[M-IS04] Interdisciplinary studies on pre-earthquake processes

[MIS04-01] Generative Stochastic Earthquake Simulations for Nowcasting with AI: A Catalog is All You Need

*John B Rundle¹ (1.University of California Davis, USA)

Keywords:Earthquake Simulations, Generative AI, Earthquake Nowcasting, Machine Learning

Earthquake nowcasting has been proposed as a means of tracking the change in large earthquake potential in a seismically active area. The method was developed using observable seismic data, in which probabilities of future large earthquakes can be computed using Receiver Operating Characteristic (ROC) methods. Furthermore, analysis of the Shannon information content of the earthquake catalogs has been used to show that there is information contained in the catalogs, and that it can vary in time. So an important question remains, where does the information originate? In this paper, we examine this question using statistical simulations of earthquake catalogs computed using a new stochastic model "ERAMSS", Earthquake Rescaled Aftershock Model for Stochastic Simulations. The current state of the art has been Epidemic Type Aftershock Sequence (ETAS) simulations, which has between 6-10 parameters that must be set by fitting the catalog using a maximum likelihood or other method. Fitting all these parameters, with their inevitable cross-correlations, can be challenging. By contrast, the ERAMSS model has but 3 parameters to describe the triggered seismicity, a rescaling parameter for time variation in the Omori-Utsu law, and two spatial parameters for the latitude-longitude migration of the aftershock patterns. All parameters can be easily be determined by analyzing the given data catalog. We further note that the space and time parameters are independent, and have no cross-correlation dependencies. Using this model, we test the hypothesis that the information in the catalogs originates from aftershock clustering. We find that significant information in the catalogs arises from the non-Poisson aftershock clustering, implying that the common practice of de-clustering catalogs may remove information that would otherwise be useful in forecasting and nowcasting. We also show that the ERAMSS nowcasting method provides similar results with the the ETAS models as it does with observed seismicity.