This thesis is an interdisciplinary work between Geographical Information Science and Wildlife Biology. It is focusing on methodological aspects that are arising when both disciplines meet. The aim of this work is the development of new analysis methods for movements of animals in the environmental space.
I would like to start off with a basic idea: Animals, in contrast to plants, once upon a time have invented locomotion. The definition of locomotion includes two fundamental concepts: Space and Time. Spatial data, the material Geography is working with, is a very complex matter. In the past few decades powerful instruments have been developed for handling and analyzing such data. The diffusion of these geographical information systems (GIS) into disciplines outside Geography started in the late 80ies and early 90ies. Today we are at the beginning of a new era for GIS where they become almost omnipresent and even available in standard office computer software packages.
Wildlife researchers are always confronted with spatial data analysis, so it is not astonishing that they also started to make use of these systems. Impressive applications were developed and results presented, but a major problem still exists in the handling of wildlife data in GIS. Although powerful in analyzing the spatial domain, GIS are nearly ignorant concerning the handling and analysis of the temporal domain. Thus, major efforts today on this topic are directed towards the representation of time within GIS.
The work presented here aims to go two steps ahead of representational issues. The first step is to extend the understanding of time as a single dimension with a perception of time as a congregation of 'multi-faceted' temporal aspects. The second and main step that goes beyond representation is the development of powerful analysis methods that incorporate both spatial and temporal aspects in an equally balanced way. Hence, the main question posed in this work is the following:
How can we include temporal aspects in the analysis of the spatial behavior of an animal within a GIS environment?
The intentions underlying the development of new concepts may be better explained by presenting the current state and its shortcomings. I shall try to do this in the form of the following list of common misconceptions and shall then explain why they need to be overcome.
The following shows some of the misconceptions and errors in reasoning in the above statements.
(1) Random is often confused with the finding that underlying patterns cannot be identified. Animals obviously have clear patterns in their behaviors, but sometimes these patterns are too complex to be identified with simple methods.
(2) The temporal dimension is left out. Without time no changes are possible. This means that two things are vital when analyzing movements: space and time.
(3) Summarizing information either means concentrating or loosing information. The art of analysis is to make complex phenomena easy to understand.
(4) Most of today's data in wildlife biology (and probably many other fields, too) are not analyzed to its full potential due to a lack of time. The ease and speed with which an analysis can be performed is crucial for its application in the everyday life of a researcher. If one needs to calculate the time to sunrise for 9000 observations by hand, no one would do it. If it can be done by pushing a button, it is performed whenever there is a chance to gain new insights into the data.
(5) The adequate representation of temporal aspects in GIS is only one step towards a Temporal GIS. The representation of a line and a polygon in a computer software does not make it a GIS until analysis functionality is added to the system. They are mostly called graphics, drawing or cartographic programs instead. The same is true for time aspects. The representation itself is only the first requirement. To make full use of the information, the analysis part of a spatio-temporal information system needs to be developed, a part that is largely missing today.
(6) The most popular way of analyzing temporal data in wildlife research is to plot the data on a separate map for each observation period. Calendar months presumably have no meaning to animals. By applying such a frame-set to the data, the researcher implicitly (and probably unintentionally) assumes that the changes he or she is looking for overlap with these time periods. If the changes in the data are shorter than the calendar months, they will not be detected. In the case the changes are occurring at a larger time range or in between them, there is a potential danger that fluctuations in the data are considered to be significant differences.
Here, a completely new approach was developed. A conceptual shift is presented from a space-centered view to a time-centered view, where time can even become the dominant dimension in the analysis (chapter 5).
With this thesis I would like to pursue the following objectives. First I would like to provide a better understanding of time and temporal aspects. It should lead to improved and more accurate analysis methods for wildlife research data within geographical information systems. In the context of GIS it will show that temporal geographical information systems not only need a sophisticated data model to represent temporal aspects, but in addition that new and improved analytical methods need to be developed to adequately analyze the data. This work is intended to provide new analytical methods to achieve better insights into changes in the spatial behavior of animals. It will also create analytical instruments to perceive and define biologically 'meaningful' phases in a wildlife researchers data. Even though most examples and statements are made primarily with a focus towards wildlife research, the methodology is of course applicable in any context concerned about spatial movements of point objects.