by Theresa Pepin
Even in our home kitchens—where we keep a small chunk of the Ice Age in a box—we all know it is not a good thing when the refrigerator’s freezer thaws, and our frozen items melt. The food spoils.
How much greater the consequences for entire Arctic landscapes where thawing of permanently frozen ground (permafrost) is occurring on a massive scale due to climate change!
The collaboration led by Ed Perfect and ORNL’s Richard Mills pairs a profound knowledge of the physical processes involved in ground freezing and thawing with massively parallel computation on leadership-class supercomputers to model the impacts of global warming. Just as it sounds, parallel computing treats many calculations simultaneously, dividing large problems into smaller ones, which are then solved concurrently.
Accurately representing and modeling subsurface processes is a daunting task because of the wide range of spatial scales involved—from tiny pores to lakes—and the wide range of time scales involved—from fractions of a second to millions of years. Notwithstanding its difficulty, the work is essential to assessing the effects of global warming on permafrost degradation in sensitive Arctic regions and the broader implications for carbon-climate feedback cycles.
Upscaling work is particularly relevant because the partner LDRD project, directed by Richard Mills, is targeted by the ORNL-led Next Generation Ecosystem Experiments (NGEE) Arctic project, in which an iterative (repetitive sequence of operations), upscaling/downscaling approach is employed to determine how fine-scale changes in hydrology and biogeochemistry can be represented in global climate-scale simulations. Mills is particularly impressed by the upscaling software utility developed by the JDRD team in its first year for forward prediction of selected ground freezing/thawing scenarios.
In year two, the JDRD team, including post-doc Chu Lin Cheng and an undergraduate student, will continue consultation on the physics of ground freezing/thawing processes, especially the phenomenon of ice segregation, and conduct additional parameter sensitivity testing at different spatial scales running the massively parallel 3-D simulator PFLOTRAN. Based on those results, the team will modify the PFLOTRAN code and compare its predictions with an independent model (MarsFlo) for selected climate change scenarios at various scales.
For too long, the characterization of hydrologic processes in current-generation land surface models (LSM’s) has suffered from over-simplification or complete omission of physical processes. Typically they feature one-dimensional representation of subsurface flow and heat transport, a unidirectional flow from surface to subsurface, and no freeze-thaw dynamics. Perfect and Mills are substantially improving the state-of-the-art by integrating detailed, high-resolution, surface-to-subsurface thermal, hydrologic, and biogeochemical reaction models within comprehensive land models.
The parallel performance engineering that has gone into PFLOTRAN, combined with the leadership-class supercomputers available at ORNL and elsewhere, will make possible a new Community Land Model (CLM)-PFLOTRAN code for undertaking novel, ambitious studies and regional-scale simulations.
Coupled simulation of hydrologic processes and terrestrial ecosystem and climate feedbacks: inclusion of soil freezing/thawing and upscaling modules in PFLOTRAN (year2)
Ed Perfect, UT Earth and Planetary Sciences Department
Coupled simulation of surface-subsurface hydrologic processes and ecosystem and climate feedbacks: from Arctic landscapes to the Continental U.S.
Richard Mills, ORNL Environmental Sciences Division