High-Resolution GPU-Enabled SCREAM RRM Simulations for Extreme Weather and Climate Events

PI Brandi Gamelin, Argonne National Laboratory
Co-PI Dimitrios K. Fytanidis, Argonne National Laboratory
Gökhan Sever, Argonne National Laboratory
Vishwas Rao, Argonne National Laboratory
Jiali Wang, Argonne National Laboratory
Danqing Wu, Argonne National Laboratory
Gamelin INCITE 2025

Dual-Multiple-Regional Refined Mesh (RRM) configuration using a GPU-enabled version of the Simple Cloud Resolving E3SM Atmosphere Model Version 1 (SCREAMv1). This configuration includes three distinct horizontal resolutions: Background (0.25⁰), Pacific Ocean (12 km), and U.S. Islands and Western U.S. (3 km). Image: Brandi Gamelin, Argonne National Laboratory

Project Description

The intensity and frequency of extreme events driven by Pacific Ocean variability has increased over time. Modeling the effects of Pacific Ocean surface temperature variability, coupled atmospheric dynamics, and long-term changes is crucial for understanding their influences on future extreme weather events in the United States, including the Pacific U.S. islands of Hawaii, Guam, Northern Mariana, and American Samoa. The insights gained will be critical for enhancing the resilience of energy systems, infrastructure, and communities against the impacts of extreme weather events. 

This INCITE project will produce high-resolution datasets for regions extending from the U.S. and across the Pacific Ocean using DOE supercomputing resources to investigate the high impact of extreme weather and climate events extending into mid-century (2015-2055). The team will deploy the Simple Cloud Resolving E3SM Atmosphere Model (SCREAM) on DOE supercomputers over a global computational domain with regionally refined areas across the Pacific Ocean, the contiguous U.S. (CONUS), and particularly the U.S. islands, which are sorely underrepresented and lacking high-resolution data for future projections. This work will advance the availability of actionable data for the analysis of extreme weather events in the U.S. and ensure that the datasets generated by this work are publicly available. This unique set of simulations will also foster collaborative opportunities with other research organizations, and the generated dataset is expected to be valuable for numerous projects funded by DOE and other agencies—particularly for projects that are interested in risk, reliability, and resilience studies to inform infrastructure planning.

Project Type
Allocations