Geography Reference
In-Depth Information
8
Exploring spatial
patterning in data values
Introduction
8.1
h is chapter introduces a variety of methods for the analysis of spatial variation in
single and multiple variables. Methods are introduced that allow for the exploration in
changes in values from place to place or in the way in which variables are related.
In the i rst case, an example problem might be to ascertain if zones tend to be more
similar to their neighbours in some parts of the study area than in others (e.g. do
neighbourhoods in some areas have similar characteristics while neighbourhoods in
other areas are quite dif erent). In the second case, we may want to address questions
such as 'How does the relationship between altitude and precipitation vary from place
to place?' (e.g. does altitude seem to have an ef ect on precipitation amount in some
areas but not others). h e initial focus of the chapter is on the analysis of spatial struc-
ture (spatial autocorrelation, i.e. the degree to which values at one location are similar
to values at neighbouring locations). Next, the chapter moves on to computation
of local statistics. h e initial concern is with univariate measures; next, regression
and correlation procedures are outlined that enable exploration of spatial relations
between multiple variables. Finally, some other approaches are mentioned briel y
before the chapter is summarized.
Spatial autocorrelation
8.2
Section 4.8 introduced the concepts of spatial autocorrelation and spatial dependence.
Recall that spatial autocorrelation refers to the nature of correlation between neigh-
bouring values while spatial dependence suggests the case where neighbouring values
Search WWH ::




Custom Search