摘要：This paper selects fault source models of typicalearthquakes across the globe and uses a volume extending 100 km horizontallyfrom each mainshock rupture plane and 50 km vertically as the primary areaof earthquake influence for calculation and analysis. A deep neural networkis constructed to model the relationship between elastic stress tensorcomponents and aftershock state at multiple timescales, and the model isevaluated. Finally, based on the aftershock hysteresis model, the aftershockhysteresis effect of the Wenchuan earthquake in 2008 and Tohoku earthquakein 2011 is analyzed, and the aftershock hysteresis effect at differentdepths is compared and analyzed. The correlation between the aftershockhysteresis effect and the Omori formula is also discussed and analyzed. Theconstructed aftershock hysteresis model has a good fit to the data and canpredict the aftershock pattern at multiple timescales after a largeearthquake. Compared with the traditional aftershock spatial analysismethod, the model is more effective and fully considers the distribution ofactual faults, instead of treating the earthquake as a point source. Theexpansion rate of the aftershock pattern is negatively correlated with time,and the aftershock patterns at all timescales are roughly similar andanisotropic.
