Logo Logo
Help
Contact
Switch language to German
Applications of ambient seismic noise: clock error detection and group velocity estimation in land and ocean bottom seismograms
Applications of ambient seismic noise: clock error detection and group velocity estimation in land and ocean bottom seismograms
Ambient seismic noise generated by ocean waves is continuously present in seismograms and has previously been considered as undesired, disturbing signal. However, it has been shown that cross-correlations of noise recorded at two seismic stations converge towards the Green's function. This function describes the underground properties between the two stations. Ever since this finding, a wide range of noise applications has been described, which are still under development. Temporal changes of noise cross-correlations can be used for detection of clock errors in seismic data, while group velocities derived from cross-correlations are the basis for tomographic studies. In the first part of this thesis, an extensive clock error study of land stations and ocean bottom seismometers (OBSs) is presented. A new multiple-component approach is applied, which enhances the accuracy (~20 ms) of the detected clock errors significantly. Moreover, this approach allows the retrieval of clock errors with high temporal resolution of 1-2 days, even for large interstation distances (~300 km). The application of the described approach to data sets with low timing quality could highly increase their usability for structural studies. The second part of this thesis deals with group velocity curves that are mainly retrieved from OBS cross-correlations with interstation distances of up to ~2000 km. A joint inversion of the noise group velocities together with earthquake data from a prior study yields a high-resolution crustal S-wave velocity model of the western Indian Ocean. This model highly reflects tectonic structures in this region, like ocean ridges and plateaus. These results demonstrate the feasibility of large-scale OBS noise tomography of ocean basins, while prior studies were limited to smaller scales. In the last part of this thesis, first steps towards a seismological image of the island La Réunion are presented. Group velocity curves are derived from noise cross-correlations between island stations. In agreement with gravity studies, the group velocities indicate a spacious high-velocity body beneath the ancient volcano of the island. The inversion of the group velocities will yield a tomographic model of the island, which may be used as starting velocity model for future seismic surveys. The majority of the seismic stations used in this thesis were installed in the western Indian Ocean on and around La Réunion during the RHUM-RUM project. This reflects the high applicability of noise cross-correlations (from data quality inspection to crustal imaging) within the same data set. The detailed method descriptions of this work provide a valuable guideline for future studies, that deal with land or OBS seismograms.
Not available
Hable, Sarah
2019
English
Universitätsbibliothek der Ludwig-Maximilians-Universität München
Hable, Sarah (2019): Applications of ambient seismic noise: clock error detection and group velocity estimation in land and ocean bottom seismograms. Dissertation, LMU München: Faculty of Geosciences
[img] PDF
Hable_Sarah.pdf

39MB

Abstract

Ambient seismic noise generated by ocean waves is continuously present in seismograms and has previously been considered as undesired, disturbing signal. However, it has been shown that cross-correlations of noise recorded at two seismic stations converge towards the Green's function. This function describes the underground properties between the two stations. Ever since this finding, a wide range of noise applications has been described, which are still under development. Temporal changes of noise cross-correlations can be used for detection of clock errors in seismic data, while group velocities derived from cross-correlations are the basis for tomographic studies. In the first part of this thesis, an extensive clock error study of land stations and ocean bottom seismometers (OBSs) is presented. A new multiple-component approach is applied, which enhances the accuracy (~20 ms) of the detected clock errors significantly. Moreover, this approach allows the retrieval of clock errors with high temporal resolution of 1-2 days, even for large interstation distances (~300 km). The application of the described approach to data sets with low timing quality could highly increase their usability for structural studies. The second part of this thesis deals with group velocity curves that are mainly retrieved from OBS cross-correlations with interstation distances of up to ~2000 km. A joint inversion of the noise group velocities together with earthquake data from a prior study yields a high-resolution crustal S-wave velocity model of the western Indian Ocean. This model highly reflects tectonic structures in this region, like ocean ridges and plateaus. These results demonstrate the feasibility of large-scale OBS noise tomography of ocean basins, while prior studies were limited to smaller scales. In the last part of this thesis, first steps towards a seismological image of the island La Réunion are presented. Group velocity curves are derived from noise cross-correlations between island stations. In agreement with gravity studies, the group velocities indicate a spacious high-velocity body beneath the ancient volcano of the island. The inversion of the group velocities will yield a tomographic model of the island, which may be used as starting velocity model for future seismic surveys. The majority of the seismic stations used in this thesis were installed in the western Indian Ocean on and around La Réunion during the RHUM-RUM project. This reflects the high applicability of noise cross-correlations (from data quality inspection to crustal imaging) within the same data set. The detailed method descriptions of this work provide a valuable guideline for future studies, that deal with land or OBS seismograms.