD decrease)
creating ramifications (ED increase). Such development
leads more toward fragmentation than diffusion.
FD ED MA
Area 8
|
| |
| oh [Ee
|
T Po. ED MA PD ED MA
psc Area { Area 16 | Mead) es
Figure 3. Diffuse cities: variation of indexes in absolute value
