COMPUTER TECHNOLOGY IN ENVIRONMENTAL EDUCATION: A QUALITATIVE MODELLING TOOL G. K. Adam Department of Planning and Regional Development University of Thessaly Pedion Areos, 38334 Volos, Greece E-mail: gadam@prd.uth.gr ABSTRACT The advances in computer technology, although are quite significant -we can talk and work on a virtual world- are still far from providing a really free of environmental problems world. In this paper a case study is presented of the research application of a qualitative modelling and simulation tool in solving environmental problems associated with certain physical phenomena, considering and estimating a large number of multi-variable parameters, and providing suggestions for the ways these phenomena should be tackled in a way that is friendly to the environment. Η ΤΕΧΝΟΛΟΓΙΑ ΤΗΣ ΠΛΗΡΟΦΟΡΙΚΗΣ ΣΤΗΝ ΠΕΡΙΒΑΛΛΟΝΤΙΚΗ ΕΚΠΑΙ ΕΥΣΗ: ΕΦΑΡΜΟΓΗ ΠΟΙΟΤΙΚΗΣ ΜΟΝΤΕΛΟΠΟΙΗΣΗΣ Γ. Κ. Αδάµ Τµήµα Χωροταξίας και Περιφερειακής Ανάπτυξης Πανεπιστήµιο Θεσσαλίας Πεδίον Άρεως, 38334 Βόλος E-mail: gadam@prd.uth.gr ΠΕΡΙΛΗΨΗ Η πρόοδος της τεχνολογίας, και ιδιαίτερα της πληροφορικής, παρόλο που είναι σηµαντική - µπορούµε να αναφερόµαστε και εργαζόµαστε σε ένα εικονικό κόσµο- απέχει ακόµα από το να παρέχει ένα περιβάλλον πραγµατικά ελεύθερο από περιβαντολογικά προβλήµατα. Σε αυτήν την εργασία παρουσιάζεται η περίπτωση µιας ερευνητικής εργασίας εφαρµογής ενός συστήµατος ποιοτικού µοντελισµού και προσοµοίωσης στην επίλυση προβληµάτων σχετικά µε συγκεκριµένα φυσικά φαινόµενα, λαµβάνοντας υπόψη και υπολογίζοντας ένα µεγάλο αριθµό πολυµεταβλητών, και παρέχοντας προτάσεις για την αντιµετώπισή τους.
1. INTRODUCTION During the last decade, the efforts for environmental protection have been increased tremendously. Although in most of the cases the problems arisen (e.g., air pollution and climatic changes) are due to certain uses of technological advances, at the same time technology provides most of the basic tools used to confront these problems [1, 2]. In that direction, a large number of research methodologies and computer applications have been investigated and applied in environmental sciences [3, 4]. However, the advances in computer technology, although are quite significant -we can talk and work on a virtual world free of environmental problems- are still far away from providing a really free of ecological problems real world. Computer simulation models of real world situations, provide significant assistance in many cases, particularly in the estimation and anticipation of various physical phenomena for civil, health and military purposes [5, 6, 7]. The results obtained so far are considerable valuable. In this paper, as a case study of the research work carried out, is presented the application of a qualitative modelling and simulation tool in solving environmental problems associated with certain physical phenomena (e.g., meteorological), considering and estimating a large number of multi-variable parameters and providing suggestions for the ways these phenomena should be tackled in a way that is friendly to the environment. After all, we must not forget that educating humans to respect and protect the environment -enviromental education- is not just one of the major requiremements to be considered, but is more like an obligation which we must always fulfil. Our qualitative modeling system QMTOOL [8] is used, in conjunction with other logical reasoning approaches [9, 10], to develop and test effective and flexible system models. The system development was based mainly on object-oriented programming techniques and fourth generation languages. 2. QUALITATIVE MODELLING AND SIMULATION METHODOLOGY In recent times the use of model-based reasoning has been constantly increasing. In particular, techniques have been developed, mainly by the Artificial Intelligence community, that try to define a system qualitatively [11, 12]. The qualitative behaviour of a system is commonly described by a set of rules connecting causes (inputs) and effects (outputs), and qualitative differential equations (confluences), which are derived from the qualitative equations for the system. Commonly, in order to construct an efficient model, one that is close to representing the real system for a given task, the governing system equations must be solved. A variety of methods and algorithms have been generated to do this qualitative reasoning [13]. In our system during the qualitative modelling process a number of system parameters, required to describe the behaviour of the system, are selected and defined (set of symbols) within the representative model. The relationships between these variables are determined (constraint equations), according to the range of landmark values that define the magnitude of the variables. These complete sets of symbols and constraints, together with a set of initial qualitative values for the parameters, form the system model. A flowchart of the qualitative modelling and simulation methodology applied in most of the cases being examined, is presented in following Figure 1.
