Scientific Software Ecosystem. The IDEAS project is advancing a scientific software ecosystem comprising BER application codes that can increasingly share components (blue disks) either directly or through common interface libraries (orange boxes). Less mature components and those using simplified representations are shown in gray. The xSDK currently includes scientific libraries, Alquimia (biogeochemistry interface library), and PFLOTRAN (subsurface processes, biogeochemistry engine). The land model interface library (light orange) is targeted for development and later inclusion. [Courtesy Pacific Northwest National Laboratory]
Advancing software productivity to accelerate scientific discovery and understanding in environmental systems
The need to understand and predict how critical watersheds and terrestrial ecosystems function under natural and anthropogenic perturbations is leading to ever greater challenges in multiscale and multiphysics modeling. Moreover, disruptive changes in computer architectures are creating significant uncertainty in programming models. This confluence of interdisciplinary challenges is driving the Interoperable Design of Extreme-scale Application Software (IDEAS) project to explore new, community-based approaches to software development. Specifically, IDEAS seeks to enhance scientific productivity through an agile approach centered on adapting modern software engineering tools, practices, and processes to build a flexible scientific software ecosystem. Multiscale and multiphysics applications through the Department of Energy’s (DOE) Office of Biological and Environmental Research (BER) offer the ideal balance of complexity and maturity to define use cases that can both guide and test the efficacy of this approach.
The initial phase of the IDEAS project, IDEAS-Classic, was jointly funded by BER and DOE’s Office of Advanced Scientific Computing Research through DOE’s Office of Science. It established an interdisciplinary, community-oriented approach as the foundation for current and future projects. In particular, the methodologies focus area advances software development tools and practices to aid library and application developers. The resulting advances in library quality, interoperability, and performance are captured in the extreme-scale software development kit (xSDK). Completing the collaborative cycle, BER use cases motivate the agile methodologies, test interoperability, and contribute domain-specific components to the xSDK.
Each use case addresses important scientific challenges through advances in software development methodologies and engagement in the growing community-driven software ecosystem.
Perturbations of the Upper Colorado River System. Biogeochemistry plays a critical role in watershed function, driving the need for integrated hydrology models (surface-subsurface) that include tracer and reactive transport. In collaboration with the Watershed Function Scientific Focus Area (SFA) project at Lawrence Berkeley National Laboratory (LBNL), IDEAS is developing this capability through advances in multiphysics frameworks and the Alquimia interface to biogeochemical engines. In addition, community benchmark tests are being developed to compare different meshing strategies for multiscale models of the East River watershed.
Arctic Thermal Hydrology. Warming trends in the Arctic and resultant thawing of frozen soil may cause significant impacts on Arctic hydrology, built infrastructure, and the carbon cycle. Field and laboratory observations and models used to help understand the observed processes focus on local scales, but the consequences of these processes must be evaluated at regional and global scales. In collaboration with BER’s Next-Generation Ecosystem Experiments–Arctic project, a mixed-dimensional regional model representing surface-subsurface thermal hydrology and reusing local-scale model components has been developed through the use of flexible, multiphysics frameworks.
Hydrologic, Land Surface, and Atmospheric Process Coupling. Feedbacks among groundwater dynamics, overland (surface) flow, land surface and vegetation processes, and atmospheric boundary layer development significantly affect system function at local and regional scales. To untangle these feedbacks, this use case is exploring the computational challenges and scientific potential of high-resolution, integrated hydrologic models at regional and continental scales. New agile methodologies and testing were used to accelerate development of the first integrated hydrologic model over the continental United States that directly simulates physically based lateral groundwater flow.
Cross-Cutting Activities. This focus targets broadly applicable topics that may fall outside the current use cases and ensures that new methodologies and capabilities are transferred efficiently across projects. Two examples are Alquimia, an exemplar for best practices provided by the IDEAS methodologies team, and a collaboration with the River Corridor and Watershed Hydrobiogeochemistry SFA project at Pacific Northwest National Laboratory to explore hybrid, multiscale modeling approaches for the groundwater–surface water interaction zone.
The IDEAS project has made significant progress in meeting its objectives for developing a flexible, scientific software ecosystem to accelerate scientific discovery and understanding in environmental systems. This progress includes completion of open-source licensing and open access to source code repositories. In addition, the code teams are adopting agile methodologies and developer workflows, improving sustainability while enhancing capabilities. Community contributions include (1) leadership in model intercomparison studies in both integrated hydrology and reactive transport; and (2) helping establish the Environmental System Science (ESS) Cyberinfrastructure (CI) working groups, which are guiding a phased approach to community-driven CI development in the ESS Data Infrastructure for a Virtual Ecosystem project.
Building on this success, the IDEAS family of synergistic projects now includes three projects. Within DOE’s Exascale Computing Project (ECP) the IDEAS-ECP project continues development of agile methodologies with ECP teams, and xSDK-4-ECP supports continued xSDK growth. The third project, IDEAS-Watersheds, is developing the next phase of BER use cases to drive advances in understanding watershed function.
November 2019 [PDF]