Image Processing Reference
In-Depth Information
a
and
()
g
a
i
t
Table 2.3
Temporal mean value of
()
l
t
for populati
on
Smoothed
()
l
a
t
Smoothed
()
g
a
i
t
Whole study area
(%)
Urban centre
(%)
Outer urban
ring (%)
Hinter-land
(%)
Cities
Years
Milan
1971-2001
−0.1
−1.2
0.6
0.9
Brussels
1981-2001
0.2
−0.4
0.3
0.2
Stuttgart
1976-2000
0.5
−0.5
−0.1
0.4
Bristol
1971-1991
0.1
−0.8
0.8
0.4
Helsinki
1990-1999
1.2
−0.5
0.5
−0.4
Rennes
1962-1999
1.5
−0.7
1.8
−0.2
a
Table 2.4
Temporal mean value of
()
l
t
and
()
g
a
i
t
for employme
nt
g
a
i
()
t
Smoothed
()
l
a
t
Smoothed
Whole study
area (%)
Urban centre
(%)
Outer urban
ring (%)
Hinterland
(%)
Cities
Years
Milan
1961-2001
0.7
−1.0
1.3
1.0
Brussels
1984-1999
1.2
−0.9
1.7
0.6
Stuttgart
1976-1999
0.4
−0.7
0.4
0.3
Bristol
1971-1991
0.4
−1.1
1.2
0.6
Helsinki
1990-1999
0.3
−1.1
1.5
−0.6
Rennes
1982-1999
1.3
−0.7
1.6
−0.6
null hypothesis that the spatial autocorrelation of a variable is zero. If the null
hypothesis is rejected, the variable is said to be spatially autocorrelated (see Anselin
1995
; Getis and Ord
1996
for a theoretical and formal description of the indicators).
As an example, when applied to population density, local indices of spatial autocor-
relation might be used to define urban centers (high autocorrelation of density
between adjacent units - similar high densities), the rural hinterland (high autocor-
relation - similar low densities), urban poles (low autocorrelation - urban poles
surrounded by rural zones, with much lower densities), and finally intermediate
zones characterized by very low spatial autocorrelation, corresponding to suburban
areas, which are a mix of more or less recently urbanized communes and other still
rural communes. In Fig.
2.5
we provide a map of the local indicator of spatial auto-
correlation for the population densities in the SCATTER case study cities.
2.4
Conclusions
This chapter has provided an overview of some of the issues that are salient to the
measurement of urban form and function. In many respects, urban remote sensing
provides an important spur to improving our understanding of the way that urban
areas grow and change. Certainly there is a sense in which our abilities to routinely
monitor incremental accretions and changes to urban shapes are not matched by socio-
economic data of similar spatial or temporal granularity. Although increasingly
