Land and water resources model integration: software engineering and beyond

Integrated modelling has arisen as a valuable tool for assisting environmental management. Many existing land and water resources models are, however, not amenable to integration. To transform and develop models that serve integrated environmental management we need to address technical and social issues that include model reformulation and design, adoption of standard specification, notation and documentation on models and the reluctance of developers to adopt new design and programming approaches. There are tools and approaches that can aid the adoption of new approaches, and these are reviewed in the paper. Other developments that need to be infiltrated into the modelling community include facilities to flexibly examine, edit, visualise and analyse data at all stages of program execution, processes for user involvement in program and interface design, tools to assist knowledge-based selection of models and processes for undertaking and recording the peer reviewing of models. Options to address these issues are discussed, and it is considered that, despite the availability of suitable tools, the likelihood of getting adoption of high quality software engineering along with appropriate documentation and design practices will be low unless an integrated development approach is made, and the benefits made apparent to both developers and users. Context abstract: recent moves towards more integrated management of land and water resources have resulted in an increased focus on model integration. This focus has included the redesign of existing models to increase their capacity for reuse, often by using object-oriented approaches to produce flexible and reusable program modules. The redesign of existing models, as well as the design of new models, has, ideally, being done in a way that allows the program modules to be easily linked and run together. The adoption of sound software engineering practices, including the use of formalised modelling languages and modular development, is helping to achieve the goal of producing natural resources management modelling tool kits that allow complex, module-based, programs to be developed through drag-and-drop environments. Beyond software engineering there lie a number of issues that must be considered before the advantages of drag-and-drop integrated environmental modelling become apparent to natural resources managers and planners. These issues, including model design, notation, documentation and review, are discussed. (C) 2001 Elsevier Science Ltd. All rights reserved.