Joint inversion of centroid moment tensor for large earthquakes by combining teleseismicP-wave andW-phase records

Abstract

SUMMARYCentroid moment tensor inversion involves the determinations of the moment tensor solution, which is usually done with low-frequency signals to meet the requirement of the point-source approximation, and the moment centroid, which requires high-frequency signals to improve the resolution. Traditional centroid moment tensor inversion techniques, such as the W-phase inversion, mainly use low-frequency data to estimate the magnitude and fault parameters, which limits the resolution of the moment centroid. In this study, we combine the P wave and W phase to constrain both the moment tensor and moment centroid. The rupture directivity is considered in the P-wave inversion, and the rupture velocity is resolved by inverting the P wave solely. The moment centroid is estimated by the rupture velocity from the P-wave inversion and the grid search in the W-phase inversion. The final moment tensor solution is determined based on the moment centroid by jointly inverting both P-wave and W-phase data. The resulting centroid moment tensor solution can fit a broad frequency band (0.001–0.1 Hz) of waveforms. Through synthetic inversion tests and applications to several large earthquakes, we demonstrate that the magnitudes and fault parameters from our joint inversion are more stable than the P-wave inversion, and the moment centroids, especially the centroid depths, seem more reasonable than those from the W-phase inversion.