Research On Intelligent Determination Of P-Wave First-Motion Polarity And Inversion Method Of Seismic Source Mechanism

Focal mechanism may reveal informations about the rupture process,fault structure,and regional stress field characteristics.P-wave first motion polarity is the basic parameter for focal mechanism inversion,especially for focal mechanism inversion of small-moderate earthquakes.More accurate identification of the P-wave first motion polarity can better constrain the focal mechanism solutions of small-medium earthquakes.In this paper,we compare three methods based on extensive research of domestic and international automatic P-wave first motion polarity determination methods,and the results show that the accuracy of the deep learning method to determine P-wave first motion polarity is higher than that of the traditional rule-based determination method.First,we use the DiTingMotion model based on massive data and deep learning training to determine P-wave first motion polarity,and combine with the grid search method(CHNYTX)to construct an automated data analysis process named as FocMech-Flow(Focal Mechanism Flow).Next,we apply FocMech-Flow to waveforms at stations in the Yangbi earthquake series and the Anninghe fault zone.In Yangbi 2021 Ms 6.4 earthquake sequence,we obtained 2389 P-wave first motion polarities from 112 events in May 18 to June 02 by using DiTingMotion,and the comparison between this study with manual picks show that the accuracy of polarity discrimination from FocMech-Flow may reach 98.49%.In addition,we obtain the focal mechanism of 112 earthquakes with the minimum magnitude of ML1.6 by using CHNYTX,which is more than 3 times the events from the waveform inversion method,indicating that FocMech-Flow is more efficient for obtaining focal mechanisms of small-moderate earthquakes.For Yangbi 2021 Ms 6.4 sequence,it is shown that the focal mechanisms of the foreshocks are consistent and the predominate type is strike-slip.However,the focal mechanisms of the aftershocks are more complex,in which the major type is strike-slip while the reverse and normal types are not rare.To further validate the method,FocMech-Flow is applied to the dense array of the Anninghe fault named as XiChang seismic array,and the focal mechanism of 115 earthquakes are obtained.The results show that the focal mechanism of events along the fault is mainly in strike-slip,including some events showing both reverse and normal types.This hints that the stress field along the fault zone is quite heterogeneous.In summary,this study shows that FocMech-Flow has the features of automation,higher efficiency and lower cost compared with the traditional methods of manually identified P-wave first motion polarity and the focal mechanism inversion.FocMech-Flow can significantly improve the work efficiency while ensuring the accuracy of the results,providing an important support for large-scale focal mechanism research.By obtaining a more detailed dataset of earthquake focal mechanism,high-resolution monitoring and study of regional stress fields would be reached.FocMech-Flow lays the foundation for further indepth study of the relationship between stress field and fault activity in mainland China,and provides important technical support for the study of large-scale stress field changes and seismic activity characteristics by using the focal mechanism.