The Science

Researchers developed and demonstrated a new approach for managing the rapidly increasing complexity of simulations of environmental systems in the critical zone near the land surface. The multiphysics Arcos framework combines modern software design principles in a novel way to create flexibly configured simulators, thus enabling significantly more complex and realistic simulations that combine many individual ecohydrological and biogeochemical processes.

The Impact

As simulations of environmental systems grow in complexity by incorporating more and more ecohydrological and biogeochemical process representations, adding new process understanding while ensuring that individual and coupled-process simulations are reliable has become increasingly difficult. A new multiphysics framework helps tame this runaway complexity, making process-rich simulations easier to develop, test, combine with data, and reconfigure for different numerical experiments. This approach to developing models provides a more natural way for scientists to collaborate on increasingly complex models and helps build confidence in the resulting simulations.

Summary

The Arcos system is based on two graph representations that interact to provide a flexible and extensible framework. The first graph is a process tree representation that defines the coupling among various environmental process representations denoted as process kernels (PKs). Two or more PKs are coupled together through multiprocess coordinators. The second graph defines how the mass and energy balances depend on primary variables (unknowns to be solved for) through a series of intermediate variables. Formal representation of these dependencies in a graph structure makes it easier to substitute new constitutive models and ensures that intermediate variables are always current and consistent among different PKs. Taken together, these two graphs make it possible to define which PKs are to be used and how they are to be coupled at run time. Such a flexibly configured and hierarchical structure is critical to systematically building up complexity supported by rigorous testing and evaluation against observations.

Principal Investigator

Ethan Coon
Los Alamos National Laboratory
ecoon@lanl.gov

Program Manager

Paul Bayer
U.S. Department of Energy, Biological and Environmental Research (SC-33)
Environmental System Science
paul.bayer@science.doe.gov

Funding

This research was supported by the Interoperable Design of Extreme-scale Application Software (IDEAS) project funded by the U.S. Department of Energy (DOE) Office of Science. This work was also supported by Los Alamos National Laboratory’s Laboratory Directed Research and Development Predicting Climate Impacts and Feedbacks in the Terrestrial Arctic project (LDRD201200068DR).

Related Links

• Reference link

References