Coupling Technologies for Earth System Models Workshop report
Jeremy Walton, Met Office Hadley Centre and UKESM core group.
The fourth workshop on Coupling Technologies for Earth System Models was held at Princeton University on March 20-22 (see https://www.earthsystemcog.org/projects/cw2017/). Richard Hill and Jeremy Walton attended from the UKESM core group; our Met Office colleagues Jean-Christophe Rioual and Mike Hobson were also at the meeting.
Coupling is an important aspect of the construction of a climate model. It manages complexity by allowing the system to be separated into domain-specific components (atmosphere, ocean, land, ice, …) each of which can be developed separately or replaced by a component of similar functionality. Coupling enables collaboration between groups working in disparate scientific domains, and facilitates the efficient use of compute resources by allowing load balancing across the components.
The workshop was focussed on recent developments in methods and systems for coupling components together in climate models. Presentations covered a variety of topics, including enhancements in coupling systems such as OASIS (as used by UKESM) and ESMF, technologies like NUOPC for the interoperability of climate components and XIOS for I/O management in climate models, whole systems including CESM and EC-EARTH, progress in benchmarking coupling systems (on current and future HPC architectures), the development of enabling technologies such as for analysis such as visualization and work on geoscience ontologies. The application areas discussed in the meeting ranged from global climate prediction to hydrology and environmental modelling.
Mike Hobson described the latest results from the LFRic project, particularly the challenges of coupling the finite-element dynamical core to finite-difference physics. Jean-Christophe Rioual spoke about scalability issues for the current NEMO/CICE configuration, which could be addressed by decoupling CICE from NEMO. Richard Hill gave an update on the development of a mixed-resolution climate model (involving the coupling of a high-resolution atmosphere to low-resolution atmospheric chemistry). Finally I described work being done in the core group to incorporate ice sheets into UKESM; here, the adaptive-mesh ice sheet model must exchange data with the atmosphere, ocean and land components. Technical challenges include the wide difference in length and time scales between the ice sheet processes and the other components, and the moving physical boundaries between domains caused by the dynamic evolution of the ice sheet. This appears to be a topic of current interest since at least two other presentations at the workshop were devoted to the same subject.