Abstract:

Of all of the short-term and imminent precursors, foreshocks are recognized as one of the most effective indicators for earthquake prediction. This paper studies the statistical characteristics of foreshock activity of 29 global great shallow-focus earthquakes with M_W≥8.0 from 1976 to 2017. Among these earthquakes, there are 23 thrusting ones.
Results show that there are 8 great earthquakes with foreshocks, accounting for 27.6% of the total. In addition, the 8 great earthquakes with foreshocks are all thrusting events, accounting for 34.8% of the total thrusting earthquakes. The maximum magnitude of foreshock is greater than 5.0. The epicenters of foreshock sequences are densely concentrated in space and near the main shock. The epicenter distance between the largest foreshock and the main shock is 10~53km, the magnitude difference ΔM between the maximum foreshock and the main shock ranges from 1.1 to 2.8, and the origin time difference Δt(the interval between the origin time of the largest foreshock and that of the main shock)is from 2 hours to 15 days for 87.5% of the earthquakes except the Kuril Islands M_W8.3 earthquake, which occurred on November 15, 2006, while its maximum foreshock of M_W6.6 occurred 45 days before the M_W8.3 main shock.
Compared with the background seismicity, the foreshock sequence has the characteristic of high frequency. Of the eight earthquakes, the seismicity of five foreshock sequences accelerated within 15 days before the main shock, for the other three the seismicity began to increase during 35~45 days prior to the main shock, and the frequency increased again one day and six days before the main shock.
The focal mechanisms of foreshocks are similar to that of the main shock. While the focal mechanism of aftershock sequences is complicated and diverse. Such consistency of focal mechanisms with main shocks did not exist in aftershock sequences.
The parameters, α-value, p-value and b-value of foreshocks and aftershocks were calculated by the ETAS(Epidemic Type Aftershock Sequence)model, and b-value was got by the maximum likelihood method. The sample size N and the lowest calculated magnitude M_j can affect the calculation results of earthquake sequence parameters. Considering the reliability of results, it is assumed in this paper that when the number of foreshocks with M_j≥M_c(M_c indicates the minimum magnitude of completeness)is more than or equal to 30, the parameters, α-value, P-value and b-value, of the foreshock and aftershock sequences can be calculated. And the lowest calculated magnitude M_j of foreshocks and aftershocks is taken as the same or very close. Among the eight foreshock-mainshock-aftershock sequences, four can be compared to calculate the parameters of foreshocks and aftershocks. It should be noted that we take the aftershock activity within one month after the main earthquake as the aftershock sequence.
The calculated foreshock and aftershock sequence parameters show that there are no common features between the foreshock and aftershock for the α-value, meaning the ability to generate higher magnitude aftershocks, and the p-value, indicating the decay of sequences. While the b-value, indicating the stress state, shows a distinct characteristic that the b-value of foreshocks is obviously lower than that of aftershocks. Compared with regional background b-value, the b-value of foreshocks is 10%to 24%lower than the regional background b-value. And the b-value difference between foreshocks and regional seismicity is 2.2~7.1 times of the standard deviation of the regional background b-value. But the b-value of aftershocks is higher than or close to the regional background b-value.
In order to discuss the stability of b-value of foreshocks, we selected two foreshock sequences with sufficient data and studied the variation of b-value with sample size. Results show that the b-value is relatively low at the beginning, and then increases gradually with increase of sample size. When the calculated samples N≥70, the b-value is basically stable.

Key words: foreshock sequence, aftershock sequence, ETAS model, b-value

摘要：

前震是短临地震预报最有效的方法之一。文中系统研究了1976—2017年全球29次M_W≥8.0浅源地震的前震特征, 发现: 1)共有8次大地震出现前震, 占总数的27.6%, 且最大前震的震级M>5.0。这8次大地震均为逆冲型, 占逆冲型地震总数(23次)的34.8%; 2)前震序列的空间分布集中, 且集中在主震震中周围。最大前震与主震的震中距为10~53km, 震级差为1.1~2.8级。87.5%的时间差(最大前震与主震的发生时间差)为2h~15d, 仅2006年11月15日千岛群岛8.3级地震前45d发生6.6级最大前震; 3)与背景地震活动相比, 前震具有高频次活动的特点。在8次大地震中, 5次在主震前15d内加速活动; 3次加速活动的开始时间超过1个月(35~45d), 但在主震前1d和6d又再次出现前震频次显著增多的现象; 4)前震的震源机制解与主震一致, 而余震的震源机制解比较复杂; 5)使用ETAS模型计算前震和余震序列的参数α、 p和b值。为了确保参数计算的可靠性, 文中约定当前震序列满足计算样本量N≥30且最低计算震级M_j≥M_c(M_c为最小完备震级)时, 才同时计算前震和余震序列参数。此外, 文中约定余震序列的持续时间为主震后1个月。在8次大地震中, 有4个可以进行对比计算。结果显示, 对于反映激发次级余震能力的α值和序列衰减快慢的p值, 前震和余震没有规律性差异; 而反映应力水平的b值则共性特征明显, 前震b值明显低于余震; 与区域背景b值相比, 前震b值低于区域背景b值的10%~24%。前震b值与区域背景b值之差是区域背景b值标准差的2.2~7.1倍, 具有显著性。而余震b值高于区域背景b值或与其相当。为了讨论前震b值的稳定性, 计算了2个资料丰富的前震序列的b值随样本量的变化。 结果显示, 在前震序列的开始阶段b值较低, 之后逐渐增大, 当计算样本量N≥70后, b值基本稳定。

关键词: 前震序列, 余震序列ETAS模型, b值