structures, projections, or levels of resolution is
especially useful in land use and land cover studies.
Quantifying temporal change often involves
using such sources as historical maps, air photos,
and satellite images. Changes in the spatial distribution
of land classes can be summarized by
overlaying maps of different dates and analyzing
their spatial coincidence. Changes from one land
class to another can be mathematically described
as probabilities that a given pixel will remain in
the same state or be converted to another state
(Johnson, 1993). The use of GIS with digitized
maps provides more precision in determining TPs
over different portions of the landscape at different
times, removing many of the difficulties in
early works of parameterizing Markov matrices
(Johnson et al., 1996).
The integration of satellite remote sensing, GIS,
and Markov modelling provides a means of moving
the emphasis of land use and land cover change
studies from patterns to processes. Data and
computational limits are becoming less significant
due to advances in remotes sensing for change
detection and in the incorporation of remotely
sensed data and auxiliary data into GIS (Baker,
1989). The most compelling research issues may be
a lack of appreciation of the power of the integration
and understanding of how to incorporate existing
knowledge in useful