Model-driven Development of Environmental Modeling Languages: Language and Model Coupling

Abstract

Common characteristics of environmental modeling and simulation (M&S) are multi-disciplinary modeling and the need to reuse models in different contexts. There is a plethora of M&S frameworks available, but still challenges remain, when models are developed across different communities and organizations with specific modeling paradigms, languages, and tools. Issues arise due to technical complexity, model reuse, and model integration. We target these issues with the provision of declarative domain-specific modeling languages (DSLs) that allow for problem-oriented and technology-independent modeling. With this approach, DSLs are used to define technology independent environmental simulation models. Executable code that conforms to a simulation technology of choice is automatically generated from these models, thus the models can be reused on different platforms. The implementation of DSLs causes effort. However, the model-driven approach for the definition of DSLs, as opposed to grammar-based approaches, allows the efficient definition of DSLs and respective tools, in particular with respect to language coupling. Language coupling is necessary for multi-disciplinary modeling, where different parts of a model may be defined using different DSLs. In this paper, we present a model-driven language engineering approach and show how object-oriented language modeling can be the base for coupling DSLs. The semantics of coupled models are based on established concepts of model decomposition and event-driven simulation. Since many generic simulation technologies implement these concepts, this facilitates model reuse on many platforms, although programming languages and interfaces might differ. We applied this approach to a DSL for Cellular Automata modeling and a simple DSL for describing computational agent models and defined code generation for one exemplary simulation framework combined with Geographic Information System technology (GIS). So far, the DSL has been used to reimplement published models of fire spread and seismicity. However, the approach is not limited to the presented DSLs, framework technologies, and application areas.