Geography Reference
In-Depth Information
Chapter 17
Spatiotemporal Change Pattern Mining:
A Multi-disciplinary Perspective
Xun Zhou, Shashi Shekhar, and Pradeep Mohan
17.1
Introduction
Given a definition of change and a dataset about a spatiotemporal (ST) phenomenon,
ST change pattern mining is the process of identifying the location and/or time
frame of shifts in phenomenon. Detecting patterns of change over space and/or
time is an increasingly important activity in application domains ranging from
climate science to public health. Researchers have developed numerous techniques
to facilitate the mining of such patterns. Addressing domain specific challenges, they
have often worked in distinct research settings, most notably time series analysis,
remote sensing, and spatial statistics. Although they tend to target different aspects
of the change pattern mining problem, there is much researchers could learn from
one another to advance their respective areas.
Related literature includes discipline-specific surveys. For example, literature on
time series analysis (Basseville and Nikiforov
1993
; Shaban
1980
; Zacks
1983
)
focused on methods for the change point detection (a.k.a., “abrupt change” detec-
tion) problem. In remote sensing, some survey papers (Radke et al.
2005
; Coppin
et al.
2002
; Singh
1989
; Im and Jensen
2005
) mainly focus on change detection
techniques using bitemporal or multi-temporal image sequences. A tutorial by Wong
and Neill (
2009
) discusses the problem of “event detection” which aims to discover
abnormal behaviors of data. The authors cover time series change detection and
Search WWH ::
Custom Search