Environmental Remediation and Water Resources

Active

IDEAS: Computational Challenges in Building Virtual Terrestrial Ecosystems

DOE-SC-Biological and Environmental Research

East River Catchment, Gunnison County, Colorado.
[Image courtesy Roy Kaltschmidt,
LBNL]

A number of flagship programs within DOE’s Office of Biological and Environmental Research (BER), including EESA’s SFA 2.0 Genomes-to-Watershed and NGEE programs, have the goal of developing a predictive understanding of the complex ecosystems under study in each of those programs. To achieve the desired level of predictive understanding, a new generation of multiscale, multiphysics models is needed for terrestrial systems—models that incorporate process couplings and feedbacks between various “pools” (i.e., vegetation, soils, subsurface aquifers, and surface waters) across wide ranges of spatial and temporal scales.

This is a formidable task. Terrestrial systems are inherently multiscale, involve more processes than traditional multiphysics applications, and have significant uncertainty in process representation and coupling at different scales. Moreover, the scientific software community is facing the confluence of disruptive changes in computing architectures and new opportunities for greatly improved simulation capabilities. New architectures, while demanding fundamental algorithm and software refactoring, are at the same time enabling new multiscale and multiphysics modeling, simulation, and analysis.

Fig. 12. Schematic Showing Integrated Software Ecosystem Needed to Realize Potential of Virtual Systems. Here, as envisioned in Chapter 3 (p. 11), productivity of the modern scientific workflow (center ring) is enhanced because the critical phases of model development, simulation, and analysis leverage expertise and capability from the interdisciplinary community. Model development leverages interoperable components generated by multiple projects and contributes new model components to this collection within a flexible framework. Similarly, significant gains in the efficiency of the analysis phase, which includes sensitivity analysis (SA), uncertainty quantification (UQ) and parameter estimation (PE) are realized through more flexible and modular designs that enable efficient collaboration between the computational science and domain science communities. [Image courtesy David Moulton, Los Alamos National Laboratory]

EESA’s Genomes-to-Watershed and NGEE-Arctic projects seek to take advantage of these new scientific software capabilities by incorporating a recently initiated DOE project, co-funded by the Advanced Scientific Computing Research (ASCR) and the Office of Biological and Environmental Research (BER) within DOE’s Office of Science, entitled Interoperable Design of Extreme-Scale Application Software (IDEAS). This project pursues the development and demonstration of new approaches for producing, using, and supporting scientific software. It will establish methodologies and tools that facilitate delivery of software as reusable, interoperable components. Software life-cycle models will be developed that are both flexible and rigorous.

Specifically for the Genomes-to-Watershed Project, Phase I of which is an intensive study of the Colorado River watershed near Rifle, Colorado, IDEAS will provide fundamental simulation of biogeochemical cycling within the East River Watershed. The objective of this work, over the next three years, will be a better understanding of aquifer redox status and climate impacts on watershed carbon and nitrogen cycling—through higher fidelity, multiscale models simulated at high spatial resolution.

Fig. 6. High-Resolution Digital Elevation Maps. Shown at similar resolution, a typical analysis might begin with (a) highresolution topography from the East River Catchment, Gunnison County, Colorado, which is used to develop coupled surfacesubsurface hydrological flow models. These flow models then are expanded by incorporating (b) surface hydrology (rock type) and (c) land cover (vegetation) distributions, which then interact with the atmosphere. [Images courtesy Christine Pribulick and Reed Maxwell, Colorado School of Mines]

Specifically for NGEE-Arctic, IDEAS will provide the foundation for multiscale, multiphysics simulations of warming tundra in the Barrow, Alaska, region. The objective of this work, over the next three years, will be to determine (through high-resolution simulation of domains extending over 10 km) how dynamic microtopography caused by thawing permafrost alters the hydrologic and carbon cycles of Arctic lowland tundra.

With IDEAS, BER envisions a software ecosystem of interoperable components to increase both software development and scientific productivity across its portfolio of projects that depend on modeling.