Extreme-scale modeling - Understanding ecohydrologic dynamics under climate change

Praveen Kumar, University of Illinois at Urbana-Champaign

Usage Details

The elevated concentration of atmospheric CO₂ increases the ratio of carbon fixation to water loss of plants or water use efficiency. This shift in ecosystem functioning is central to understanding the cycles of water, energy and carbon under climate change. However, the magnitude of the effects on ecohydrologic dynamics, such as soil moisture content and surface runoff controlled by microtopographic variability, has been notoriously difficult to characterize over large areas. Part of the problem is that the underlying vegetation acclamatory processes that affect plant photosynthesis and transpiration under elevated CO₂ conditions are not captured in current multi-dimensional ecohydrologic models. In addition, large-scale simulations of ecohydrologic processes at the emerging Lidar-data resolution are numerically expensive due to the density of the computational grid and the iterative nature of the algorithms for solving nonlinearity. The development of high-performance computing, efficient and robust predictive models for the mechanism of vegetation acclimation and it links with ecohydrologic dynamics will greatly aid in understanding of how will climate change affect ecohydrologic systems. In this study, we develop a parallel, scalable and hybrid CPU-GPU modeling framework that links a vertically resolved model of canopy-root-soil biophysical processes with a distributed physically-based integrated surface—sub-surface flow model to investigate the impacts of climate change on large-scale ecohydrological dynamics. In this coupled model, while canopy processes in the aboveground systems are simulated in parallel in CPU using MPI, hydrologic processes on the land surface and in the belowground systems are simulated using GPU parallel computing.