Geography Reference
In-Depth Information
b
0.8
0.7
0.6
W1=[14E, 20E]
0.5
0.4
0.3
0.2
0.1
0
NA
11.3
16.8
22.2
27.6
33.0
Latitude
Fig. 17.9
(continued)
Change interval may also be observed along a path taken in a spatial field. For
example, on the spatial path shown in Fig.
17.3
, the value decreases rapidly from
high level (rain forest and savanna) to very low (desert), as shown in Fig.
17.9
b.
This helps researchers understand the spatial footprint of the transitional zone (a.k.a.
ecotone) which may shift in response to climate change. A formal definitions of
this pattern has been proposed as an instance of the interesting sub-path/interval
discovery problem (Zhou et al.
2011
) and a data mining approach to the problem
was provided.
Besides interval of change, terms describing this kind of change include gradual
change in boundary analysis in ecology (Fitzpatrick et al.
2010
) and in time series
analysis for climate science (Chapin III
2004
).
17.4.2.2
Spatial Boundaries
A number of applications aim to find a curve on two dimensional space that
separate different regions. Given a spatial point, polygon or raster dataset, the
output pattern consists of multiple curves where value difference on two sides of the
curve is significantly higher than what would be expected in a randomly generated
dataset. Figure
17.2
c shows the top 20 % change boundaries (highlighted) found
on the global GDP growth data (CIA
2013
). This problem is also referred to as
spatial wombling (Womble
1951
), where techniques have been proposed to discover
boundaries on point, polygon and raster data. Terms such as rapid change (Liang
et al.
2009
), abrupt change (Lu and Carlin
2005
) are also used in the definitions of
a similar pattern, whose footprint is also referred to as “Edge” or “Zones of Abrupt
Change” (Liang et al.
2009
), etc.
Search WWH ::
Custom Search