The design and implementation of an ecological simulation development environment

Abstract

Researchers have suggested two approaches for designing software development environments (SDEs) aimed at alleviating the software crisis: an evolutionary approach that supports the contemporary software life cycle model and a revolutionary approach that supports an automated life cycle model. An alternative approach is presented that is based on a "profession oriented" automated SDE and that is possible with today's technology. The design and implementation of an ecological simulation development environment (ESDE) for aiding the non-professional programmer in the construction of ecological simulation models is described and demonstrated. The ESDE is composed of (1) a simulation language, (2) a program editor, (3) a report generator, and (4) a program generator. The simulation language has been designed at the specification level and is used to textually specify conceptual ecological system models defined using simple box-and-arrow diagrams. The program editor is used to enter and maintain specifications and has been designed to prevent syntax and semantic errors from occurring using the grammar rules of the simulation language. The program generator and report generator use the specifications to automatically produce portable FORTRAN-77 source code computer models and to print summary reports, respectively. Facilities have been provided for extending the language with user-written FORTRAN function subprograms and for performing sensitivity analyses on model parameters. The ESDE has been designed for automated model development and maintenance on a microcomputer workstation. It is demonstrated using a theoretical fish harvest model. An evaluation of the ESDE from a graduate class in systems analysis indicates that the ESDE can reduce the amount of time and effort required for modelers with little or no programming or modeling experience to build a computer model. Future extensions to the ESDE also are discussed. The ESDE is shown to have implications in the areas of ecological simulation modeling and simulation modeling in general and in the areas of programming language and SDE design.