Although earthquakes remain impossible to predict precisely, scientists are developing advanced computational models that significantly improve understanding of how seismic events interact with the Earth’s subsurface. A magnitude 7.0 earthquake that struck Alaska on December 6, 2025, highlights the importance of this work. While large events are relatively rare, seismic activity is constant: roughly 55 earthquakes occur worldwide each day, amounting to about 20,000 annually. The physical damage, economic costs, and human trauma caused by these events make improved risk assessment a critical scientific goal.
The central challenge in earthquake science is not identifying where faults exist, but understanding how seismic waves travel through complex underground structures. The Earth’s subsurface is not uniform; it consists of layers of solid rock, fractured zones, sand, clay, and other materials. Each of these affects how seismic waves slow down, speed up, or amplify as they move toward the surface. These variations largely determine why earthquakes of similar magnitude can cause dramatically different levels of damage in different locations.
To capture this complexity, researchers use a technique called Full Waveform Inversion (FWI). The process begins with an initial “best guess” of the subsurface structure in a region. Scientists then simulate synthetic earthquakes on a computer and calculate how seismic waves would propagate through that assumed underground model. These simulated waveforms are recorded at virtual seismograph locations, mirroring how real instruments collect data during actual earthquakes.
The key step is comparison. The synthetic waveforms are matched against real seismograms—graphical records of ground motion from observed seismic events. Differences between the simulated and real data reveal inaccuracies in the subsurface model. The model is then adjusted, and the simulation is run again. This iterative process continues until the simulated data closely matches real observations. Through many cycles of refinement, researchers build a detailed and physically realistic image of the subsurface, improving understanding of how earthquakes interact with specific geological settings.
However, this approach is computationally demanding. Traditional FWI requires thousands of simulations, each involving millions of parameters, and a single run can take hours on powerful computing clusters. This makes real-time monitoring or large-scale risk assessments impractical using conventional methods.
To overcome this limitation, Kathrin Smetana and an interdisciplinary team developed a reduced-order modeling approach. Their innovation dramatically shrinks the mathematical system that must be solved—by roughly a factor of 1,000—while preserving the accuracy of the physical predictions. Instead of recalculating every detail from scratch, the reduced model captures the most important patterns in wave behavior, allowing seismic simulations to run far faster across many parameter variations.
This efficiency transforms what is possible. While the model cannot predict when earthquakes will occur, it enables rapid testing of many subsurface scenarios, improving estimates of shaking intensity and damage potential. The same framework could also support faster tsunami simulations following offshore earthquakes, where even modest time savings can improve emergency response.
In essence, the model works by combining physics-based simulation, real seismic data, and mathematical reduction techniques to reveal the hidden structures that shape earthquake impacts. By making high-fidelity subsurface imaging computationally practical, this work represents a major step toward more resilient societies in earthquake-prone regions, even as earthquakes themselves remain fundamentally unpredictable.
https://scitechdaily.com/this-clever-math-trick-could-change-how-scientists-study-earthquakes