0755-B1

Qualitative Geomatics for Sustainable Development

Geoffrey Edwards[1]


Abstract

The use of geomatics in support of sustainable development has been restricted, to date, to mapping and modelling the natural resources available and, to a lesser extent, to mapping the demographics of the population that uses or exploits those resources. The use of geomatics to further the study of the social dimensions of the problems related to sustainable development is limited partly because of the large methodological divide between the natural and social sciences. In particular, the latter draw much more heavily on qualitative analysis and descriptive study. Current geomatics technologies do not support such an approach. However, a new generation of geomatics tools is under development that addresses such needs more directly. These new tools range from methods for representing qualitative human reasoning processes on a computer, on the one hand, and computer architectures for emulating human cognitive responses to environments on the other. The new tools should lead to environments involving geographic data that better support social analysis, and, when combined with existing geospatial tools, may provide a set of tools designed to address the full range of issues related to sustainable development.


1) Introduction

Geomatics technologies are presently limited in their application scope. They are used when precise information is available or is needed for visual purposes. In general, today's geomatics tools are poor at handling less precise information. In a sense, they assume too much precision. More than a decade of research has been devoted to the development of geomatics methods and technologies that are more expressive for spatial information that is qualitative in nature. As a result, a number of lines of inquiry are yielding new methods that should support more effectively qualitative spatial descriptions of the world, and qualitative analysis. In this paper, some of these lines of inquiry are presented along with the new tools that are becoming available.

2) New tools for qualitative spatial representation and analysis

2.1 Qualitative spatial reasoning (QSR)

Qualitative spatial reasoning is an area of intense interest and research in computer science over the past several decades. The domain is concerned with conceptual frameworks for spatial reasoning, with formal methods of reasoning, and with logics and calculi. Also, qualitative spatial reasoning is strongly linked to the efforts to formalise ontologies (Bittner and Edwards 2000). Likewise, qualitative spatial reasoning overlaps with work in computational linguistics aimed at supporting language-based spatial reasoning (Edwards et al. 1996).

From a practical point of view, qualitative reasoning may be viewed as a way of formalising expressions of existence of cases, situations, etc. when there is not enough support to make inferences about the properties of groups of cases, such as would be required by statistical techniques. Hence, for example, if an interview reveals the existence of a certain activity, then this fact may be used to support inferences based on the existence of the activity, even though there is no population from which one could infer other information. Likewise, qualitative reasoning may be applied to ordered symbols when the order is not based on numerical information.

2.2 Cognitive representations

Human beings maintain a large number of mental representations of the world. Many if not most of these representations incorporate largely qualitative information. A recent development of some interest is the emergence of GIS capabilities that accommodate human conceptualizations of space. Sometimes called Perceptual GIS, these are standard GIS that incorporate support for defining nested spaces that correspond to different scales of perception (e.g. Reginster and Edwards 2001).

Several scales may be defined - figural spaces, body spaces, vista spaces, environmental spaces and geographic spaces (Montello 1993). Figural spaces are viewed from a single head position. Body spaces are defined by the regions that can be touched. Vista spaces are defined by the region that can be sensed by a person from a single location. Environmental spaces are defined as the set of vista spaces accessible via locomotion. Environmental spaces are a function of the means of locomotion and the time scale over which the locomotion occurs. Finally, geographic spaces are those that are too large to be viewed by moving through them. Existing GIS is largely concerned with geographic spaces. The other types of space can be measured and modeled using a variety of techniques with geomatics.

Perceptual spaces are templates for assessing other variables, such a variables of environmental quality, as a function of the perceptions of different types of agents. That is, they determine the parts of space that impinge on the perceptions of people. They are a kind of negative space. On the other hand, lifelines are a related tool, but they are positive spaces. Lifelines are space-time trajectories of individuals or groups of individuals. Understanding the interactions between lifelines, and between these and regions of space, is increasingly being used to study social constructs within geography. Lifelines are derived from work on time geography (Hagerstrand 1970).

Both lifelines and perceptual regions require information concerning the displacements of people, which is very demanding in terms of data acquisition, often involving extensive interviews, protracted self-monitoring, or the use of modern spatial location instruments such as digital loggers equipped with portable GIS. Furthermore, displacement trajectories are complex and analysis is non trivial. In the original study devoted to the determination of perceptual regions, the use of rules to approximate perceptual regions was investigated (Reginster and Edwards 2001) to alleviate this problem. Varying rules for different target groups may be constructed to generate different forms of perceptual regions.

Another qualitative means of characterising environmental space from a cognitive point of view is the use of image schema. Image schema may be understood as structural primitives or spatial configurations that characterise spaces (Johnson 1987). Examples include the container schema, the conduit schema, and the surface schema. Hence a given region may be viewed as a container (into which objects may be placed), a surface (onto which objects may be put), or a conduit (through which objects may be passed). Image schema have been used as a visual language for characterising navigable spaces in ways that are meaningful to users. Furthermore, studies of the formal properties of combining cumulatively local information about space into global representations shows that people's survey representations of even unstructured spaces partition the space into "rooms" and "gateways", albeit at different hierarchical levels (Edwards and Ligozat 2002). Using formal tools, it is possible to transform perceptual regions into mental representations of space that follow these structures and hence to construct models of human mental representations of navigable space.

