Short-Distance Attenuation Characteristics and Ground-Motion Model for Induced Earthquakes in the Rongxian Shale-Gas Region, Sichuan Basin, China

The Rongxian shale gas field in the southern Sichuan basin has experienced increasing induced seismicity, with frequent earthquakes adversely impacting local communities and even causing casualties. In this study, we investigate the short-distance attenuation characteristics of strong ground motions from induced earthquakes with local magnitudes (ML) ranging from 2.0 to 3.9, using 7249 high-quality recordings selected from a total of 11,161 accelerograms captured by a dense network of 52 microelectromechanical systems accelerometers. To ensure data reliability, we evaluate a deep-learning denoising technique for the acceleration records and apply a magnitude-dependent usable-period cutoff. Our analysis reveals that the induced earthquakes in Rongxian exhibit regional attenuation and source-scaling features distinct from those of the selected reference ground-motion models (GMMs). Specifically, the BSSA14 model (Boore, Stewart, et al., 2014), developed within the Next Generation Attenuation-West2 project, shows distance-dependent residual trends for peak ground acceleration (PGA) and selected short-period intensity measures, whereas the induced-seismicity NAA18 model (Novakovic et al., 2018) provides a more compatible reference but retains period- and magnitude-dependent residuals. To address these discrepancies, we developed a region-specific GMM using nonlinear mixed-effects regression. Within the observed moment magnitude range (Mw 1.8–3.7), the model successfully reproduces the main attenuation and source-scaling behavior of horizontal RotD50 PGA, peak ground velocity, and spectral accelerations with residuals generally centered near zero. For larger-magnitude scenarios, the GMM smoothly transitions from the empirically calibrated model to relative magnitude–distance scaling derived from regional stochastic simulations, with extrapolated predictions explicitly distinguished from the empirical calibration range. These findings highlight the ground-motion characteristics of induced seismicity in the Rongxian shale-gas region and provide a regional ground-motion basis for rapidly estimating short-distance shaking and the spatial extent of felt or affected areas, including the consideration of selected source-model epistemic uncertainty beyond the observed magnitude range.