Feedback loops and temporal misalignment in component-based hydrologic modeling

• Published in: Water resources research 2011-12, Vol.47 (12), p.n/a
• Main Authors: Elag, Mostafa M., Goodall, Jonathan L., Castronova, Anthony M.
• Format: Article
• Language: English
• Subjects: Boundary conditions; Communication; computational methods; Computer science; Data exchange; Geographic information systems; High performance computing; hydrologic modeling; Hydrologic systems; Hydrology; Information systems; Information technology; Interfaces; modeling frameworks; Sediment-water interface; Simulation; Software; Water resources management; Watersheds
• ISSN: 0043-1397; 1944-7973
• DOI: 10.1029/2011WR010792

In component‐based modeling, a complex system is represented as a series of loosely integrated components with defined interfaces and data exchanges that allow the components to be coupled together through shared boundary conditions. Although the component‐based paradigm is commonly used in software engineering, it has only recently been applied for modeling hydrologic and earth systems. As a result, research is needed to test and verify the applicability of the approach for modeling hydrologic systems. The objective of this work was therefore to investigate two aspects of using component‐based software architecture for hydrologic modeling: (1) simulation of feedback loops between components that share a boundary condition and (2) data transfers between temporally misaligned model components. We investigated these topics using a simple case study where diffusion of mass is modeled across a water‐sediment interface. We simulated the multimedia system using two model components, one for the water and one for the sediment, coupled using the Open Modeling Interface (OpenMI) standard. The results were compared with a more conventional numerical approach for solving the system where the domain is represented by a single multidimensional array. Results showed that the component‐based approach was able to produce the same results obtained with the more conventional numerical approach. When the two components were temporally misaligned, we explored the use of different interpolation schemes to minimize mass balance error within the coupled system. The outcome of this work provides evidence that component‐based modeling can be used to simulate complicated feedback loops between systems and guidance as to how different interpolation schemes minimize mass balance error introduced when components are temporally misaligned. Key Points Componentization is a promising approach for building community models It is possible to simulate fully‐coupled systems as separate model components Interpolation schemes can be used to rescale data between misaligned components