"Full Waveform Ambient Noise Inversion"

Speaker: Korbinian Sager, ETH Zurich

Abstract: In earthquake tomography, modern tomographic methods exploit waveforms for the benefit of improved resolution. However, these techniques cannot be applied to noise correlation functions without knowing the distribution of noise sources. To overcome this limitation, we develop a method – referred to as full waveform ambient noise inversion – that is valid for arbitrary noise source distributions in both space and frequency, accounts for 3D heterogeneous and attenuating media and the full seismic wave propagation physics. The fundamental idea is to drop the principle of Green function retrieval, which is the basis for current noise tomographic studies, and to establish correlation functions as self-consistent observables in seismology.
Many studies, which were previously impossible or complicated to conduct, are now rendered feasible. For a first use case, we investigate the emergence of signals in correlation functions in general and in particular of body waves. In contrast to surface waves, body waves likely emerge from very localized sources and their interpretation as pure body waves is prone to misinterpretation due to significant cross-terms. In addition, we present a first application of full waveform ambient noise inversion with real data, a global data set focusing on the Earth's hum period band. We jointly invert for the distribution of noise sources and Earth structure.