ALBERT

Description: An important aim of modelling complex environmental systems is to embed scientific outcomes into applications. With this characteristic, not only the modelled systems, but also the required technical realizations are complex. Managing interdisciplinary, interorganisational modelling tasks and dealing with complex heterogeneous technologies are challenges the modellers and respective institutions are faced with. In practice, issues arise due to technological heterogeneity across organisational and disciplinary boundaries, model reuse within different application contexts, and integrative system specification. Service-oriented Architecture (SOA) with Web Services and Workflow Management provide a conceptual framework and a technological basis to deal with these problems. As part of a web-based disaster information platform established in Germany, we implemented the SOA-based prototypical distributed information system NaDiNe-Dike. It combines a numerical dike breach simulation model, spatial data handling, simulation management, and a web-based user interface. All functionalities of the system are provided by Web Services. The execution of these services is managed by a Workflow Management System. In this paper, we describe how various existing technologies, including heritage simulation and other off-the-shelf software (i.e. OGC-WS), have been integrated using SOA. This approach is particularly flexible by supporting the reuse and recombination of functionalities (e.g. simulation models) in different contexts and applications. We discuss the main characteristics of the implemented system and conclude that the main advantages using SOA in such a way are the possibility to distribute functionality of a system spatially and functionally across institutions, programming languages and operating systems. Moreover, SOA provides enough flexibility to integrate simulation models with data management and presentation technologies that originate in different technological spaces and may be provided and maintained by different spatially distributed organizations. The possibility to reuse software components helps to reduce redundant implementation efforts and eases the transfer from science to the application domain. The main disadvantages of the approach followed here are the use of the network as the coupling media, the need to redesign heritage software, and the lack of specific simulation support. Moreover, the introduction of the component-based software implementation paradigm into the participating organizations is required, which must be accompanied by adequate system specification techniques. However, SOA provides the technological basis for a software infrastructure that reflects the distributed nature of today's modelling tasks, enhancing the division of responsibilities, thus, making modelling more efficient in the long run.

Description: Simulation models are commonly developed by scientists to evaluate scenarios that show potential developments of the Earth system in the past, present and future. In order to provide adequate information systems that facilitate access to simulation models to a broad, heterogeneous user community, the deployment of such models on web servers provides a technical fundament. However, many aspects need to be taken into account for setting up operational, user-friendly web-based systems that include access and administration tools for simulation models. Data integration, data exchange, scenario management, and visualisation are among the most important functionalities to be accounted for, while usability needs to be aimed at by choosing an appropriate abstraction level and providing a careful interface design. Usually simulation models encapsulate complex algorithms, which have been developed by domain experts and implemented based on very diverse technologies. In order to provide the functionality of such models to users over the Internet, standardisations such as the Web Processing Service developed by the Open Geospatial Consortium (OGC) help to specify the technological framework, but do not provide concepts for guaranteeing the aforementioned functionalities and usability. An additional requirement from the administrator and developer perspective is to offer a minimum level of flexibility in information architectures in order to adapt and exchange single components such as a simulator or data base. In many cases, this flexibility stands in conflict with a rapid, use-case specific development. In this paper, different integration concepts for hydrological simulation models into web-based management systems are compared to each other. All concepts were developed to fulfil the requirements of heterogeneous user groups, ranging from scientists to re-insurance companies. Their implementation in prototypical realworld systems was performed in inter-disciplinary groups of experts in Hydrology and Information Technology. While the first three integration concepts focus specifically on functionality (legacy model encapsulation, integration of real-time data, scenario management) and usability (user interface, visualisation) for single simulation models or static process chains, the fourth use case outlines a way towards more generic service composition based on a workflow management system. A comparison of the potential and limitations of these architectures results in a discussion of aspects to be taken into account for making simulation models accessible and usable for science, industry and governmental agencies. From our experience of designing, setting up and running the developed systems we conclude that functionality and usability are in the main interest of the end users of such systems. Each user group has different requirements, depending on their expertise and objectives. However, a clear, easy-to-use user interface is far less error-prone and avoids semantic problems for lay users, while experts require complex control mechanisms to run, calibrate or even re-design their modelling infrastructure. Integrating third-party data sources is possible, but requires well-defined machine readable user interfaces. For system administration and sustainability, system architectures incorporating a higher flexibility and implementation effort in the setup phase are seen to pay off in the long term. It is very important that all relevant aspects have to be specified in the design phase of a web-based management system for simulation models. Depending on this specification, the target system focuses either on implementation speed or flexibility, which comes with the cost of a more complex service-based infrastructure. The demand for using and accessing simulation models has increased in number and complexity in recent years. With the availability of appropriate concepts and technologies from information technology, integrating such systems into a web environment is a worthwhile, yet challenging task for the modelling and Information Technology communities.