Top > Visiting researchers > GuoJie Meng

GuoJie Meng

Self Introduction

I am grateful to Earthquake Research Institute (ERI), University of Tokyo for offering me a significant opportunity to work with Japanese colleagues here. I appreciate greatly Professor Kato and Ms. Watanabe for their generous help in making arrangements for my visit. I am a Research Professor at the Institute of Earthquake Science, China Earthquake Administration. My principle research interests cover GPS-derived crustal deformation at regional and continent scales, fault activity, earthquake cycle deformation, and dynamic ground shaking, recorded by 1 Hz GPS data, caused by medium-sized or large earthquakes. My research involves GPS field measurements, GPS data processing and crustal deformation modeling. Recently I am particularly interested in modeling the rupture process of large earthquakes using high-rate GPS waveforms. During my 3 months long visit to ERI, I will concentrate, in collaboration with Professor Kato and Dr. Yokota, on understanding the dynamic surface deformation and the slip distribution of the 2008 Sichuan Ms8.0 earthquake by the inversion analysis of 1 Hz GPS data, in combination with some insight into strong motion. ERI is a leading institute worldwide in the domain of earthquake science, and has obtained many impressive results in various branches of earthquake research. I am looking forward to learning about the work of my ERI colleagues.

Research Report

During my three-month visit to ERI, my main research was focused on the three-fold joint inversion of high-rate GPS data, strong motion and teleseismic P-phase waveforms in an effort to elucidate the rupture history of the 2008 Wenchuan, China Ms8.0 Earthquake, which caused great casualties and property losses. I am greatly grateful to Professor Kato and Mr. Yokota for their profound help in my research.

High-rate GPS data provides essential complement to seismic waveforms. Inertial accelerometers can not distinguish between accelerations caused by rectilinear motions and accelerations that arise when a seismometer is tilted in the Earth’s gravity field(Hammond, et al., 2010). Strong motion instrument, which measure acceleration, must be integrate twice to obtain displacement, amplifying noise. Double integration of seismic data also requires extreme linearity of seismometer system as well as the knowledge of initial ground velocity at the start of integration. Several previous researches showed that high-rate GPS can provide direct measurement of arbitrarily large dynamic and static ground displacement, and can be used to finite-source inversions in the same way as seismic waveforms. Real-time high-rate data can also be possible to be incorporated into early warning system by quick acquisition of total displacement waveforms(Crowell, et al. 2009).High-rate GPS is directly sensitive to displacement. High-rate GPS derived displacement time series can be a profound constraint to source rupture .Thus joint inversion of high-rate GPS data, strong motion and teleseismic waveforms will be a significant way to infer source rupture history.

The Wenchuan Ms8.0 earthquake occurred on May 12, 2008. The origin time of the main shock is 6:28:01UTC(6:28:15 GPS time) at an epicenter of (31.021°N, 103.367°E), with a focal depth of 14 km. The associated fault rupture is mainly thrust with strike-slip components. Dynamic ground displacements of the Wenchuan Earthquake were captured at one-second intervals by a GPS networks, encompassing 12 continuous operating stations. GPS phase data was processed by the GAMIT/BLOBK software, in the way of relative positioning. Then the 1 Hz GPS data was filtered with a 0.2 Hz low-pass filter. Only the recordings of 7 strong motion stations were employed in the following-on inversion analysis in terms of their high signal-noise ratios, although more than 460 stations recorded acceleration waveforms of the main shock. All the strong motion data were filtered with a band-pass filter between 0.03-0.3 Hz. We collected the P-phase waveforms of 21 teleseismic stations worldwide and further processed with a band-pass filter of 0.02-0.3Hz. We adopted a source model of 2 faults, each incorporating 15×9 subfaults with a size of 10km by 5km. The Green’s functions for 1 Hz GPS data were computed in the frequency-wavenumber(FK) method developed by Zhu and Rivera(2002),which can reproduce both dynamic and static displacement components. The Green’s functions for strong motion waveforms were computed using the reflectivity method, originally developed by Kohketsu(1985). The Green’s functions for teleseismic P-phase waveforms were calculated following the procedure of Kikuchi and Kanamori(1991). The 1-D velocity structure for the calculation of Green’s functions was constructed based on Zhao etc. (1997). Ramp functions with a rise time of 3s were used as the temporal basis functions, with 5 time windows at each subfault. Slip vectors were represented by a linear combination of two components in the direction of 90±45°, 75°and 165° for the south fault, and, 135°and 225° for the north fault.

The inversion was subject to a smoothness constraint with a discrete Laplacian in space and time. The weights for the smoothness constraints were determined to minimize the Akaike’s Bayesian Information Criterion(ABIC)(Akaike, 1980). Identical weights are applied to the horizontal and vertical data for all the GPS stations, in spite of relatively lower signal-noise ratio of vertical displacement components.

The rupture velocity was assumed to be 3.0km/s, which triggered time of the first time window at each subfault. The positive and smoothness constraints were applied during the inversion. The preliminary resultant inversion indicates that the fault rupture initiated on the south-western end of the 300 km long fault plane and developed into a complex static rupture pattern. Major slips are located 40-120 km to the northeast of the epicenter. The total seismic moment is 1.07×1021 N.M, approximating the energy of an earthquake of Mw7.9-8.0. A large asperity was recorded close to 60km northeast of the hypocenter approximately at a depth of 12km. A small asperity was also inferred 200km northeast to the hypocenter. The snapshots of slip distribution for successive time windows show the dynamic part of the source process, indicating rupture propagate unilaterally from the hypocenter to the northeast, and slips mainly occurred in the duration of 10-60 seconds after the initiation of rupture. The joint inversion inferred an extension of the rupture toward the ground surface. With high-rate GPS data encompassed, our three-fold joint inversion shows a larger static slip pattern, both in magnitude and distribution, than that estimated from strong motion and teleseismic P-phase waveforms.

We found that 1 Hz GPS data at most stations could be fitted well except that at PIXI station, possibly due to that structural velocity utilized now is too small. Further inversion analyses are needed by changing the velocity structure beneath the station.