3) Applying qualitative methods

We now return to our initial problem. How can these new tools be used to address application areas not well handled by existing geomatics tools, such as GIS, remote sensing, and GPS, that assume high precision?

3.1 Navigation and access

Navigation problems, although often viewed as being highly quantitative in that precise information about position is usually required, incorporate significant qualitative reasoning elements. Humans solve navigation problems constantly with very little recourse to quantitative methods. Hence navigation is one of the most important applications for the new, qualitative methods under development.

Applications in navigation that involve qualitative methods are essentially focussed on the problem of assisting people in making decisions such as selecting routes, describing them or communicating information about them to others. Selecting routes is quite a complex task and in most real contexts leads to computational complexities that are not easy to solve. People revise their destinations on an ongoing basis, and hence navigation problems often involve a fixed origin but an open destination. Strategies for route selection that incorporate such changeable behaviour are required. The best route also depends on the means of transport, and the effort involved in the actual displacement. These are choices about which issues of social equity and sustainability may indeed arise, since constraints on route selection are likely to be more severe for people with disabilities, for people with limited access to costly transport means, or for people who are encumbered. Also, families must often plan routes as a function of the needs of several individuals, and this further complicates the process of route selection. Sustainable transport is an area of major concern where a good understanding of human travel and navigation behaviour is essential.

3.2 Planning and design

Planning and design are more obviously appropriate tasks for addressing issues of social equity and sustainability. Planning usually occurs at several different levels and scales. Qualitative approaches might conceivably model part of the planning process or support this in a direct way. Planning involves sophisticated and complex reasoning, judgements and decision-making, including the determination of objectives, an examination of the materials and means available for achieving these objectives, analysis of the current situation and likely obstacles and a set of scenarios for getting from the current situation to the desired goals. Planning usually involves more than one person, hence a negotiation process may also be involved that calls for questioning elements already developed and sometimes a reframing process for lateralizing problem parameters (Goel and Grafman 2000). While there is little interest in reproducing the entire process on a computer, the use of formal reasoning methods to examine the logical entailments of certain assumptions about the current situation, the objectives, the material and means and the obstacles may be explored using QSR.

Similar considerations arise in the process of design. Design, in general, addresses multifaceted problems with many different, interacting levels of issues. To channel design too early into fixed objectives, as is done in the planning process, is to encumber the process of finding original solutions that meet a wide range of needs. As indicated earlier, one of the results of the design process may be to determine appropriate sets of objectives for meeting these needs.

Several formal methods of design exist. One method that has gained widespread use is a method for urban design developed by Christopher Alexander in the 1960's. This method uses a pattern language to break down and describe the interlocking problems of urban design and to provide solutions that likewise interlock with each other, over several different scales (Alexander et al. 1977). This design pattern methodology has been adopted by the software design community, and is now finding use in an increasingly broad range of disciplines.

The specific needs of segments of the population that are disenfranchised, or the need to generate solutions that are more sustainable in general, can be addressed directly within these design paradigms. Hence, for example, the pattern language approach breaks large complex problems into smaller problems for which solutions consist of a certain configuration of functionality. In the urban context, the problem of maintaining access to both green and urban areas can be solved using a finger-type spatial configuration in the cities (Alexander et al. 1977), while the need for spiritually important sites can be solved using an island-type spatial configuration at a different scale.

3.3 Description and analysis

While support for quantitative analysis has been one of the strengths of geomatics tools, support for qualitative analysis and description has been much weaker. Perceptual GIS is one new form of description that is useful. The difficulty with these methods is that they may require a lot of data. The recent development of multi-agent systems with cognitive architecture provides an alternative that may serve a complementary role. Hence to the extent that such a model can reproduce household mobility behaviour, the model can be used to generalize this behaviour to a broader population and a larger spatial region. The results will contain a certain amount of uncertainty, but can be used both to deepen understanding and to support planning and design activities as described earlier.

As was the case for planning and design, qualitative spatial reasoning methods may be used to enhance the expressivity of descriptions. They provide a formal language for describing spatial configurations and the relations between them. This formal language can then be used to validate reasoning carried out with regards to such configurations. Such a language could also be used as a means of encapsulating spatial information present in existing textual descriptions. Today, much of the work aimed at the development of policy is based on the use of indicators. Indicators are a pseudo-qualitative form of information and are not always the best method of supporting policy.

4) Example of how these techniques could be applied

A problem involved in sustainable development is that of the behaviour of the poor in rural areas. These people, who are highly marginalized in our modern societies, are often characterised by strong cognitive constraints that affect their sense of independence and ability to survive. Hence the analysis of the problem requires an integrated approach that does not only account for observed behaviour but also cognitive motivational factors. Furthermore, because of their distance from population centres and their scattered spatial distribution, it is often difficult to develop reliable data about these people.

