Adjoint Moment Tensor Inversion for Non-Double-Couple Earthquake

Adjoint method provides an elegant framework for computing the gradient of the misfit function with respect to model parameters. For the case of full-waveform adjoint moment tensor inversion, this can be achieved by recording the adjoint strain field in the source location. This adjoint strain is the product of adjoint simulation which can be achieved by solving the wave equation backwards in time from the receiver to the earth model and using waveform misfit as the adjoint source. In this paper, several test cases were defined to investigate the potential and stability of adjoint full-waveform inversion to invert moment tensor of non-DC earthquakes. The results from the test cases shows that misfit convergence to the reference model can be achieved if proper pre-processing step is applied to the waveform. Even for non-DC earthquake inversion that was using DC earthquake as a starting model. If this pre-processing step is not applied, this method can introduce a non-DC noise originating primarily from a non-accurate velocity structure and poor misfit convergence. Keywords: adjoint method, full-waveform inversion, moment tensor