Introduction - Computational Movement Analysis - page 2

1.1 Motivation
Similar to many other research fields, the spatial sciences have made a transition
in the last decades from computation and data poor to computation and data rich
environments (Miller and Han 2009 ). The revolutionary character of this transi-
tion is especially evident in the application field animal navigation research, where
the movement behavior of individual animals has been studied for decades as a
fundamental knowledge base. Not so long ago, tracking of moving entities was a
very cumbersome and costly undertaking: For example, turtles were tracked using
thread-trailing (Claussen et al. 1997 ) and all that flying birds would reveal was their
vanishing bearing in release experiments (Walker 1998 ). By contrast, to date it is
now common to track flying birds using GPS devices with a sampling rate of seconds
(Shamoun-Baranes et al. 2012 ). Hence, the previous lack of fine-grained movement
data is a first reason why CMA is a relatively young and little-developed research
field.
Secondly, within GIScience, the legacy of cartography's static view of the world
slowed down the development of CMA. For decades, dynamic processes and change
was primarily captured in a snap-shot manner, where the arrival of a new areal or
satellite image only allowed a comparison with previous states of the world (Worboys
and Duckham 2004 ). Finally, geography's snap-shot view unfortunately found a
match in the concept of sporadic updates in the databases underlying early geographic
information systems. Conventional databases are designed for handling mostly static
data with occasional updates (e.g., a change of ownership or an area change in
a cadastral map), but not entities that change continuously (e.g., the permanently
changing location of a moving car).
The increased availability of movement data goes hand in hand with a growing
interest of the application fields in exploiting that new resource. Be it behavioral
ecology, transportation and mobility research, surveillance and security, or even
sports analysis, all fields interested in movement show significant interest in studying
and analyzing movement. For example, only for the field of movement ecology,
Holyoak et al. ( 2008 ) list thousands of papers addressing organismal movement,
ranging from seed dispersal to bird migration. The question arises if the described
data and information revolution requires new scientific foundations with respect to
methods. With richer and more complex data comes a need for more sophisticated
tools for managing, exploring and analyzing that data. Arguing for the need of new
scientific fundamentals CMA first of all means identifying the characteristics of the
new challenge. Similar to Anselin ( 1990 ) seminal question about “What is special
about spatial?”, it is now fair to ask “what is special about movement?”.
The following list summarizes a set of properties of movement (data) that under-
line the need for a new theory of computational movement analysis. The list is inspired
by revisiting seminal texts framing the theory of GIScience (Anselin 1990 ; Goodchild
1992 , 2001 ), and adapting the challenges therein to the movement domain.