Seismic sources
Investigating seismic source properties is important not only for understanding their nature, but also to avoid mapping large source uncertainties into structural heterogeneities.
Dynamic earthquake rupture
(P. Galvez, J.-P. Ampuero, L. Dalguer, T. Nissen-Meyer)Earthquake rupture is a highly complex process, for which no singular physical theory has been established. We implement heuristic models of dynamic fault rupture (e.g. slip-weakening) into the flexible spectral-element method SPECFEM3D such that the entire process from dynamic earthquake initiation along a complex fault through wave propagation in 3D heterogeneous media is achieved. We benchmarked the implementation for branching and splay faults against existing solutions provided by SCEC, and are in the process of modeling the complex Tohoku M9 rupture 2011. Ultimately, the benefit of this implementation is the joint treatment of rupture and large-scale wave propagation including calculation of adjoint-based sensitivity kernels within one code.
Source imaging and discrimination
(T. Nissen-Meyer, M. Mai, J.-P. Ampuero)We embark on characterizing earthquake sources by backprojection with full teleseismic wavefields (similar to reverse-time migration in exploration seismology). It is computationally trivial to reach the highest desirable frequencies (e.g. Hz-range for global distances) with AXISEM. Moreover, this method relies on a separate treatment of moment-tensor elements and can therefore help in discriminating volumetric, explosive wave- field characteristics from indigenous (i.e. traceless, shear-dominated radiation) sources within this full-wave teleseismic inversion framework. Developments are also under way to incorporate generic teleseismic Green’s functions generated with AXISEM into the finite-source rupture model database of Martin Mai. The fact that most kinematic finite-fault models rely on very few discrete points in depth makes our axisymmetric approach highly efficient, also in light of finite-fault inversions.
Origin of ambient noise
(P. Basini, S. Hanasoge, T. Nissen-Meyer, L. Stehly)Ambient-noise cross-correlations have opened new doors in mapping Earth structure in regions with little seismicity. However, the Green’s function is only retrieved upon a series of assumptions, most notably the uniform distribution of ambient noise. We attempt to constrain and correctly incorporate the non-uniform origin of ambient noise with two different approaches: Based on global correlations of modeled and observed noise amplitudes in 1D global reference Earth models, we attempt to invert for the global noise origin in a time-varying manner. In more local settings (e.g., Europe), we compute the gradient of cross-correlation misfits with respect to the noise source location using adjoint methods in 3D structural models as an integral part of non-linear inversions of ambient noise.