The Environmental System Science Community considers questions such as the prediction of water availability and quality from watersheds under prolonged drought and warming temperatures. Integrated multidisciplinary projects in the DOE Office of Biological and Environmental Research (BER) are developing mechanistic process-rich models of nutrient and carbon cycling and are collecting field and laboratory data to help inform and constrain these models.
These science questions raise many software challenges. While legacy codes must be refactored to enable sharing of mechanistic modeling capabilities and coupling to describe more complex systems, most codes do not have comprehensive automated tests that enable refactoring with confidence. Most existing multiphysics frameworks have limited flexibility and extensibility, as these have been based on a priori static enumeration of a few coupling scenarios. In contrast, models of terrestrial ecosystems are increasingly process-rich, requiring the iterative addition and removal of processes. To this end, teams are exploring agile methodologies to support common interface designs, runtime-configurable multiphysics and multiscale frameworks, model-data integration workflows, and the componentization of legacy codes.