S19: Exascale Computational Mechanics for Earth, Environmental and Sustainability Solutions
Co-chairs:
Andrew McBride* (University of Glasgow)
Todd A. Ehlers (University of Glasgow)
Christopher Jewell (Lancaster University)
Tobias Keller (University of Glasgow)
Michèle Weiland (EPCC)
Keywords: Exascale and GPU computing; environmental modelling; geophysical fluid and solid mechanics; Earth observation data; data assimilation and inverse problems; uncertainty quantification; digital twins.
ABSTRACT
Computational mechanics is central to understanding and predicting the behaviour of the natural environment. Simulations of geophysical flows, flooding, coastal and tidal hydrodynamics, ice and sediment transport, geohazards and solid-Earth deformation inform responses to climate change, the management of natural hazards and the transition to sustainable energy. Exascale computing, delivered increasingly through GPU-based architectures, offers the capacity to resolve these coupled, multiscale processes at unprecedented fidelity, but realising that potential demands new numerical methods and software, and new ways of combining models with observations. These challenges are the focus of the NERC-funded ExaGEO Doctoral Landscape Award (Exascale computing for Earth, Environmental and Sustainability Solutions), a partnership between the University of Glasgow, EPCC at the University of Edinburgh and Lancaster University, and they sit squarely within the interests of the UKACM community.
This mini-symposium invites contributions on computational methods for Earth and environmental systems from researchers within and beyond ExaGEO. Topics include, but are not limited to: numerical methods and software for environmental models on GPU-based and exascale architectures, including scalable solvers, performance portability and code modernisation; the analysis of large-scale Earth observation and simulation datasets; data assimilation, inverse problems, model calibration and other approaches that combine models with data to understand environmental change, including digital twins and machine-learning surrogates; and uncertainty quantification in environmental prediction. Applications of interest span geophysical fluid and solid mechanics, including hydrodynamics and flood forecasting, tidal energy, cryosphere dynamics, geohazards, and landscape and solid-Earth evolution. The mini-symposium aims to bring together the computational mechanics, Earth and environmental science, statistics and high-performance computing communities, and particularly welcomes contributions from doctoral students and early-career researchers.