Geoscience Reference
In-Depth Information
Figure 3.1. From Berry's spatial information matrix to the exploration
of spatio-temporal data [BER 64]
The progress made in these two areas has been transferred to geography at
different times, with varied issues (1960 for the development of quantitative
methods and 1980 for that of geographical information systems). If spatial analysis
has been particularly influenced by the developments of these two areas, this is
mainly because they impact simultaneously on the methods that it uses and on the
information which it works upon. This impact is all the more important given the
increase in the availability of information, its diversity and the richness of its
granularity. A detour describing this landscape is useful to place our positioning.
With regards to the statistical methods, which are concerned here, there have
been numerous discussions between statisticians and geographers, highlighting
differences in points of view. These interactions, which are extremely rich ([CLI 73,
ANS 95, GRI 10]), have resulted today in a set of methods upon which spatial
analysis is based. According to O'Sullivan and Unwin [OSU 10], three types of
methods can be distinguished:
- analysis of geographical data , which is based on descriptive and exploratory
statistical methods to identify statistical relationships (in the sense of Berry's
“associations”) between various spatial phenomena [CHA 87, LEB 06, DUM 11];
- spatial statistics which have been developed to test the significance of the
random or organized characteristic of a spatial configuration [CRE 93, ANS 95,
GRI 91, FOT 00, RIP 05, DUB 14);
- spatial modeling which relies on models taking into account a spatial
dimension, such as Reilly's model to define catchment areas, or the gravity model to
predict flows between places [HAG 65, TAY 77, PUM 97].
Search WWH ::




Custom Search