Locations: United States of America
For the vast regions of continental U.S. in and east of the Rocky Mountains, there is a lack of empirical near-field ground motion data. Predictions of ground motion from future damaging earthquakes in these regions fundamentally rely on the issue of how similar these future earthquakes will be to the previously recorded earthquakes that are either smaller, or larger but recorded in other regions such as California. Currently the seismological community is divided on the answers to several key questions regarding the similarities among earthquakes in different magnitude ranges, or in different tectonic environments. These questions include: (1) how do seismic moments (M0) scale with corner frequencies (ƒc)? (2) which of the various 1-ƒc and 2- ƒc source models best describes the shape of the true source spectra? (3) do static stress drops tend to remain constant with varying event sizes and tectonic environments? And (4) do ratios of the radiated energy to seismic moments (or, the apparent stress) tend to remain constant? The answers to these questions have direct impacts on the ground motion attenuation relationships used by the probabilistic hazard maps. Moreover, they lead to implications for source physics, such as whether strength and status of stress of seismogenic faults vary with the general tectonic environment, the far-field loading rate and deformation styles. To strongly constrain the answers to these questions, we propose to invert for source spectra from earthquakes recorded in the past two-to-three decades in the various regions in continental U.S. using the seismic Lg phase. The strength of the proposed research lies on first, the Lg is the dominant phase on regional seismograms in the frequency band between 10 -1 to 101 Hz, and is practically the only viable phase for a comprehensive study. Second, the inverse algorithms to be used are well-calibrated as a result of a decade of effort of development and improvement. In particular, if used carefully, they are capable of removing the well-known trade-off between source spectra and path Q in spectral inversions. A bi-product of this research will be path-variable Lg Q values that can be used to develop a tomographically regionalized Lg Q model in much of the continental U.S.
Cross Cutting Themes:
Hazards and Risk