The second seismic moments can be used to estimate finite source attributes such as rupture length,
width, duration, velocity, and propagation direction from the apparent characteristic durations of the
apparent source time functions (ASTFs) without a predefined rupture model. The ASTFs can be
obtained by deconvolution of the empirical Green's function (EGF).
The use of second seismic moments to resolve finite-source models has been shown to be effective
in small to moderate earthquakes in California (McGuire et al., 2001; McGuire, 2004; Meng et al.,
2020) and appears to be a promising tool for improving seismic hazard assessment, including for
future use in real time.
Since information on source parameters and rupture propagation is crucial for seismic hazard
assessment, it is important to evaluate the sensitivity and thus the reliability of the result in the
presence of uncertainties in the input variables, which can be large, especially in real time and when
the coverage of the network varies. To this end, we performed a test with a synthetic seismic source
and computed the ASTFs. We then applied 1000 random perturbations to the hypocentral depth,
focal mechanism, number of recording stations, and observed characteristic source duration
separately, and then performed the inversion to evaluate how much each data-induced uncertainty
can affect the accuracy of the resulting characteristic rupture size, propagation, and rupture velocity
in terms of both mean and dispersion.
