Prof. Dr. Cecilia Van Cauwenberghe, PhD, MSc, BS, MBA, is Research Director at Everest Group. Here, she explores seismic resilience in the age of artificial intelligence and advanced remote sensing, focusing on how science and technology transform earthquake preparedness
Introduction: From detection to prediction
Earthquakes remain among the most devastating and unpredictable natural disasters, causing catastrophic loss of life and infrastructure damage (Aksoy et al., 2024). However, recent breakthroughs in artificial intelligence (AI), machine learning (ML), remote sensing, and cost-benefit modeling are transforming our ability to detect, predict, and mitigate seismic hazards.
Historically, earthquake preparedness has relied on reactive detection systems and post-event damage assessments. However, cutting-edge research is shifting the paradigm toward proactive forecasting and real-time reconnaissance, enhancing both emergency response efficiency and long-term resilience planning.
Innovations such as the ShakeAlert® Earthquake Early Warning System, AI-driven forecasting models, transformer-based building damage assessments, and benefit-cost analyses (BCA) for seismic resilience mark a new era in earthquake science – one that is data-driven, predictive, and action-oriented.
AI-driven forecasting
Between 2019 and 2023, the ShakeAlert® Earthquake Early Warning System issued 95 alerts for earthquakes of 4.5 or higher, with 94 confirmed events (Lux et al., 2024). While this marks a high detection accuracy, seven false alerts and four missed earthquakes highlight network coverage and precision gaps, particularly in edge regions.
AI-driven approaches are being developed to move from detection to prediction to address these limitations. Anbazhagu et al. (2025) highlight how deep learning models, seismic anomaly detection, and neural networks can improve forecasting accuracy by analyzing large-scale geophysical datasets.
Similarly, Olaoluwa et al. (2024), as part of NASA’s Gateways to Blue Skies Competition, introduced an AI-driven earthquake forecasting system that significantly advances beyond ShakeAlert’s reactive model. Their system integrates: 1) Synthetic Aperture Radar (SAR) satellite imaging for large-scale seismic monitoring; 2) drone-mounted geophones to capture localized ground motion data; 3) Total Electron Content (TEC) perturbation analysis in the ionosphere to detect seismic precursors.
These ML models achieved 85–90% accuracy in differentiating seismic wave phases, while generative adversarial networks (GANs) enhanced prediction accuracy, even with limited historical data. According to the developers, by detecting earthquakes before seismic waves reach the surface, this AI-driven approach offers a transformative shift from reaction to prevention, allowing earlier evacuations and infrastructure protection measures.
Enhancing post-earthquake damage assessment with synergistic innovations
Once an earthquake strikes, rapid and precise damage assessment is critical for directing emergency response efforts. Traditionally, building damage evaluations have relied on manual inspections, which are time-consuming, hazardous, and inefficient for large-scale disasters.
Singh et al. (2024) developed a metadata-enriched transformer-based model to improve efficiency and accuracy. The researchers synergistically integrated high-resolution optical and SAR satellite imagery, ground motion intensity data, and soil property analysis. It automates damage classification, enhances the distinction between structural damage levels, and ensures explainability in AI decision-making.
Meanwhile, Giardina et al. (2024) demonstrated how integrating SAR data with ground-based surveys improves the identification of building collapse and earthquake-induced landslides. Their hybrid reconnaissance framework overcomes the limitations of optical satellite imaging (weather dependency) and manual assessments, enabling near real-time, high-resolution disaster mapping.
These AI-powered damage assessment systems provide a faster, more scalable, and precise approach to disaster response and urban resilience planning.
The economics of earthquake resilience: Evaluating cost vs. benefit
Zhang et al. (2024) conducted a comprehensive benefit-cost analysis (BCA) to evaluate the financial trade-offs of modern building codes, above-code structural designs, and seismic retrofitting.
