[M-IS09] Interdisciplinary studies on pre-earthquake processes

[MIS09-17] Incorporating Non-Seismic Precursors into Earthquake Probabilistic Forecasting Model

*Yiqun Zhang^1,2, Jiancang Zhuang², Kaiguang Zhu¹, Peng Han³ (1.Jilin university, 2.The Institute of Statistical Mathematics, 3.Southern University of Science and Technology)

Keywords:Earthquake Probabilistic Forecasting Model, ULF magnetic anomaly, CO anomaly, temporal ETAS model

Increasing reports show that various non-seismic anomalies can be observed before strong earthquakes, such as changes in geomagnetic field or gas emission. Many of these anomalies have been statistically linked to earthquakes, suggesting their potential as precursors. However, additional forecasting information provided by non-seismic signals beyond clustering effect in seismicity, which is the most significant predictable component, still needs to be well evaluated. Traditionally, precursors are typically studied separately from seismicity research, despite their potential complementarity. In this study, we develop a probabilistic model by incorporating precursors into temporal Epidemic-Type Aftershock Sequence (ETAS) model, to evaluate earthquake forecasting potential of ultra-low frequency (ULF) magnetic anomaly and carbon monoxide (CO) anomaly. The model is applied to M 4.0+ earthquakes between 2001 and 2010 around the Kakioka (KAK) station, Japan. Results indicate that 15.2% of events originally attributed to the Poisson background in temporal ETAS can be predicted by the external excitation of precursors. The proposed model improves the probability gain by 3.8% compared to ETAS. Although individual precursors may have limited forecasting capability, these findings highlight the potential for enhancing short-term earthquake forecast performance by incorporating precursors. In particular, combining multiple precursors within the model further improves the probability gain of earthquake forecasts.