Earthquake ground-motion predictions have long relied on statistical models built from historical records, but those records are sparse in many regions and often lack data for the largest events. Seismologists are increasingly turning to physics-based simulations to fill those gaps, yet the approach has rarely made it into engineering applications or building codes. At the US National Conference on Earthquake Engineering in Portland, Oregon, researchers gathered to ask why.
The conference, hosted by the Earthquake Engineering Research Institute, included a dedicated workshop on validating and applying simulated ground motions. Researchers noted that current probabilistic models depend on curated datasets collected over roughly a century of earthquake recording, a period too short to capture the full range of possible magnitudes and shaking patterns. The data gap is especially stark for high-magnitude earthquakes at short distances from the fault, since only a handful of such events have been recorded since modern instruments were deployed.
Physics-based simulations can meaningfully augment those datasets and improve conventional ground-motion models, according to researchers presenting at the meeting. One example discussed was the Cascadia subduction zone, a fault running hundreds of kilometres along the Pacific Northwest coast of the US and southwestern Canada. The fault has not produced a significant earthquake in the modern era, but geological evidence shows several large events over thousands of years. A simulated magnitude 9 earthquake on that fault has been used to study how the built environment would be affected, including cascading impacts such as landslides and tsunamis.
Despite the promise, physics-based simulations remain largely absent from national seismic hazard models, and where they are used, the scale is negligible. Stanford University’s Greg Deierlein told the workshop that simulations are most useful when they provide different answers compared to conventional methods. The missing ingredient, he and others said, is trust: whether simulations can represent complex fault geometry and rupture mechanisms, how reliable the velocity structure models are, and whether simulations can reproduce empirical data.
Trust within the research and practitioner community is built through testing and validation, and the workshop included discussions on standardized validation protocols and open repositories of reproducible workflows. One key challenge is establishing that simulated ground shaking scales across parameters in patterns consistent with observed recordings. Jonathan Stewart of the University of California Los Angeles noted that while trust in the absolute amplitude from physics-based simulations is low, the way simulated ground motion scales across various parameters is what holds value.
The conference featured at least one session each day on developing, validating, and applying physics-based simulations. One presentation used simulated ground motions to reveal damage patterns across a building portfolio that conventional models cannot capture, suggesting that different models can produce very different views of risk. That divergence creates challenges for insurance companies trying to price risk accurately and for planning future infrastructure in earthquake-prone regions.
Researchers left the meeting with more questions than answers, but the energy around physics-based simulations is growing as computational capacity and geophysical understanding of earthquake cycles improve. The path forward, they agreed, depends on building the validation frameworks that will let engineers and policymakers trust what the simulations show.
