Modelling and architecture of a generic framework for integrative environmental simulations

In the face of Global Climate Change the scientific evaluation of possible impacts on the natural and social environment became more and more important in recent years. In this context computer-based simulation has emerged as a means of choice in many scientific disciplines as it facilitates predictions about future impacts of climate change based on well-known physical and societal relations. While in the past simulation models were developed within single disciplines one is nowadays particularly interested in interdisciplinary modelling which can be achieved by coupling of formerly independent simulation models from different disciplines. Thus the complex, mutually dependent processes occurring in nature and in socio-economic systems can be studied and analysed in the model. Coupling of simulation models from various disciplines is a non-trivial task; the problems one has to cope with range from the usage of different (physical) units, over diverse granularity of the models concerning space and time, up to completely different modelling and simulation approaches. It emerged – even in the context of this work – that methods of computer science render considerable support in this context. In this work the development of Generic Framework for Integrative Environmental Simulations is presented, which enables the coupling of spatially distributed, time-discrete simulation models exchanging data among each other via interfaces at run time. The framework was developed by applying a particular software engineering methodology – which is also presented within this thesis – based on modelling on different abstraction levels and several system views. For each abstraction level (requirements, design, components) precise rules for modelling as well as conditions for the consistency of structural and behavioural models are given. Moreover, an intuitive refinement relation between the single levels is introduced. For the Generic Simulation Framework under consideration the three system views “Data Exchange between Simulation Models”, “Modelling of the Simulation Space”, and “Time Coordination of Integrative Simulations” were identified and – emanating from a common base architecture – independently elaborated in the course of this thesis. Finally, the architectures of the different views are integrated to obtain the complete framework architecture. Furthermore, we provide guidelines for the application of the framework for creating an integrative environmental simulation system and describe an implementation of the framework as distributed system in Java. Finally we report on the successful application of the framework