Purpose Definition Task Specification System Definition System Variables System Range System Relations Model Data System Simulation Behaviour Generation Prototype Model compare System Verification modify Modified Behaviour Evaluated Model Implementation Desired Model FIGURE 1. Qualitative modelling and simulation methodology. Although is known that qualitative modelling approaches do not always provide such a completely accurate model representation, compared to conventional (analytical) approaches, they have proved particularly useful and effective in certain domains. Analytical approaches, although are capable of detailed system design analysis, they often require a considerable amount of technical data; even model building is a time-consuming procedure based on extensive written specifications. In contrast, our qualitative modelling approach utilizes simpler computational mechanisms, symbolic programming, flexible and interactive AI techniques, that offer multiple ways of reasoning about model prototypes and rules associated with their simulated behaviour, based on relatively small amount of input data (qualitative information). This qualitative basis produces more comprehensive results, similar and closer to human reasoning and understanding on systems behaviour. Such an approach has been successfully applied in cases where conventional methods, traditional modelling systems, are too complex or less effective [14, 15]. It has been shown to require less numerical computations, and often provides unique solutions. We define a system model as a structure of interconnected system components (set of variables and their connections), that determine the important characteristics of the system being modelled. A set of variables is represented by a number of software entities (objects) termed 'input', 'state', and 'output', the attributes of which are specified qualitatively by assigning qualitative values to them, such as 's' (small), 'l' (large), etc. The set of relationships describing how these variables are linked together is represented explicitly by graphical connection via the user interface, and maintained by a 'connection' system object. Similarly these connections are also specified qualitatively and assigned associated qualitative values.
In a few words, in order to avoid lengthy explanations of the internal conversion and value calculation mechanism, the basic steps of this process are: the operating range of a variable is determined and divided (qualitative partitioning) by the amount of qualitative values this variable can obtain, and the result is multiplied by the actual sign the variable is assigned to. The numerical value is obtained by multiplying the obtained result with a numerical factor that corresponds to this qualitative value. However, in practice we have accepted the approach that this value should actually be between its respective step-range determined by the upper and lower limit values. The qualitative modelling tool produced, based on direct manipulation, allows the designer to concentrate on the design alternatives rather than be continually building models through programming. In particular, the tool incorporates and deals with some of the following features: fuzzy relationships between system variables (using symbolic value representations 'xl', 'l', 'm', 'xs', 's', and qualitative partitioning); symbolic computation (presented as set of 'input', 'state' and 'output' visual objects embedded with behavioural rules and functionality); mapping mathematical equations (relationships) into qualitative descriptions (using functional 'M + ', 'M - ', arithmetic 'ADD', 'MINUS', and derivative 'incr', 'steady', 'decr', constraints); generating behaviour as a sequence of time-varying qualitative states (represented graphically as cartesian plots); System models are created at the design specification stage by the user selecting objects to represent the functional elements in the system. The user can then interconnect the elements to produce a representative model of the system. The number of variables selected to form the system model depends on the specific task. However, the user must ensure that the number of variables chosen within the model is capable of describing the behaviour of the system, as this will define the precision of the generated results. At run-time the user can monitor the behaviour of important elements, by graphical analysis to verify overall system performance. An example of a system model representation is shown in Figure 2. The actual relations between the variables shown could be represented qualitatively in the following way: M + (Stvar,Invar), M - (Stvar,Stvar), f(outvar,stvar), M + (...,..), M - (..., ), f(..., ), where M +, M - and f similarly to QSIM [12], could simply indicate that there is a relationship (influence) between these variables. Running the simulation of this model, the behaviour derived from this formal structure is generated, and knowledge is acquired that could be used to compare its behaviour with that of the physical system, or for further tasks. During this simulation phase, the user has the opportunity to perform a final evaluation of the system design and adjust the model's structure, so that its performance corresponds to a predicted behaviour. At this stage a detailed analysis can assist understanding of the behaviour of the real system. For this reason, in every simulation cycle, when the calculation process is completed (a complete cycle), the variables outputs (cartesian plots) are graphically displayed on the canvases. The changes in the qualitative values of the monitored variables are mapped into the y cartesian axis, while x axis indicates the number of discrete time steps (cycles). The simulation can also run in single steps that allows the performance of each variable to be traced individually.
