日本地球惑星科学連合2022年大会

講演情報

[E] ポスター発表

セッション記号 S (固体地球科学) » S-CG 固体地球科学複合領域・一般

[S-CG44] Science of slow-to-fast earthquakes

2022年6月3日(金) 11:00 〜 13:00 オンラインポスターZoom会場 (23) (Ch.23)

コンビーナ:加藤 愛太郎(東京大学地震研究所)、コンビーナ:田中 愛幸(東京大学理学系研究科)、山口 飛鳥(東京大学大気海洋研究所)、コンビーナ:波多野 恭弘(大阪大学理学研究科)、座長:永冶 方敬(東京大学大学院理学系研究科)、Anca Opris(Research and Development Center for Earthquake and Tsunami Forecasting)

11:00 〜 13:00

[SCG44-P28] Why do Gutenberg-Richter b-values decrease with time prior to large earthquakes? Insights from rate-and-state earthquake cycle simulations with frictional heterogeneities

*伊東 良1金子 善宏1 (1.京都大学)


キーワード:b値、地震サイクルシミュレーション

Understanding the mechanism of foreshock generation is an important issue for earthquake forecasting as well as our fundamental understanding of earthquake physics. Observations of seismicity prior to large earthquakes show that a slope of Gutenberg-Richter magnitude-frequency relations, known as a b-value, often decreases with time before the occurrence of a large earthquake. For example, Nanjo et al. (2012) reported that b-values in the hypocentral region of the 2011 Mw9.0 Tohoku-Oki earthquake had decreased prior to the mainshock. Yet, interpreting observed temporal changes of b-values remains challenging. Here we use numerical simulations of earthquake cycles with frictional heterogeneities and attempt to simulate the temporal variations of b-value. We first identify a parameter regime in which the model gives rise to an accelerating foreshock behavior prior to the mainshock. We then focus on the spatiotemporal pattern of foreshocks. We find that, like in observations, a complex pattern of migrating foreshocks can be seen in this model. Next, we analyze the resulting foreshock statistics and find that b-values decrease with time prior to the mainshock. In this case, afterslip in the creeping (or velocity-strengthening) patches due to numerous foreshocks makes these patches more susceptible to future co-seismic slip, increasing the likelihood of large ruptures and resulting in smaller b-values with time. Interestingly, our model also shows that the average shear stress over the entire fault slightly decreases with time. From this result, we conclude that the cause of a temporal decrease in b-values would not be due to increased average shear stress, but instead, the afterslip of foreshocks that makes the stably-slipping regions more susceptible to co-seismic slip, leading to larger foreshocks.