When an earthquake hits, its effects can vary wildly from neighborhood to neighborhood, shaking one area's buildings violently while just a short distance away, tremors hit with noticeably less force.

One reason for this difference is local subterranean variations known as “site effects." For instance, soft or loose soils, such as those found in many coastal areas, can significantly amplify earthquake shaking, causing much more damage to structures. 

Understanding site effects is critical for designing safe and resilient infrastructure, including buildings, bridges, and transportation systems. Yet modeling and designing for them is one of the most persistent challenges in earthquake engineering.

Adrian Rodriguez-Marek, professor of civil and environmental engineering, and a group of collaborators from the University of Canterbury in New Zealand recently won the outstanding paper award from the Earthquake Engineering Research Institute for research that tackles this problem.

The team developed a framework for evaluating how well different modeling approaches capture near-surface soil effects in earthquake simulations. The paper in Earthquake Spectra, “Methods to account for shallow site effects in hybrid broadband ground-motion simulations,” was recognized for research that meaningfully advances earthquake engineering practice and improves seismic risk assessment.

"The key to improving earthquake predictions is understanding how to make the best use of local data,” said Rodriguez-Marek. “When we know which methods work best, we can collect more meaningful information and make more accurate predictions."

Improving how models reflect earthquake behavior

The study examined computer models used to predict earthquake shaking, known as hybrid broadband ground-motion simulations. They combine physics-based models of large, slow-moving seismic waves with more data-driven or statistical methods that estimate the faster shaking that can cause significant damage. 

The researchers found that these different methods can produce meaningfully different predictions of how strongly the ground will shake, especially in areas with softer soils that amplify earthquake motion. The biggest differences occurred in the ranges of shaking that affect mid-rise buildings and other infrastructure. By comparing models with recordings from earthquakes in New Zealand, the team identified which approaches work best under different soil conditions and with different kinds of available data. 

“This approach allows earthquake simulations to be customized for specific site conditions without additional computational cost, improving both efficiency and accuracy,” said Felipe Kuncar, a postdoctoral fellow at the University of Canterbury in Christchurch, New Zealand, who worked on this project as a visiting scientist at Virginia Tech in 2023. 

Tradeoffs between simplicity, data, and accuracy

One of the study's key findings is that no single method performs best in every situation.

Simple engineering methods are often practical and useful when detailed soil information isn't available, but they may not accurately predict how a specific site will respond if the underground soil is deep or behaves in complex ways. On the other hand, more advanced computer models can provide a more accurate picture of how the ground is likely to respond during an earthquake, but they come with increased computational cost and a need for detailed geotechnical information about local soil conditions that is not always available to engineers, according to the study.

In short, whichever modeling approach engineers choose can change predictions of how the ground will shake during an earthquake, and in turn, influence how they estimate the seismic forces that buildings need to withstand.

“This research can improve how we predict earthquake shaking in a wide range of situations, from assessing the safety of critical infrastructure such as dams to creating the seismic hazard maps used to guide building design,” said Rodriguez-Marek.

Looking ahead

By systematically comparing these modeling approaches, the study provides guidance on when simpler methods are sufficient and when more detailed site-specific analysis is necessary. Ultimately, the research provides engineers with better tools for predicting earthquake impacts, supporting safer designs and stronger, more resilient structures.

"This study has the potential to change how engineers model earthquake shaking in seismically active regions,” said Rodriguez-Marek. “By improving our understanding of how local ground conditions influence shaking, we can make more informed decisions that lead to safer infrastructure and more resilient communities."

Original study: DOI 10.1177/87552930241301059

Contact:

Tags

Share this story