Ambient seismic source inversion
Ocean waves and other phenomena occurring at the Earth’s surface interact with the Earth’s crust and cause faint seismic signals that can be measured at great distance. The sources of these ambient vibrations are of long-standing interest in seismology, both in their own right, as they carry informa...
| Main Author: | |
|---|---|
| Other Authors: | , , , |
| Format: | Doctoral or Postdoctoral Thesis |
| Language: | English |
| Published: |
ETH Zurich
2017
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/20.500.11850/209095 https://doi.org/10.3929/ethz-b-000209095 |
| Summary: | Ocean waves and other phenomena occurring at the Earth’s surface interact with the Earth’s crust and cause faint seismic signals that can be measured at great distance. The sources of these ambient vibrations are of long-standing interest in seismology, both in their own right, as they carry information about environmental processes and conditions, and because they persistently probe the Earth’s interior, providing signals for nearly continuous imaging and monitoring of subsurface structure even in areas of low seismicity. Here, we present the first application of an iterative inversion method for the sources of ambient seismic noise with a three-dimensional Earth model. In a step leading up to inversion, we investigate how robust information about noise source properties can be derived from cross-correlations of continuously recorded ambient noise. Signal energy ratios of ambient signals traveling in opposite directions can be used to rapidly elaborate first-order estimates of ambient noise source distribution at a regional and global scale. At the regional scale, windowed signal energy measurements taken on the cross-correlation reflect the rapidly changing ambient noise field excited by passing storms. Based on a Green’s function database approach, we numerically model cross-correlations of the ambient seismic noise in a three-dimensional Earth model with laterally varying seismic structure. This allows us to construct a gradient-based, non-linear iterative inversion for the time-, location- and frequency dependent source power spectral density of ambient noise that honours the three-dimensional structure of the Earth’s interior. We apply this inversion to ten-year averaged observations of vertical-component ambient noise recorded in North and South hemisphere winter, in order to image the sources of the Earth’s hum, which is the long-periodic background seismic signal. The results reveal seasonally varying, narrowly delineated areas of high hum excitation, predominantly located at Pacific shelves or coasts during North Hemisphere winter, and at Southern Ocean locations of shallow bathymetry, as well as South Pacific shelves or coasts during austral winter. The investigated inversion method contributes to the development of full-waveform inversion with ambient noise cross-correlations. Future extension to horizontal- and mixed-component cross-correlations, as well as applications to ambient noise at the regional scale, may help advance our understanding of ambient noise excitation processes. |
|---|