Abstract:

The Yangbi M_S6.4 and Maduo M_S7.4 earthquake occurred successively on May 21~22, 2021 in Dali, Yunnan Province and Guoluo, Qinghai Province of China. The earthquakes caused deformation of boundaries with density difference and changed the density of rocks around fault due to volumetric strains, thus, disturbing the earth’s gravity field. The Earth’s time-varying gravity field contains rich information about distribution and migration of materials in the Earth system and provides very important constraints for the structural and kinematics characteristics, the interaction of various layers and coupling mechanisms of the solid Earth. Therefore, gravity observation means a lot for earthquake monitoring.

The background noise is the continuous high-frequency oscillation on the observation instruments placed on the surface of the earth, such as seismometers and gravimeters, which is affected by many factors such as oceans, atmosphere, wind fields, earthquakes, human activities, etc. The background noise of the gravimeter may vary in different times and locations. However, temporal and spatial variation of background noise before and after an earthquake has not been studied yet. The existing research has observed gravitational disturbance signals before major earthquakes, but it is difficult to capture them directly from the original gravity data without pretreatment.

Permutation entropy(PE)can characterize the randomness of the signal or detect signal mutation, and has been widely used in biomedicine, finance, mechanical vibration time series. In this study, we use PE to detect gravitational disturbance signal from raw data and study the background noise changes before and after the earthquake.

In this paper, 1Hz sampling records from 15 continuous tidal gravimeters(including the types from PET/gPhone to OSG)of Continuous Gravity Network of China, with spanning from 1th Jan to 30th June, 2021, were obtained and analyzed. Firstly, A bandpass filter(0.1~0.18Hz)was employed to extract gravity disturbance, after removing firstly the earth tides and atmospheric effects with the DDW model and a barometric admittance(-0.3×10^-8m·s^-2/mbar). The short time Fourier transform was used to determine the time-frequency characteristics of non-tidal gravity signals. Then, we calculated the PE and background noise magnitude(SNM)in seismic frequency band(200~600s)of the records. The results show that: 1)All kinds of gravimeters can record high-precision solid earth tide, and respond well to high-frequency fluctuation signals caused by most earthquakes above magnitude 6 in the world. 2)There was a set of gravity disturbance signals recorded by most stations on May 15~18th, and there were two other sets of disturbance signals at coastal stations. The disturbance amplitude was ±(10~100)μGal, and there was no evidence to show that the smaller the epicenter distance, the larger the disturbance amplitude. The large disturbance amplitude of coastal stations and the other two groups of disturbances may be related to sea wave pulsation or local rainfall. 3)The PE value of the original gravity record basically oscillated near a high value, in which the spring gravimeter was close to 1, and the superconducting gravimeter was 0.7. From this point of view, we find that the signal-to-noise ratio of superconducting gravimeter is higher than spring gravimeter; Without any preprocessing procedure on original gravity records, PE could effectively explore the abrupt-change signal in the records. The continuous and significant drop of PE corresponded to perturbation signal before the earthquake, and the instantaneous downward pulse of PE agreed with the tremor signal caused by the seismic wave. 4)We find that the background noise had a clear upward trend occurring two months(early March)before the earthquake; The spatial distribution of SNM indicates that the background noise of the gravimeter located in the northern Qinghai-Tibet Plateau and the Bayan Har block has a significant increase before the earthquake. 5)We preliminarily speculate that the preseismic gravity disturbance is the low-frequency tremor generated by the slow slip of seismogenic fault, and the increase in background noise before the earthquake may be related to the increased activity of seismogenic tectonic block.

Key words: Yangbi M_S6.4 earthquake, Maduo M_S7.4 earthquake, gravity perturbation, seismic background noise, permutation entropy

摘要：

2021年5月2122日, 云南漾濞、 青海玛多先后发生了 M_S6.4 和 M_S7.4 地震。为检测震前是否存在重力扰动与重力仪背景噪声异常信号, 文中基于中国连续重力台网中15台重力仪的秒采样数据, 成功提取了震前重力扰动信号, 并计算了排列熵(PE)及重力仪在地震频段(200~600s)的背景噪声等级(SNM)。结果表明: 1)大地震发生前多数台站在2021年5月1518日存在1组重力扰动信号, 其中沿海台站同时还存在其他2组扰动信号, 扰动幅度为±(10~100)μGal, 扰动频率集中在0.15~0.25Hz。沿海台站的扰动幅度普遍大于内陆台站, 震中距与扰动幅度无明显相关性, 沿海台站扰动幅度较大及产生其他2组扰动的原因可能与海浪脉动或局部降雨有关; 2)不需要对原始重力序列作任何预处理, PE能够快速检测到数据中的突变信号, PE时间序列中持续显著下降的信号对应震前扰动信号, PE瞬时向下的脉冲信号对应地震波引起的颤动信号; 3)多个台站的SNM时变曲线显示, 震前2个月(2021年3月初)直至主震发生时背景噪声水平显著升高, SNM的空间分布表明, 震前1~2个月, 位于青藏高原北缘、 巴颜喀拉地块附近台站的背景噪声有明显增大的趋势, 震后噪声水平显著降低。综上, 我们初步推测: 震前的重力扰动可能是由发震断裂慢滑移产生的低频颤动信号, 震前背景噪声的增大可能与发震地块的活动性增强有关。

关键词: 漾濞M_S6.4地震, 玛多M_S7.4地震, 重力扰动, 地震背景噪声, 排列熵