Recurrence analysis on slow slip events in Japan using renewal processes

Abstract SSEs are closely linked to megathrust earthquakes, as they occur near seismically locked zones and can perturb stress conditions along plate boundaries. Understanding their timing is, therefore, crucial for assessing seismic hazards. This study investigates the recurrence intervals of slow slip events (SSEs) at four regions in Japan: Kii peninsula, Yaeyama islands, Boso peninsula, and Bungo Channel. Using renewal processes based on the Brownian Passage Time (BPT) model and the Gamma model as well as the moment method, the study estimates the mean recurrence intervals and their coefficients of variation (the standard deviation divided by the mean) along with the estimation uncertainties in each region. Final estimated mean recurrence intervals are 0.96 years (Kii), 0.51 years (Yaeyama), 3.97 years (Boso), and 3.16 years (Bungo) for all methods. Coefficients of variation reflecting event periodicity in the four regions (0.42−0.46, 0.25−0.40, 0.44−0.54, 0.26−0.28) suggest that SSEs are quasi-periodic, being neither completely deterministic nor highly random. Compared to earthquakes with recurrent patterns of occurrence, which are typically less sensitive to small stress changes, SSEs generally exhibit higher variability in the recurrence, suggesting greater sensitivity to external perturbations. However, Bungo SSEs, with higher stress drops, exhibit lower variability, indicating more stable recurrence. The bootstrap analysis also confirmed regional characteristics. In Kii, SSEs occurring at the intermediate area, with the distance from the Nankai Trough axis to the fault center is approximately 25–36 km, show greater timing variability. In Yaeyama, recurrence appears linked to slip size, suggesting time-predictable behavior. In Boso, the 2011 Tohoku–Oki earthquake likely influenced SSE timing. In Bungo, the unique 2014 SSE, possibly affected by the Tohoku–Oki event, was highlighted. In addition, the study performs predictive model comparisons for the baseline renewal process and a retrospective predictability analysis for all the regions, and provides forecasts of the next SSE in the Kii region along with prediction intervals for its occurrence time.