25 May 2021 in Bulletin of the Seismological Society of America
The curvilinear-grid finite-difference method (FDM), which uses curvilinear coordinates to discretize the nonplanar interface geometry, is extended to simulate acoustic and seismic-wave propagation across the fluid–solid interface at the sea bottom. The coupled acoustic velocity-pressure and elastic velocity-stress formulation that governs wave propagation in seawater and solid earth is expressed in curvilinear coordinates. The formulation is solved on a collocated grid by alternative applications of forward and backward MacCormack finite difference within a fourth-order Runge–Kutta temporal integral scheme. The shape of a fluid–solid interface is discretized by a curvilinear grid to enable a good fit with the topographic interface. This good fit can obtain a higher numerical accuracy than the staircase approximation in the conventional FDM. The challenge is to correctly implement the fluid–solid interface condition, which involves the continuity of tractions and the normal component of the particle velocity, and the discontinuity (slipping) of the tangent component of the particle velocity. The fluid–solid interface condition is derived for curvilinear coordinates and explicitly implemented by a domain-decomposition technique, which splits a grid point on the fluid–solid interface into one grid point for the fluid wavefield and another one for the solid wavefield. Although the conventional FDM that uses effective media parameters near the fluid–solid interface to implicitly approach the boundary condition conflicts with the fluid–solid interface condition. We verify the curvilinear-grid FDM by conducting numerical simulations on several different models and compare the proposed numerical solutions with independent solutions that are calculated by the Luco-Apsel-Chen generalized reflection/transmission method and spectral-element method. Besides, the effects of a nonplanar fluid–solid interface and fluid layer on wavefield propagation are also investigated in a realistic seafloor bottom model. The proposed algorithm is a promising tool for wavefield propagation in heterogeneous media with a nonplanar fluid–solid interface.
9 October 2026 in Bulletin of the Seismological Society of America
Earthquake forecasting along active faults is important for reducing future seismic losses, especially in active zones such as Yunnan Province. However, the major earthquake recurrence patterns of faults in Yunnan remain understudied, and the time-dependent probabilities are still unclear. Here, we collected paleoseismic data on 20 fault segments in Yunnan, fit these data using four types of renewal processes, and calculated the probability of at least one major earthquake occurring on each segment over the next 30 yr. Approximately half of the fault segments show evidence of quasi-periodic recurrence, as indicated by the 95% credible intervals of their shape parameters being inconsistent with the random recurrence boundary. The Weibull renewal process is more likely to best reproduce recurrence times, followed by the lognormal renewal process. Five (≥4%) fault segments have relatively high earthquake occurrence probabilities. We also estimated the distribution of next-event times and found that fault segments with higher earthquake probabilities generally tend to have earlier expected times for the next earthquake. In this study, we conducted long-timescale recurrence analysis under restrictive assumptions, without taking into account multisegment rupture, stress interaction, and magnitude-specific modeling. Therefore, further research is required in these aspects in follow-up research.