Estimating Dynamic Source Properties of Giant Megathrusts

The energy released during an earthquake includes contributions from seismic radiated energy E_R and fracture energy E_G, which includes inelastic cracking but not frictional heat generation. Uncertainties in our measurements of E_R and stress drop directly affect our estimate of E_G, and therefore may bias our understanding of earthquake mechanics. We explore observations of E_R from the body-wave far-field velocity source spectrum, and also consider surface-wave contributions to the radiated energy for shallow events. We find that uncertainties in the high-frequency fall-off rate of the source spectrum are the dominant contributor to uncertainties in the scaled radiated energy E_R/M₀, in contrast to the corner frequency uncertainties that have the largest affect on dynamic stress drop estimates.

We investigate the sources of uncertainties that arise from commonly used approaches to remove the path effects from velocity seismograms, i.e., empirical corrections for radiation pattern differences on amplitude and empirical Green’s functions (eGf). Recent dynamic rupture simulations (Kaneko and Shearer, 2014a, 2014b) have highlighted the role of take-off angle, rupture velocity, and source directionality on single-station estimates of stress drop and scaled energy from both P and S source spectra. We thus investigate the likely errors in these parameters that arise in real data from an incomplete sampling of the focal sphere.

We re-evaluate E_R/M₀ estimates for large recent M8.5+ megathrust earthquakes and their uncertainties associated with intrinsic variability in the radiated energy, from removing the path effects, and from incomplete sampling of the focal sphere.