FIGURE 2. Qualitative Modelling and Simulation Tool sample view. 3. EXPERIMENTAL CASE STUDIES The overall system mentioned above is still under development and continuous integration with further functionality and tools. At the time being, the system has been used mainly into experimental educational applications and some research in conjunction with a geographical information management system [16]. In particular, it has been applied in modelling and simulation of environmental data, in geographical data processing and evaluating various physical phenomena, considering a large amount of variables from various web sources [17]. In most of the cases reference maps have been modelled for the purpose of land usage identification, and in consequence perform their analysis based on simulation to derive valuable information, such as population distribution, standard deviations, subtotals and groupings, etc. In Figure 3, a report of such environmental data is presented, produced on a search query with given criteria. FIGURE 3. A sample report of environmental data analysis search.
Based on observations, comments and further feedback about the system's usage and applications, the trainee is able to understand the basic concepts that describe the environmental information systems and follow step by step the normal flow of the information provided. On each step, all the required detailed information for the requested subjects is provided, as well as examples for their comprehension. As a result the trainee learns about the nature and structure of the actual environmental data, as well as the procedures employed for their manipulation (insertion, classification, etc.). He is capable to perform various data selections according to the task he is engaged with, and use the appropriate procedures to acquire them. Further on, the trainee could proceed with data analysis procedures provided at the system, select and use the methods he chooses as appropriate for each case study, and in this way learn about the analysis methods provided within this environment. Finally, produce results and conclusions in various presentation forms (text, tables, graphs, etc.). 4. CONCLUSIONS Full realization of environmental protection requires the progressive and proper use of information technology in all aspects of technical, organizational and political life. In this work, we have dealt with a qualitative modelling tool for education and research in environmental systems applied in local research case studies. Models were build through direct manipulation and interconnection of system components selected from the components library. Model execution produces graphical outputs that describe the behaviour obtained from system models, upon which analysis is performed and conclusions are derived. The development work was based on classical and modern techniques (symbolic computational methods), further analysed and merged for this qualitative modelling and simulation tool to be achieved. The final toolkit consisted of a system model creation and execution units, developed to assist system design, process analysis and implementation tasks. The incorporated user interactive utilities provided an easy creation and a friendly manipulation of system models. Studying the behaviour of such models has increased our interpretation and understanding of real world physical systems behaviour, and has made our task easier in providing multiple ways of reasoning, and more details of the environmental processes. Currently, the system is going under further development aiming at being used as a distance learning tool in environmental research and education. In the near future the integration of spatial models into the system is scheduled, in conjunction with the geographical information management system mentioned above [16], and their implementation into the evaluation of environmental information. REFERENCES 1. Lepper M., H. Scholten and R. Stern (1995) The added value of Geographical Information Systems in Public and Environmental Health, Kluwer Academic Publishers. 