Menu Close

New paper on empowering machine learning-based labquake forecasting

Sadegh Karimpouli published a paper in Journal of Geophysical Research – Machine Learning. This is a continuation of our work that started with study published in EPSL on labquake forecasting for rough faults using deep learning methods, followed up by the unsupervised classification of the system state presented in GJI. In JGR-ML Sadegh continues the story applying an extended set of physics-informed features that are now calculated in space and time. In this way, we transmit much more information to the machine learning algorithms, increasing the accuracy of labquake forecasting. In addition we find that separation of catalog into background and clustered seismicity and processing these subsets individually we again increase the forecasting accuracy. Have a read! The paper is open access!

Reference:

Karimpouli, S., G. Kwiatek, Y. Ben-Zion, P. Martínez-Garzón, G. Dresen, and M. Bohnhoff (2024). Empowering Machine Learning Forecasting of Labquake Using Event-Based Features and Clustering Characteristics, Journal of Geophysical Research: Machine Learning and Computation 1, no. 2, e2024JH000160, DOI: 10.1029/2024JH000160. [ Article Page ] [ Download open-access article ]
Posted in Publications