[S09] AM-2

[S09-22] Japan M5 or over forecast for 2025 based on rate/state stress transfer projects strong influences of the 2011 M 9.0, 2024 M 7.5 and M 7.1 events

*Shinji TODA¹, Ross S Stein² (1. International Research Institute of Disaster Science, Tohoku University, 2. Temblor Inc.)

We contend that there is sufficient evidence that earthquakes promote and inhibit failure on nearby faults, a process governed principally by, whose effect decays with time. We reproduce the time decay and non-linearity of the response of seismicity to stress changes by rate/state friction (Dieterich 1994). Instead of calculating the stress on mapped faults, here we calculate the stress imparted to background focal mechanisms, because wherever there is a focal mechanism, a fault must exist and has indeed slipped. In rate/state friction, stress increases amplify the background seismicity rate, and stress decreases suppress it, and so the background seismicity model is crucial to the calculation. We have experimented with both the GEAR1 earthquake rate model (Bird et al 2015) and the observed JMA background rate for a sufficiently long period above the magnitude of completeness. We use the 200 M≥6.0 earthquakes since 1995 as the sources of stress, represented by published or USGS finite fault models, or point sources if the events lack finite fault models. We calculate annual retrospective forecasts of M≥5 earthquakes for the years 1995-2023 in and around four main Japan’s islands ranging of longitudes 128°-148° and latitudes 30° - 46°. To score the skill of these models, we devised a new metric which measures the difference between the observed and forecast cumulative number of events. In more than half of the testing years, the rate/state Coulomb model for a given year using JMA background outperforms the JMA background itself (the average behavior), and it also outperforms the GEAR1 model background, or the rate/state Coulomb model that uses GEAR1 to amplify the stress changes. This probably arises because GEAR1 includes ~100 Tohoku aftershocks, whereas the JMA background does not. We then present prospective forecast for 1 Jan – 31 Dec 2025.