International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
3. corridors are qualified by a decrease of PD and ED and an
increase of MA (figure 4). Corridors differentiate from
concentrated cities through MA absolute value which is
always over 2.88 acres for concentrated cities and under
2.32 acres for corridors. Corridors, like concentrated cities,
tend to consolidate existent structures.
|
|
|
^ i ce zm A 1 i. IM | ed a m cuf |
ED MA / | PD Fes MA je )| ED MA |
Vea 12 Area 13 i Aen 14
Figure 4 .Corridors: variation of indexes in absolute value
4. diffuse urbanisation is qualified by an increase of PD, ED
and MA (figure 5). That means that a tendency toward
diffusion is still prevailing.
E pe um
: — =
PD ED MA PD ED MA PD ED MA
Arca 15 Area [6 Area 17
Figure 5. Diffuse urbanization: variation of indexes in absolute
value
Built N* PD ED MA
aurfaces% | PATCH| 100 hectares | m / hectare | hectare
CONCENTRATED CITIES
Area2, 1990 508 268 1226 103,86 4,16
Areal, 2000 5929 165 755 7955 7,685
DIFFUSE CITY
Areas, 1990 1520 1982 19,46 7354 0,81
Areal, 2000 2684 1395 1359 8327 1.95
CORRIDORS
Area13, 1990 2253 1296 20,13 9089 4,41
Area13, 2000 3055 45 1342 9778 232
DIFFUSE URBANIZATION
Area 15, 1930 9,23 1391 4727 5224 0,53
Area 18, 2000 18,14 1572 1950 7440 0.22
Table 1. A sample for each settlement typolog
5. SECOND WORK
The second work analyses the same area with different
urbanizations, identified in various maps of the built space
derived from images at different resolution: (Landsat, Spot,
IKONOS).
Additionally, two maps, called IKONOS. 5 and IKONOS 10,
have been made by applying a buffer of 5m and 10m to the
IKONOS map, in order to verify whether such operation allows
to get the features of Landsat and Spot maps.
5.] Built surfaces
A first analysis compares built surfaces in the various maps
(figure 6).
It results that:
| Landsat has the higher percentage of built space (51%)
while it decreases a little in the Spot (43%) because of the
higher resolution.
N
lowest percentage occurs in IKONOS (26%), since built
space includes only buildings and roads instead of a
continuous urban space like in Landsat and Spot.
3. IKONOS_5 built surface is near to the one of Spot as well
as IKONOS_10 to the LANDSAT. Buffers indeed fill
voids between objects (roads and buildings) and increase
the area of objects.
BUILT SURFACES (%)
Landsat [Spot [IKONOS __ [IKONOS 5 [IKONOS 10|
Be | 438. | Bo [£8 1 B36
60
]
30
0 \
Landsat Spot IKONOS IKONOS 5 IKONOS 10
Figure 6. Built surfaces in the various maps
5.2 Mapping built space density
Maps of local density have been produced by processing built
space maps with a low pass filter (Pesaresi, 1993), whose
kernel corresponds in our case to | Km? of surface (that means
33 x 33 pixels for Landsat, 99x 99 for Spot and 999 x 999 for
IKONOS).
BUILT
SPACE
DENSITY
[JA 81-1009,
[1B 61-80%
Bc 41-60%
[.]D 21-40%
(JE 0-20%
Ss
LANDSAT
IKONOS
Figure 7.
5 intervals of local density named from A to E have been set.
Values refer to the percentage of built pixels over total
number: A = 100-81%; B = 80-61%, C = 60-41%; D = 40-
21%, E = 20-0%.
LANDSAT SPOT IKONOS IKONOS_5 IKONOS 10 |
A 15.35 10.86 7.90 9.82 14.44
B 18.08 17.23 7.91 18.11 18.39
C 26.30 21.48 20.51 22.97 26.35
D 30.15 29.20 21.79 24.81 27.19
E 10.12 21.23 35.89 24.29 13.68
Table 2 . Built space density classes over 5 images
Results reported in the above table show that:
I. Landsat gets the highest percentage of areas ranked in A
(A =15%), and IKONOS the lowest (A= 8%).
2. behaviour over intervals of density is similar in Landsat
and Spot with a maximum for D (D=30% for Landsat and
D=29% for Spot)
3. behaviour over intervals of density in IKONOS is quite
different from Landsat and Spot. IKONOS shows higher
values for class E (E= 36%).
4. with IKONOS. 5 and IKONOS 10 we obtain, for classes
A, B and C, percentage values similar to Spot and Landsat
respectively.
Internat
—
101
TOT
To pr
means
distrib
anothe
