Re-Imagining Computation and Storage Resources in Climate- and Weather-dedicated Cyberinfrastructures

The use of climate models is extremely important to understand the future of our planet and the local impacts of a changing climate. These models are however considerably expensive, as they require weeks or even months to run and a large amount of storage space. This research project develops new software that is able to save a considerable amount of computational time and space using the latest advances in applied mathematics, statistics and machine learning as well as the most advanced computational resources available. The software is applied to wind behavior to better understand the risk of wildfire spread.

Climate and weather models are fundamental tools to understand the future state of the Earth's atmosphere, which will affect the human habitability of our planet at all scales (urban to global). Model-based environmental science is hampered by its ever-increasing computational and storage cost, and even research institutions with the most advanced cyberinfrastructure are able to perform only a limited number of simulations, thus leading to an inadequate and inaccurate uncertainty quantification.

The main research goal of this proposal is to propose key methodological developments in order to aid computation (with Stochastic Partial Differential Equations) and storage (with deep reservoir computing) in advanced climate- and weather-dedicated cyberinfrastructures. These methods will be then implemented on a cyberinfrastructure for modeling atmospheric flows in complex terrain.

This project is funded by the National Science Foundation's National Discovery Cloud for Climate initiative through the Office of Advanced Cyberinfrastructure.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.