A Coming of Age: Geospatial Analysis and Modelling in the Early Twenty
First Century Forty years ago when spatial analysis first emerged as a
distinct theme within geography's quantitative revolution, the focus was
largely on consistent methods for measuring spatial correlation. The
concept of spatial au- correlation took pride of place, mirroring
concerns in time-series analysis about similar kinds of dependence known
to distort the standard probability theory used to derive appropriate
statistics. Early applications of spatial correlation tended to reflect
geographical patterns expressed as points. The perspective taken on such
analytical thinking was founded on induction, the search for pattern in
data with a view to suggesting appropriate hypotheses which could
subsequently be tested. In parallel but using very different techniques
came the development of a more deductive style of analysis based on
modelling and thence simulation. Here the focus was on translating prior
theory into forms for generating testable predictions whose outcomes
could be compared with observations about some system or phenomenon of
interest. In the intervening years, spatial analysis has broadened to
embrace both inductive and deductive approaches, often combining both in
different mixes for the variety of problems to which it is now applied.
