Skip to Content

Earthquake Source Inversions Based upon Adjoint Methods

Earthquake source inversions are important to seismic hazard analysis.The Global Centroid- Moment-Tensor (GCMT) project( ) routinely determines point source parameters for earthquakes with magnitude (Mw ) greater than 5 based on the normal mode theory in 1-D earth model since 1981 (e.g. Dziewonski et al. 1981; Ekstrm 2006). For a large earthquake (Mw ≥ 8.5), one point source representation might not be enough. Then it is reasonable to use multi-point source to model the earthquake(e.g. Tsai et al. 2005). Another common routine is finite fault inversion, which maps spatial and temporal distributions of slip on a fault plane (e.g. Olson & Apsel 1982; Hartzell & Heaton 1983; Ji et al. 2002).

In this talk, I present multiple moment-tensor solution of two earthquakes, namely, 2004 December 26 Sumatra event and 2010 February 27 Chile event, based upon adjoint methods. A misfit function Φ(m), where m is the multiple-point source model, measures the goodness of weighted waveform fit between data and synthetics. Synthetics are calculated by spectral-element simulations (SPECFEM3D GLOBE) in a 3D global earth model S362ANI to reduce imperfect structural effects, and the Frechet derivatives are calculated based upon adjoint methods implemented in SPECFEM3D GLOBE. Our initial source models are obtained by monitoring the time-integrated adjoint strain tensors around the presumed source region. We then utilize the misfit value and Frechet derivatives in a preconditioned conjugate-gradient methods to iteratively invert for the source model.