Geoscience Reference
In-Depth Information
3.1. Statistical data and spatio-temporal analysis
The practices in spatial analysis have been strongly influenced by the different
methodological developments in statistics and in computer science. On the basis of
these disciplinary fields, geographers have devoted their time to developing
formalisms adapted to their questions, with a particular attention to the taking into
account of geographical space. We sketch out here a few of those stages and then go
back to the layout of the data.
3.1.1.
Spatial analysis evolution in a context influenced by statistics and computer
science
As early as 1960, in the context of “the quantitative revolution”, Brian Berry
[BER 64] proposed in a seminal article a system of generic representation that
allows the integrating of the variety of empirical approaches of regional science.
Starting from the “geographical fact” that he describes as being either the
observation of the same phenomenon in several places (topical approach), or several
observations in the same place (regional approach), he proposes the concept of a
“geographical matrix” (which is also known as a “spatial information matrix”). This
concept allows these two points of view to be integrated. The rows of this matrix
designate different measurable characteristics and the columns designate the places.
At the intersection of a row and a column, the observed value is found for a certain
characteristic in a given place. The geographical matrix is a representation system
for the analysis of spatial variation (variability over a row) and that of spatial
associations (spatial variability over a column). For Berry, this representation is
compatible with and supports a systemic point of view: all the described locations
are interact and make up a spatial system. The geographical matrix allows the
evolution of such a system to be conceptualizes by taking a third dimension into
account, the temporal dimension. A move is then made from variations and spatial
associations to variations and temporal associations (Figure 3.1). The same
components are found in this formalism as in the one developed by Peuquet
[PEU 84] presented in the previous chapters.
Berry's proposals may seem very “modern” nowadays but they were inserted
into a context where automatic calculation was only just beginning to emerge.
Quantitative geography (and spatial analysis) has benefited from the strong progress
that the two disciplinary fields on which it relies have experienced:
- the statistics for all the methodological aspects and data processing;
- computer science for all aspects related to geographic information and data
modeling.