Their findings highlight that: 1) modern building codes significantly reduce structural damage in high-risk zones, but their economic viability in moderate-risk areas remains uncertain; 2) above-code seismic designs provide greater long-term resilience yet require further validation of cost-effectiveness; 3) retrofitting older buildings is often cost-efficient but must be optimized for cost-performance balance.
Future research will explore BCA integration with environmental benefits, mainly how earthquake-resistant infrastructure reduces carbon footprints, and assess policy-driven incentives to encourage resilience investments.
Policy and research roadmap: NEHRP and NSHM
The National Earthquake Hazards Reduction Program (NEHRP) has guided U.S. seismic risk reduction since its inception, undergoing multiple reauthorizations (1990, 1997, 2000, 2004) (Zhu et al., 2024).
In May 2023, NEHRP released its Strategic Plan for 2022–2029, consolidating its objectives into four integration areas, 18 support tasks, and eight key investment priorities – defining the roadmap for seismic risk reduction in the coming decade.
Meanwhile, the 2023 U.S. National Seismic Hazard Model (NSHM) represents a landmark advancement in hazard assessment (Anderson et al., 2024). The latest update incorporates expanded fault databases, new geodetic deformation models, and improved earthquake occurrence rate calculations.
Developed using the best available science principles, the NSHM enhances seismic risk modeling and improves building codes, risk management strategies, and emergency planning frameworks.
A future defined by AI, remote sensing, and resilience
Emerging technologies can minimize uncertainties, improve forecasting capabilities, and accelerate post-disaster response. To fully realize this vision, continued investment in AI-powered earthquake forecasting, real-time reconnaissance technologies, and optimized benefit–cost strategies for resilience infrastructure is crucial. A global collaborative effort among governments, researchers, and technology leaders will be key in scaling these innovations for widespread adoption.
With machine learning-enhanced forecasting models, high-resolution remote sensing, and robust economic planning, seismic resilience is entering a new era that could redefine how we anticipate and mitigate earthquake disasters worldwide.
References
- Aksoy, Cevat Giray, et al. “Unearthing the impact of earthquakes: A review of economic and social consequences.” Journal of Policy Analysis and Management (2024).
- Anderson, John G., et al. “Recommendations on best available science for the United States National Seismic Hazard Model.” (2024).
- Anbazhagu, U. V., et al. “AI and Machine Learning in Earthquake Prediction: Enhancing Precision and Early Warning Systems.” Modern SuperHyperSoft Computing Trends in Science and Technology. IGI Global Scientific Publishing, 2025. 1-32.
- Giardina, Giorgia, et al. “Combining remote sensing techniques and field surveys for post-earthquake reconnaissance missions.” Bulletin of Earthquake Engineering 22.7 (2024): 3415-3439.
- Lin, Zhu, et al. “Introduction to the National Earthquake Hazards Reduction Program and the latest strategic plan.” Progress in Earthquake Sciences 54.2 (2024): 140-146.
- Lux, Angela I., et al. “Status and performance of the ShakeAlert earthquake early warning system: 2019–2023.” Bulletin of the Seismological Society of America 114.6 (2024): 3041-3062
- Olaoluwa, Adewoye, Ian Gabriel Mondares, and Alivia Ross. “Improving Earthquake Prediction with Artificial Intelligence and Machine Learning.” NASA’s Gateways to Blue Skies Competition Advancing Aviation for Natural Disasters, 2024.
- Plevris, Vagelis. “AI-driven innovations in earthquake risk mitigation: a future-focused perspective.” Geosciences 14.9 (2024): 244.
- Singh, Deepank, Vedhus Hoskere, and Pietro Milillo. “Multiclass PostEarthquake Building Assessment Integrating Optical and SAR Satellite Imagery, Ground Motion, and Soil Data with Transformers.” arXiv preprint arXiv:2412.04664 (2024).
- Zhang, Yating, et al. “Benefit–Cost Analysis for Earthquake-Resilient Building Design and Retrofit: State of the Art and Future Research Needs.” Natural hazards review 25.3 (2024): 03124001.
