Abstract: Seismic interferometry imaging can accurately determine the source location. The main shock and some aftershocks of Sichuan Changning M_S6.0 earthquake occurring on 17th June, 2019 are located with seismic interferometry imaging in this study. This technique is different from the traditional earthquake location method in that it does not use the phase arrival data in the earthquake catalog. Due to the existence of seismic phase picking errors, the traditional earthquake location method has a basic limitation on the amount of location error reduction. Interferometry imaging method directly applies the waveform records for source location and gets the travel time difference information from cross-correlation calculating, thus, greatly reduces the phase reading error. There are three main processes of interferometric imaging location technique, that is, waveforms cross-correlation of onset phases between station pairs, interferometric waveforms migration, and superposition. However, the complex focal mechanism and radiation pattern of natural earthquakes will cause the polarities of the first arrival phases to be different. When performing superposition processing, the addition of the positive and negative amplitudes will reduce the superimposed energy. By calculating the characteristic functions of the original waveform records, the inconsistency of the polarity of the initial phase caused by the different radiation patterns of the source in different azimuths is eliminated. Since the seismic stations used for location are regional network, the distance between station pairs is relatively close, and waveforms cross-correlation can eliminate some velocity disturbances. Besides, there is a certain coupling between origin time and source location. Therefore, the origin time and source location should be decoupled during locating. The waveform cross-correlation between station pairs can subtract the same origin time, thus, eliminating the location error caused by the disturbance of origin time. In addition, an advantage of the interferometric imaging location method is that it increases the amount of available data. The non-repetitive pairing between stations makes the travel time difference data far more than the direct wave phase data, and these travel time difference data are independent of each other and have no correlation. In this study, the natural earthquakes are located by applying interferometric imaging technology with cross-correlation migration kernel function. After migration and superposition of interferometric waves, the horizontal position and depth of the sources are imaged. The location of the main shock is(28.38°N, 104.88°E, 8.0km), and it is on the west of the aftershock belt. We compared the result from this study with the results of USGS(28.40°N, 104.95°E, 10km), GFZ(28.43°N, 104.94°E, 10km)and CENC(28.34°N, 104.90°E, 16km). The epicenter positions of the four results are relatively consistent, and the deviation is within 0.07°. The depth from our study is consistent with the results from USGS and GFZ, and the difference is 2km. In this study, the depths of M_S4.0~4.9 aftershocks are 5km. Three M_S5.0~5.9 aftershocks are distributed in the depth of 8~10km. The Changning earthquake sequence is located at the western end of the Changning anticline, which is the main geological structure in the Changning area and trending NWW-SEE generally. The anticline is about 100km long from east to west, and about 20km wide from north to south. The Changning anticline was formed in the Mesozoic Era. It was pushed by the NE-SW trending tectonic stress at that time, and accompanied by multiple small faults, shown as high-angle compressive thrust faults. The epicenter distribution shows that the aftershock belt is distributed along the NWW direction. After interferometric imaging location, the average travel time residual is 0.6s. In this study, we used four different velocity models(CodaTomo, Crust1.0, IASPI91 and SIMPLE)for calculating the theory travel time for migration. The effects of four different velocity models on the location are tested and the results show that the seismic interferometric imaging location method is stable. The average travel time residuals of four velocity models are 0.66s, 0.68s, 0.80s, and 0.71s. By calculating the array/network response function, the influence of the station distributions and the length of the characteristic function window on the positioning result are evaluated. The network response functions with four dominant frequencies at 0.01Hz, 0.05Hz, 0.1Hz and 0.25Hz were calculated and compared. The network response functions have fewer local maximums but converge slowly at 0.01Hz, 0.05Hz, and 0.1Hz. In the depth direction, the resolution is very low. The dominant frequency of the eigenfunction calculated in this study is about 0.25s. At this frequency, the network response function shows good convergence and stability in both horizontal location and source depth.

Key words: seismic interferometry, earthquake location, interferometic imaging, migration and superposition, Sichuan Changning earthquake

摘要： 文中利用干涉成像定位方法对2019年6月17日四川长宁M_S6.0主震及部分余震进行定位。 天然地震波振幅的极性和大小随震源机制和辐射花样的变化而变化, 利用原始波形的特征函数可消除震源在不同方位上远场P波的初动极性和大小的不一致性。 文中将干涉成像技术应用于天然地震定位, 通过对干涉波形进行偏移和叠加处理, 分别对震源的水平位置和深度进行偏移成像, 确定了主震及较大余震(M_S≥4.0)的震源位置参数, 其中主震的位置为(28.38°N, 104.88°E), 震源深度为8.0km。 此外, 还测试了4种不同的速度模型对定位的影响, 结果显示利用震源干涉成像定位方法获得的结果较为稳定。 通过计算台阵、 台网的响应函数, 评估了台站分布及特征函数周期长短对定位结果的影响。 计算得到的特征函数优势周期约为4s, 该周期的台阵、 台网响应函数在水平和垂直方向均显示了较好的收敛性和稳定性。

关键词: 地震干涉, 地震定位, 干涉成像, 偏移叠加, 四川长宁地震