Complex short-period teleseismic P-waveforms (consisting of the direct P wave and surface reflections pP and sP) are observed from many earthquake sources. It is often not possible to easily interpret these waveforms in terms of those three phases. This is necessary to obtain accurate earthquake depths and P and S wave radiation patterns. This thesis examines the contribution made by various factors to P-wave seismogram complexity using both synthetic and real data. First, using a number of synthetic waveforms it is confirmed that long duration sources can contribute significantly to the complexity of short-period waveforms. However, it is highlighted that by using broadband recordings much of this complexity can be accounted for, and attributed to the limited passband of the short-period recording system. In addition, S-to-P mode conversions at near-source structure can also contribute significantly to the complexity of the short-period waveform. Second, the causes of differences in the complexity of the short-period waveforms from the 1987 Whittier Narrows and the 1991 Sierra Madre earthquakes are examined. Originally these earthquakes were thought to be separated by a distance approximately the size of the first Fresnel Zone, and hence should, in theory, have indistinguishable near-source structure, when seen at teleseismic distances. Using relative amplitudes, the published CMT focal mechanism for these events is confirmed . In the case of Sierra Madre earthquake it was also possible to positively identify the one surface reflection, visible on the shortperiod seismogram, as pP. Even with complex waveforms the relative amplitude method can be used to place constraints on the focal mechanism of the Whittier Narrows earthquake. Using forward modelling, with a simple kinematic source model, synthetic seismograms are matched to the observed broadband seismograms for both earthquakes. Using this simple source model, the variation in the source duration, caused by the difference in source rupture areas between the two earthquakes, is sufficient to account for the first-order variations in complexity seen. To second order, the near-source structure is sufficiently different even at the limit of resolution of the data, to contribute to some extent to the observed complexity variation, most likely due to the large thickness of sediment west of the Sierra Madre Fault. Third, a suite of seismograms from the 29 October 1995 Caspian Sea earthquake is examined. Using relative amplitudes, the surface reflection on these seismograms is correctly identified and the actual depth estimated to be 48 km. From this it is shown that an arrival mis-identified by the Prototype International Data Center as a surface reflection is most likely to be be a mode conversion at an interface 80 km beneath the source. Forward modelling of the broadband and short-period waveforms shows that these mode-conversions are enhanced by the downward propagating line rupture, and are best seen when the position of the stations are at a node in the P-wave radiation pattern. This produces an apparently complex waveform. Visible S waves from this earthquake at European stations show the very low attenuation in the mantle path and this may contribute to the greater than usual complexity observed for this event. Finally several earthquakes that appear to show seismogram complexity that cannot be explained using a simple kinematic source model or path effects are examined. By modifying an existing finite-difference fault modelling code I present a possible dynamic source model that may provide one explanation for this additional complexity. This model includes real source physics (friction law, rupture criteria) and material heterogeneities. It produces complex farfield pulse shapes that vary with fault length, material heterogeneity, initial state of stress and attenuation.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:669297 |
| Date | January 2003 |
| Creators | Snowden, Conor B. |
| Publisher | University of Edinburgh |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://hdl.handle.net/1842/11422 |
Page generated in 0.0024 seconds