Modelling agricultural production

In modelling in general and biological modelling in particular two approaches may be distinguished: a descriptive and an explanatory approach. In descriptive models the system and its behaviour are described at the same level at which the observations about it are made. A good example are the chilling unit models that were discussed in this symposium and are used to calculate at what time the temperature demand is met to break the dormancy of buds. Another example is the statistical analysis of crop-weather relations. In the process of eliminating non-significant terms, the latter may lead to the well known regression equations that relate yield to for instance radiation, temperature and rainfall in certain months or decades of the year in linear, quadratic and cubic terms. The use of this statistical blunderbuss-approach, to borrow a term of Monteith (1981), is facilitated by the widespread availability of computers and software for the purpose. However, it remains not only clumsy, but also to a large extent uninstructive because it ignores the underlying mechanisms and their interactions and builds hardly on existing knowledge. At best the results describe observed situations, but they hardly help to organize thoughts about the possible. The other approach is the construction of explanatory models in which the processes that occur in the system form the basis of the model. It is then not the behaviour of the system as a whole, but the underlying processes that are analysed and described. Explanatory models suggest themselves in biology because various levels of organisation are distinguished in this science. These levels may be classified according to the size of the elements that are distinguished as those of molecules, cell structures, cells, tissues, organs, individuals, populations and ecosystems. Models that are made with the objective to explain are bridges between such levels of organisation: they allow the understanding of larger systems on the basis of knowledge gained by experimentation on smaller systems. In this way the properties of membranes may be understood better by studying molecules and the properties of ecosystems by studying species. The "blunderbuss models" are both static and descriptive, but there are also static models of the explanatory type. An example is the model of Penning de Vries et al. (1974) that contains all the necessary information to calculate the relation between respiration and growth on basis of the underlying biochemical processes. Another example are light-models that calculate the transmission, reflection and internal distribution of light by crop canopies, based on architecture, leaf properties, amount of direct and diffuse light, solar position and so on. By now, this latter family of static, explanatory