Author(s): Jean-Michel Perraud; Biao Wang; Jai Vaze
Linked Author(s):
Keywords: No Keywords
Abstract: The Catchment Water Yield Estimation Toolset (CWYET) is a software toolset for estimating water yield over up to hundreds of catchments, featuring capabilities for calibration, catchment cross-verification, ensembles of models and scenario modelling such as impact of climate change. These uses require building alternate model configurations by alternating measured or hypothetical climate inputs, model parameterization, objective functions, state initialization, etc. Figuratively, this is not different from the assembly of Lego? blocks. This information needs to be managed for the traceability and reproducibility of scientific experiments. We present in this paper an entity-relationship data model (ERM), i. e. a conceptual representation of the data, in the broad sense, associated with these catchment model “Lego” building blocks. While this ERM information modelling technique is widespread in the business systems design, it seems to be uncommon in scientific software design. CWYET has multiple requirements, sometimes conflicting, for this ERM. One central requirement is to build and execute catchment simulation models from a software scientific workflow system, the Hydrologists’Workbench. To capture the design of the data model, we use the Microsoft Entity Framework, using the so-called “model first” approach. In order to maintain some capacity to evolve and adapt the data model for future need, we use code generation to reduce the coding tedium to a minimum. We present an assessment of the benefits and smaller inconvenients of this approach, notably compared to the more ad hoc approach to the management of model configuration that existed prior to this endeavor.
Year: 2012