Lay Summary

Accurately predicting wildfire behavior is vital for saving lives and protecting ecosystems, but realistic fire simulations require vast computational resources. This paper examines the deployment of advanced wildfire modeling tools on European High-Performance Computing (EuroHPC) supercomputers. The authors discuss key technical challenges, including processing massive spatial datasets, handling real-time weather uncertainty, and scaling code across thousands of processor cores. They highlight how leveraging supercomputing infrastructure allows emergency responders to run high-resolution, faster-than-real-time simulations for improved disaster preparedness and active fire control. (83 words)

Full Abstract

High-resolution prediction of wildfire propagation demands massive computational capabilities to process complex terrain, variable weather patterns, and fuel dynamics in real time. This paper explores the adaptation and execution of wildfire simulation models on EuroHPC petascale and exascale supercomputing infrastructure. We analyze computational bottlenecks associated with parallel data input/output, spatial domain decomposition, and uncertainty quantification. Furthermore, we demonstrate how exascale-ready parallelization enables faster-than-real-time wildfire predictions, offering critical decision support for emergency responders and land management agencies during severe fire events.

Metadata

• Publication Date: 2024
• Author:  David Caballero, Leydi Laura Salazar, Ángela Rivera, Luis Torres
• Publication: Proceedings of EuroHPC User Forum / Environmental Modelling Proceedings