Constructing a library of domain knowledge for automated modelling of aquatic ecosystems

Conceptual mathematical modelling of aquatic ecosystems comprises a considerable amount of knowledge reflected through a vast variety of different models that can be found in literature. While there is a growing interest in developing unifying documentation systems that allow storage of these models, not much work has been done yet on formalization and storage of the modelling knowledge itself. Such formalization would allow for better sharing and exchange of knowledge between experts on one hand and make it available to computational methods for modeling on the other. The knowledge library we develop here covers the knowledge in the domain of food web modelling in lakes based on differential equations. We illustrate the generality of the knowledge in the library through reconstruction of three well-known models of different complexity from the library, i.e. [Vollenweider, R.A., The Scientific Basis of Lake and Stream Eutrophication with Particular Reference to Phosphorus and Nitrogen as Eutrophication Factors. Organisation for Economic Cooperation and Development, Paris, 1968; Imboden, D., Phosphorus model of lake eutrophication. Limnol. Oceanogr. 19 (1974) 297-304] and SALMO model [Bendorf, J., A contribution to the phosphorus loading concept. Int. Revue ges. Hydrobiol. 64 (2) (1979) 177-188; Recknagel, F., Systerntechnische Prozedur zur Modellierung und Simulation von Eutrophierungs-prozessen in stehenden und gestauten Gewdssern: Sektion Wasserwesen, TU Dresden, Dresden, 1980]. We also illustrate how computational methods for model induction from data can benefit from the developed library of knowledge. (c) 2005 Elsevier B.V All rights reserved.