Geography Reference
In-Depth Information
techniques. This is essentially the approach used
for the analysis of deforestation in Madagascar
described above. However, the analysis of
satellite data to determine change can be more
complex. Johnson and Kasischke (1998) have
recently given an overview of the problem.
They suggest that change detection using
remotely sensed images can be approached in
two general ways.
The first approach identified by Johnson and
Kasischke depends on making a spectral
classification of the images either separately for
different time periods or by combining the multi-
data imagery to identify areas of change. In the
future, it is likely that much more sophisticated
classificatory approaches will be developed with
the use of classification routines that are based on
the idea of spatial objects rather than pixels
(Hinton 1998). It is likely, for example, that such
methods will be used for the census of land cover
using remotely sensed data in Countryside Survey
2000.
The second general approach to change
detection identified by Johnson and Kasischke
(1998) depends on the analysis of radiometric
differences between images. Differences between
bands for different dates can, for example, be
compared once images have been corrected
radiometrically to ensure that direct comparison
between them can be made. Alternatively, band
ratioing, principal component analysis or methods
based on the calculation of indices based on
particular band combinations can be used (see
discussion of NDVI below). With these methods,
some form of radiometric correction is required
to take account of different atmospheric or
illumination conditions at the times the images
were acquired, or any differences in sensor
characteristics. Examples of the application of
these methods have recently been provided by
Prakash and Gupta (1998), who looked at the
problem of mapping land-use change in a coal-
mining area of Jharia, India; Sunar (1998), who
has considered the problem of mapping land cover
change in Istanbul, Turkey; and, Mino
et al
. (1998),
who have used multi-date image data to monitor
grassland improvement in Japan.
Detecting changes in dynamics
While some problems can be reduced to the
analysis of change in relation to some baseline
state, other environmental problems are often
more complex. Particular difficulties arise, for
example, in the context of environmental systems
that are
dynamic
in character. In these situations, it
is not a simple change of state that we need to
consider but a change in the
behaviour
of the
system over time (Figure 40.3).
The importance of understanding
environmental change in terms of change in
system dynamics can be illustrated by recent work
on El Niño, which is a disruption in the ocean-
atmosphere system in the tropical Pacific. The El
Niño effect can be seen, for example, by the
analysis of sea-surface temperatures recorded from
an array of buoys distributed across the Pacific on
each side of the equator (Figure 40.5). These point
measurements can be used to interpolate surface
temperature patterns using a GIS. Under normal
conditions, the western side of the Pacific is
warmer than the east. However, with the El Niño
oscillation, a warm tongue of water extends
eastwards to the coast of Equador and Peru,
disrupting the upwelling of the cold waters that
enrich the Pacific coast of South America. There
are many web sites that discuss El Niño, with
excellent graphics and animations that illustrate
how the dynamics of the atmosphere-ocean
system change (see additional sources below). The
importance of understanding and detecting such
changes lies in the widespread economic and
social consequences that arise as a result of the
fluctuation in this environmental system.
On land, the problems posed by the need to
monitor the behaviour of an environmental
system can be illustrated by work that uses the
Normalised DifferenceVegetation Index (NDVI).
In Figure 40.2, we have already seen how NDVI
can be used to analyse the dynamics of land cover
using the 'change vector analysis' approach
described by Lambin and Strahler (1994). Analyses
based on NDVI are now routinely used
throughout the world to help to resolve problems
related to fluctuations in terrestrial productivity.
