Jeroen Keppens and Qiang Shen
Division of Informatics, University of Edinburgh, Email: fjeroen,qiangsg@dai.ed.ac.uk
Abstract: This paper is concerned with the use of compositional modelling for effectively building and (re-)using repositories of ecological models, with a focus on the important issue of model granularity. The techniques presented herein differ from existing work in that a meta-modelling approach to grain choices is taken by transforming an emerging model's level of detail through disaggregation. The result is a means of choosing a model's level of detail that is sufficiently flexible for eco-modelling and that allows grain choices to be described in terms of scenario-level concepts. Model disaggregations are implemented as model transformations. Thus, different disaggregations of a model can be composed themselves. As such, a small number of disaggregations may encompass a much larger selection of models at different levels of detail. The paper shows that due to the use of dynamic constraint satisfaction techniques, grain choices are not necessarily made from the outset, but can be postponed until after an initial model is constructed.