Environnement : mod�lisation et mod�les pour comprendre, agir ou d�cider dans un contexte interdisciplinaireEnvironment: modelling and models to understand, to manage and to decide in an interdisciplinary context

Abstract

If the physical sciences were the cradle of modelling approaches, now almost all disciplines are concerned by this methodology. First based on mathematical language, the modelling formalisms become more and more diverse. Computer sciences contribute widely to the extension of the methodology at two levels : the first to make easier the use of mathematical models, the second to develop new formalisms, such as cellular automata or neural networks. The fields of basic and applied research on the environment and related technological developments give good examples of applications of this methodology, from global modelling at the planet level, for example for the study of climate changes, to local and specialised models, for example in population dynamics or water transfers in soils. Models are also developed to reach various goals : in basic research, to understand how environmental systems work, in technological development, to design adapted techniques, and in decision making process, as an argument, among others, to elaborate a decision. The recent expansion of modelling, to almost all disciplines implied in environmental research, makes it also a tool in interdisciplinary dialog. The model becomes a link between disciplines, for example between natural and social sciences, by combining dynamics of ecological and social systems. But the use of models is not exempt of hazards. Used in bad conditions, off its limits of validity, or running with a deficient simulator, it may lead to incorrect results. It may also be a screen to hide an insufficient knowledge. It such cases, it can be a wrong decisive argument in decision process or to lead to a bad technical device or protocol. Anyway, built and used carefully it is a precious and efficient tool. So, the state of the art and actual methodological problems and applications in environmental basic and technical research are presented. The foreseeable advances and endeavours are also envisaged. The text is essentially illustrated by examples developed by scientists and engineers of the Cemagref. Finally, the environment is an example of a complex system. Modelling is a good tool for complexity analysis and methodological advances in environmental research would be useful for other systems which are qualified “complex”, and then have a larger range of application than studies on environment. Tous les secteurs scientifiques sont maintenant concernés par la modélisation. Méthodologie en plein devenir, elle est source de progrès, mais son emploi n'est pas sans risque. Les recherches sur l'environnement et les technologies afférentes concernent de nombreuses disciplines ; chacune d'entre elles a sa propre expérience de la modélisation. Outre son intérêt propre, le champ de la recherche sur l'environnement constitue donc un bon laboratoire où s'échangent des savoirs et des savoir-faire et où une méthodologie commune peut être développée. Tout en privilégiant un point de vue sur la modélisation, celui s'intéressant à la dynamique des systèmes, il s'agit d'en décrire les principes, de citer les principaux problèmes environnementaux où la modélisation intervient de façon déterminante, d'en cerner les limites et d'identifier les progrès à faire. On remarquera, qu'en plus de son intérêt propre, pour comprendre et résoudre un certain nombre de problèmes, la modélisation devient de plus en plus un moyen de dialogue entre les disciplines. Enfin, l'environnement est un bon exemple de système complexe. La modélisation apparaît efficace pour l'étude de tels systèmes et les progrès méthodologiques réalisés en matière de recherches sur l'environnement ont sans doute une portée plus générale et peuvent être utiles pour aborder d'autres systèmes, eux-aussi qualifiés de complexes.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 27 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.