A framework for a modular modelling approach for agroforestry

There are a number of problems with conventional methods for implementing models of agroforestry systems, including the difficulties of building, modifying and understanding them. One way of addressing these problems is to adopt a modular modelling approach, in which a model is constructed by linking a number of submodels, which in turn may consist of smaller computational units.This paper presents a basic framework for the design of a modular modelling approach suitable for agroforestry modelling. This approach is based on a close correspondence between the notion of module in modelling and of subroutine in a programming implementation of a model, so that selecting and linking modules is analogous to choosing and issuing calls to subroutines. This approach has the potential for enabling the complex modelling structures that are required for agroforestry modelling, such as spatial aspects and the representation of individual trees, to be handled in a consistent and uniform manner.The approach requires that subroutines should be held as symbolically-represented structures, rather than just source code in a programming language, so that they can be referenced as discrete units, and internally modified by symbol-processing operations. It is shown that the logic-programming language Prolog is well-suited to the task of representing subroutines in template form, ready to be fleshed out and combined into a single program in response to the selections made by a user. Prolog's suitability for the development of a user-friendly model design interface is also discussed, enabling models to be constructed simply by selecting options from a menu of design choices.