EUVIMEDEUVIMED entdecken
Crossref

Toward Earthquake Prediction: Combining Physics-Based Models with Machine Learning

Swagatika Menon

American Journal for Young Scientists · 14.07.2026

DOI: 10.67194/ajfys.v2i3.002 ↗

Kurz gefasst

Fachartikel in der Zeitschrift American Journal for Young Scientists, online erschienen am 14.07.2026. Verfasst von Swagatika Menon.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Earthquake prediction is difficult because stress in the Earth builds up and releases in complex ways. This work combines a physics-based model with machine learning to study how earthquakes can be predicted. The Olami–Feder–Christensen (OFC) model is used to simulate how stress changes over time on a grid. The model is adjusted so that it follows known earthquake patterns like the Gutenberg–Richter law and Omori’s law. The simulated stress fields are then used to train a convolutional neural network (CNN) to predict whether a large event will occur soon. The model achieves a ROC-AUC of 0.67 and a PR-AUC of 0.90, showing good performance in identifying rare large events. Visualization of the stress fields suggests that the model learns useful patterns, especially regions where stress is concentrated. This approach shows that combining physics-based simulations with machine learning can help improve understanding of earthquake behavior and may support better early warning systems in the future.

Zitieren

APA

Menon, S. (2026). Toward Earthquake Prediction: Combining Physics-Based Models with Machine Learning. American Journal for Young Scientists. https://doi.org/10.67194/ajfys.v2i3.002

Vancouver

Menon S. Toward Earthquake Prediction: Combining Physics-Based Models with Machine Learning. American Journal for Young Scientists. 2026. doi:10.67194/ajfys.v2i3.002

BibTeX
@article{Menon2026,
  author = {Menon, Swagatika},
  title = {Toward Earthquake Prediction: Combining Physics-Based Models with Machine Learning},
  journal = {American Journal for Young Scientists},
  year = {2026},
  doi = {10.67194/ajfys.v2i3.002}
}
Kostenlos starten →