Seismic wave propagation in Iran and eastern Indian shield

Abstract

This dissertation addresses several important aspects of observational earthquake seismology: 1) methods for data management and processing large datasets, 2) analysis of seismic wave propagation at local to regional (up to about 700 km) source-receiver distances, 3) analysis of seismic coda, and 4) critical re-evaluation of the fundamental problem of seismic wave attenuation and measurement of the seismic “quality” factor (Q). These studies are carried out using new and previously analyzed earthquake data from Iran. In each of the four application areas above, innovative methods are used and significant new results are obtained. First, for efficient managing and processing of large earthquake datasets, I use a flexible, exploration-style open-source seismic processing system. Custom and problem-oriented scripts using Matlab or Octave software are included as tools in this processing system, allowing interactive and non-interactive analysis of earthquake records. In the second application, I note that the existing models for body-wave amplitudes are hampered by several difficulties, such as inaccurate accounts for the contributions of source and receiver effects and insufficient accuracy at the transition between the local and regional distances. Finding a reliable model for body-wave amplitudes is critical for many studies. To achieve such a reliable model, I use a joint inversion method based on a new parameterization of seismic attenuation and additional constraints on model quality. The joint inversion provides a correct model for geometrical spreading and attenuation. The geometrical-spreading model reveals the existence of an increase of body S wave amplitudes from 90 to about 115 km from the source which might be caused by waves reflecting from the crust‐mantle boundary. Outside of this distance range, amplitude decays are significantly faster than usually assumed in similar models. Third, in two chapters of this dissertation devoted to coda studies, I consider the concept of the frequency-dependent coda Q (Qc). Although this quantity is usually attributed to the subsurface, I argue that because of subjective selections of model assumptions and algorithms, Qc cannot be rigorously viewed as a function of surface or subsurface points. Also, frequency dependence of the measured Qc strongly trades off with the subjectively selected parameters of the measurement procedure. To mitigate these problems, instead of mapping a hypothetical in-situ Qc, I obtain maps of physically justified parameters of the subsurface: exponents of geometrical spreading (denoted ) and effective attenuation (denoted qe). For the areas of this study, parameter  ranges from 0.005 s-1 to 0.05 s-1 (within Zagros area of Iran) and 0.010 s-1 to 0.013 s-1 (within the eastern Indian Shield). Finally, from both body- and coda-wave studies, I derive estimates of seismic attenuation within the study areas. In two areas of Iran and within the Indian Shield, weak attenuation with Q-factors of 2000–6000 or higher is found. In particular, coda envelopes can be explained by wave reverberations within elastic crustal structures, and the Q-type attenuation appears undetectable.