The use of standard geomatics tools in their current form is hence largely precluded. However, spatial constraints (or rather, spatio-temporal constraints) affect deeply the behaviours of this group of people. This is an area where appropriate geomatics tools might provide powerful means for understanding such marginalized behaviours and developing services and policies that are better adapted. The methods used must integrate the social, cognitive and spatial dimensions of the problem, however.

The problem involves both description/analysis to develop a better understanding and planning/design to develop a response. Furthermore, the spatial behaviour of these people may be described using our knowledge about the navigation and access issues. The problem of characterising the situation may be addressed in several distinct ways. Statistical techniques might be, and traditionally are used, but data are often scarce. Qualitative reasoning methods could be used to formalize a range of facts about such behaviours, and hence to test the logical entailments in scenarios of change. Throughout much of the world, the behaviour of the rural poor is a dominant problem of marginalisation, and, this is a population about which very little reliable data exists (FAO 2001). Here, the problem addressed may be partially a question of providing services, but an important issue is simply understanding the impact of subsistence living activities on the landscape.

One means of addressing this issue is to work with perceptual GIS methods. Using rules inferred from subsistence populations concerning the perceptual regions and lifelines involved, it should be possible to develop methods for determining the body, vista and environmental spaces of these groups. These spaces may then be used to aggregate other forms of environmental data in ways that correspond to the perceptions of these groups, and hence to develop an understanding of how they interact with the landscape. In addition, for nomadic subsistence living, the techniques will require appropriate additional modifications.

5) Discussion and conclusions

In this paper, a range of methods for handling qualitative spatial information has been presented. It has been argued that the emergence of such tools will provide stronger means for supporting improved understanding of the problems to be found in connection with sustainable development, including the development of appropriate policies and a better understanding of the impacts of these policies.

The techniques described are not yet widely used for these issues. They are the result of research undertaken by a broad spectrum of disciplines. Many are still undergoing intensive study in the laboratory, by a variety of groups in several different disciplines throughout the world. Hence no off-the-shelf capability exists to support these approaches. However, the development of a better appreciation for both the strengths and the weaknesses of these new tools is a requirement for their application by domain experts to real world problems. As the tools begin to be used outside of the theoretical contexts within which they were developed, there will be a growing need to work with domain experts to ensure that they are used wisely and appropriately, and to identify further problems that might be addressed by the research community to improve their utility.

6) Acknowledgements

This work is the result of extensive collaboration with a range of very insightful people over many years to whom I'm very grateful and was particularly stimulated by the networking generated within the GEOIDE network. This work was financially supported by a Canada Research Chair in Cognitive Geomatics.

7) References

C. Alexander, S. Ishikawa, M. Silverstein, M. Jacobson, I. Fiksdahl-King and S. Angel. 1977. A Pattern Language, New York: Oxford University Press.

Bittner, T.E. and G. Edwards. 2001. Towards an Ontology for Geomatics, Geomatica, Journal of the Canadian Institute of Geomatics, Volume 55(4), 475-490

Edwards, G., G. Ligozat, A. Gryl, L. Fraczak, B. Moulin, C.M. Gold, 1996. A Voronoï-based pivot representation of spatial concepts and its application to route descriptions expressed in natural language. Proceedings of the 7th International Conference SDH'96, Spatial Data Handling, M.J. Kraak, M. Molenaar (eds), Delft, The Netherlands, August, p. 7B1-7B15.

Edwards, G., and G. Ligozat 2002, A formal model for structuring local perceptions of environment space, in preparation

FAO, 2001, Forest Resource Assessments and Global Warming Report.

Goel, V., and J. Grafman. 2000. The Role of the Right Prefrontal Cortex in Ill-structured Planning. Cognitive Neuropsychology, Volume 17(5), p. 415-436.

Hagerstrand, T. 1970. What about people in regional science? Papers of the Regional Science Foundation, Volume 24, 7-21.

Johnson, M. 1987. The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reasoning. The University of Chicago Press, Chicago.

Montello D.R. 1993. Scale and Multiple Psychologies of Space, Proceedings of the International Conference COSIT'93, September, Elba Island, Italy, Lecture Notes in Computer Science, Volume 716, 312-321.

Reginster, I., and G. Edwards, 2001. The Concept and Implementation of Perceptual Regions as Hierarchical Spatial Units for Evaluating Environmental Sensitivity, Journal of URISA, Volume 13, No. 1, p. 5-16.


[1] Canada Research Chair in Cognitive Geomatics, Centre de recherche en géomatique, Département des sciences géomatiques. Le réseau GEOIDE, Pavillon Casault, Université Laval, Sainte Foy, Québec, Canada G1K 7P4. Email: geoffrey.edwards@geoide.ulaval.ca