Database Reference
In-Depth Information
events that appear as a cascade, capturing the partial orderedness of the
event relationships. Another class of approaches have focussed on find-
ing spatio-temporal patterns in a database of moving object trajectories
(e.g., hurricane tracks and mobile users). In [75, 76], a spatio-temporal
association rule (STAR) mining technique was proposed to capture the
frequent appearance of moving objects at varying locations and time.
Figure 15.6. (a) Rise in the average winter temperatures of British Columbia,
Canada; (b) Increase in pine beetle infestation (shown as points in red); (c) Occur-
rence of forest fires at these locations. Relationship between pine beetle infestation
and forest fire events is unresolved [32, 66] and needs further study (Source: NASA
& Environment Canada).
The ability to automatically extract such complex relationships from
global-scale spatio-temporal data is essential to advance our current un-
derstanding of changes that can be attributed to natural and human-
induced forcings and in turn advance our knowledge about global ecosys-
tem dynamics at large. We next present an illustrative example of re-
lationship mining for detecting teleconnections in climate data such as
climate dipoles [41, 42], which are pairs of spatially distant locations
exhibiting a relationship in their climate anomalies.
 
Search WWH ::




Custom Search