Geography Reference
In-Depth Information
17.4.3
Spatial Regions of Change
Spatial regions of change refer to areas in the spatial framework that have signifi-
cantly different data in comparison to the rest of the data or the expected value in
the same area. Specifically, there are two patterns in this category, namely, spatial
clusters, and changes between images.
17.4.3.1
Spatial Clusters
Spatial clusters refers to the gathering pattern of point data in space. In real
world datasets, such as disease reports and crime reports, data points are not
uniformly distributed over space (a.k.a. heterogeneity). The area with dense reports
are considered as having a higher risk of disease or crime. Figure
17.2
bshows
the footprint (highlighted polygons) of spatial cluster in the given point (crime)
dataset. In some scenarios, the expected numbers of reports is available from
historical records or theoretical models. For example, the number of disease reports
is usually assumed to follow a Poisson distribution (Banerjee et al.
2003
). Areas
with a statistically significant higher number of reports are also considered as spatial
clusters, which may indicate an ongoing outbreak of disease or crimes (Kulldorff
and Nagarwalla
1995
). On areal-aggregated or raster datasets, a cluster may also
refer to a number of polygonal areas or a collection of spatial grids where the
intensity (e.g., the number of reports) inside is much higher than outside. Such
clusters are also referred to as “spatial hot-spots” in spatial statistics (Chainey et al.
2008
).
A number of techniques have been proposed to detect spatial clusters and
hotspots. Among them, scan statistics (Kulldorff
1997
) is a representative method
that has been widely used in epidemiology. We discuss scan statistics in more detail
in the next section.
17.4.3.2
Change Between Images
In remote sensing, it is of interest to identify spatial regions of change between two
or among a series of raster images of the same spatial area. This problem is referred
to as “change detection” in remote sensing and image processing (Coppin et al.
2002
; Radke et al.
2005
). Change detection on satellite images help identify new
buildings or suspicious object movements in a region. For example, in the example
in Fig.
17.4
b, the area in the red box is one of the identified change regions (near
Washington Dulles International Airport) where land use has significantly changed
during the years between the two snapshots. The change region can be identified as
whole, or on a pixel-by- pixel basis.
Search WWH ::
Custom Search