A System of Systems Approach to Integrating Global Sea Level Change Application Programs

The global sea level change application community has numerous disparate models used to make predications over various regional and temporal scales. These models have typically been focused on limited sets of data and optimized for specific areas or questions of interest. Increasingly, decision makers at the national, international, and local/regional levels require access to these application data models and want to be able to integrate large disparate data sets, with new ubiquitous sensor data, and use these data across models from multiple sources. These requirements will force the Global Sea Level Change application community to take a new system-of-systems approach to their programs. We present a new technical architecture approach to the global sea level change program that provides external access to the vast stores of global sea level change data, provides a collaboration forum for the discussion and visualization of data, and provides a simulation environment to evaluate decisions. This architectural approach will provide the tools to support multi-disciplinary decision making. A conceptual system of systems approach is needed to address questions around the multiple approaches to tracking and predicting Sea Level Change. A systems of systems approach would include (1) a forum of data providers, modelers, and users, (2) a service oriented architecture including interoperable web services with a backbone of Grid computing capability, and (3) discovery and access functionality to the information developed through this structure. Each of these three areas would be clearly designed to maximize communication, data use for decision making and flexibility and extensibility for evolution of technology and requirements. In contemplating a system-of-systems approach, it is important to highlight common understanding and coordination as foundational to success across the multiple systems. The workflow of science in different applications is often conceptually similar but different in the details. These differences can discourage the potential for collaboration. Resources that are not inherently shared (or do not spring from a common authority) must be explicitly coordinated to avoid disrupting the collaborative research workflow. This includes tools which make the interaction of systems (and users with systems, and administrators of systems) more conceptual and higher-level than is typically done today. Such tools all appear under the heading of Grid, within a larger idea of metacomputing. We present an approach for successful collaboration and shared use of distributed research resources. The real advances in research throughput that are occurring through the use of large computers are occurring less as a function of progress in a given discrete algorithm and much more as a function of model and data coupling. Complexity normally reduces the ability of the human mind to understand and work with this kind of coupling. Intuitive Grid-based computational resources simultaneously reduce the effect of this complexity on the scientist/decision maker, and increase the ability to rationalize complexity. Research progress can even be achieved before full understanding of complexity has been reached, by modeling and experimenting and providing more data to think about. Analytic engines provided via the Grid can help digest this data and make it tractable through visualization and exploration tools. We present a rationale for increasing research throughput by leveraging more complex model and data interaction.