Tsunami Forecasting

A groundbreaking approach to tsunami forecasting is being developed by scientists at Lawrence Livermore National Laboratory, leveraging advanced computational techniques to enhance prediction accuracy and speed. The research focuses on the Cascadia subduction zone, where tectonic plates can trigger massive earthquakes and subsequent tsunamis. Traditional methods of tsunami prediction have struggled with a trade-off between speed and accuracy, often resulting in unreliable forecasts that either arrive too late or trigger false alarms.

The team, led by computational mathematician Sanyo Kolev, has created a system that utilizes a digital twin of the seafloor and a powerful supercomputer, El Capitan, to run complex physics-based simulations. This innovative approach allows for the bulk of the computational work to be completed in advance, enabling real-time predictions within seconds of an earthquake occurring. By integrating data from proposed ocean floor sensors, the system can reconstruct the seafloor's movements and provide timely tsunami forecasts with high accuracy.

The project not only demonstrates the potential for rapid and reliable tsunami predictions but also showcases the capabilities of high-performance computing in addressing real-world challenges. The successful simulation, which involved over 55 trillion degrees of freedom, has garnered recognition, winning the prestigious 2025 Gordon Bell Prize. This advancement signifies a shift in scientific computing, with implications for various fields, including national security and environmental monitoring, as researchers continue to explore complex problems more efficiently.

Tsunami warning systems are either fast or accurate. LLNL is demonstrating that they may someday be both.