2. Antenucci J., K. Brown, P. Croswell and M. Kevany (1991) Geographic Information Systems - A guide to the technology, Chapman & Hall. 3. Mattheck C. and I. Tesari (2000) Design in nature, Proc. 8 th Int. Conf. Development and Application of Computer Techniques to Environmental Studies, (eds. G. Ibarra-Berastegi, G.A. Brebbia and P. Zannetti), Bilbao, Spain, June 28-30, 2000, pp. 217-226. 4. Rossi F., M. Folino and F. Lamberti (2000) An advanced set of tools for the information management of natural hazards, Proc. 2 nd Int. Conf. Management Information Systems
Incorporating GIS & Remote Sensing, (eds. C.A. Brebbia and P. Pascolo), Lisbon, Portugal, June 14-16, 2000, pp. 91-101. 5. Karavezyris V. and J. Aizpuru (2000) Modelling systems of waste economy - A simulation example, Proc. Int. Conf. Protection and Restoration of the Environment V, (eds. V.A. Tsihrintzis, G.P. Korfiatis, K.L Katsifarakis and A.C. Demetracopoulos), Thassos Island, Greece, July 3-6, 2000, pp. 819-826. 6. Adam G. K., (2000) Expanding Environmental Education Using GisTool, Proc. Int. Conf. Protection and Restoration of the Environment V, (eds. V.A. Tsihrintzis, G.P. Korfiatis, K.L Katsifarakis and A.C. Demetracopoulos), Thassos Island, Greece, July 3-6, 2000, pp. 1215-1220. 7. Zidek J., Jean Meloche, Nhu D Le and Li Sun (2000) Combining statistical and computer models for health risk assessment, Proc. 15 th Int. Workshop Statistical Modelling, (eds. V. Nunez-Anton and E. Ferreira), Bilbao, Spain, July 17-21, 2000, pp. 95-106. 8. Adam G. K. and E. Grant (1994) QMTOOL- a qualitative modelling and simulation CAD tool for designing automated workcells, Proc. Int. IEEE Conf. Robotics and Automation, San Diego, California, USA, May 8-13, 2000, pp. 1141-1146. 9. Law M. A. and D. W. Kelton, (1991) Simulation Modelling and Analysis, McGraw-Hill International Editions. 10. Sommervile I. (1995) Software Engineering, Addison-Wesley. 11. De Kleer J. (1984) How Circuits Work Journal of Artificial Intelligence, Vol. 24, pp.205-280. 12. Kuipers B. (1986) Qualitative Simulation Journal of Artificial Intelligence, Vol. 29, pp. 289-338. 13. Diaz A., R. Orchard, J. Amyot, and J. Brahan (1991) Qualitative modelling techniques for process control and diagnosis, Proc. TAPPI Engineering Conference, Institute for Information Technology, National Research Council Canada, Ottawa, Ontario, Canada. 14. Adam G. K., E. Grant and K. Adam (2000) Qualitative modelling and control of industrial processes, Proc. Int. IASTED Conf Modelling and Simulation, (ed. M.H. Hamza), Pittsburgh, Pennsylvania, USA, May 15-17, 2000, pp. 477-482. 15. Adam G., S. Tzortzios and N. Gitsakis (2000) Management of agricultural data of specific cattle breed using qualitative models, Proc. Int. IFAC Conf. Modelling and Control in Agriculture, Horticulture and Post-harvested Processing, Wageningen, the Netherlands, July 10-12, 2000, pp. 153-158. 16. Adam G. K. (1999) GisTool: A tool for geographical data management and processing, Proc. PanHellenic Conf. Geographical Information Systems I, Athens, Greece. 17. Adam G. K. (2000) GisTool: An integrated environment for GIS data management, Proc. 2 nd Int. Conf. Management Information Systems Incorporating GIS & Remote Sensing, (eds. C.A. Brebbia and P. Pascolo), Lisbon, Portugal, June 14-16, 2000, pp. 267-273.