Inversion For The Rupture Process Of The 2016 Meinong Earthquake Using High-rate GPS And Strong Motion Data

A magnitude Mw=6.4 earthquake struck the Meinong district of Kaohsiung city in Taiwan on February 6,2016.This event caused several building collapse,crack and soil liquefaction in Tainan,including 117 fatalities.The source rupture process refers to the propagation process of the rupture in the source area after the initial rupture occurs.It mainly explains the evolution of the seismic dislocation energy or seismic moment on the fault plane over time from the beginning of the earthquake to the end.Studies of source rupture process is of great theoretical significance to the source dynamics and of practical significance to post-earthquake rescue and relief.High-rate sampling GPS observation is the distance between the satellite and the receiver,which involve not only the static displacement but also the dynamic ground displacement caused by the earthquake.Because its amplitude does not suffer from saturated,it is widely used in fault slip inversion,tsunami and volcano monitoring.The digital strong motion seismometer records the acceleration of ground motion,it has the advantages of wide frequency band,high sampling rate and large dynamic range,etc.However,affected by the defects of the strong motion seismometer,saturated amplitude and baseline shifts during seismic shaking is inevitable.The two kinds of data complement each other and are widely used in seismic research.In this paper,the nonnegative linear least squares point-by-point inversion method was used to infer the source rupture process of the Meinong earthquake using 1-Hz GPS and strong motion data.In this study,coseismic surface deformation imaged by ascending and descending InSAR was used to determine the fault geometry of Meinong earthquake.The preferred fault geometry parameters are obtained with strike of 271.7°,dip of 22.7°,the average slip angle of 1.3°,length of 60 km along the strike direction and width of 60 km along the down-dip direction,based on simulated annealing algorithm.In order to obtain detailed information of rupture,the fault plane was further divided into sub-faults sized of 3 km×3 km along the fault strike and down-dip,and each sub-fault was regarded as a point source.To infer the source rupture process,the 1-D velocity structure model of Taiwan region was obtained first and then we employed the frequency-wavenumber approach to compute the Green's functions at horizontal layered model.In order to ensure the stability and reliability of the inversion result,we appended the constraints explained bellow and build the inversion model.The Laplace smoothing constraint is introduced to avoid large variations in dislocation between adjacent sub-faults,and the scalar seismic moment minimization constraint is introduced to make the inversion result more practical.We explored the trade-off relation between residual error with 1-D velocity structure model and average solution roughness.Thus,the preferred rupture velocity and smoothing factor are determined.The joint inversion of geodetic and strong motion data shows that the rupture is a mixed motion of thrust and left-lateral strike-slip,with maximum slip of around 1m and the main slip area located 10¹7 km underground.The calculated seismic moment is about 1.47×10¹⁹ Nm,equivalent to moment magnitude Mw 6.7.We also analyzed the rupture process of Meinong earthquake,the result shows that the entire duration was around 16s and the propagation was directional.The rupture initiated in the vicinity of hypocenter and propagated towards north along down-dip and west along strike,it can be divided into three main stages.In 1⁴ s,the slip was very small in the vicinity of hypocenter.In 5¹3 s,the rupture propagated with more slip.Finally,the rupture gradually decreased in 14¹6 s.