Current Journal of Applied Science and Technology

This study evaluates the performance of two Ground Motion Prediction Equations (GMPEs) for inslab and interplate earthquakes when applied to the Trans-Mexican Volcanic Belt (TMVB) in central Mexico, addressing the lack of validation of forearc-calibrated models in this backarc region. A retrospective analysis was conducted using instrumental records from six moderate-to-large earthquakes (M_W 6.0-7.7) that occurred between 2011 and 2022, independent of the original GMPEs calibrations. Records were obtained from permanent seismological networks. For each event, observed Peak Ground Acceleration (PGA_obs) was computed as the quadratic mean of the two horizontal components and compared with the PGA predicted by the corresponding GMPE (PGA_pred). Spatial contour maps of the PGA_obs/PGA_pred ratio were generated using minimum-curvature interpolation. Stations with critical underestimation were analyzed using the Horizontal-to-Vertical Spectral Ratio (HVSR) technique on earthquake records. The results show that, for interplate earthquakes, the corresponding GMPE systematically underestimates PGA with a median PGA_obs/PGA_pred ratio of 1.73 (95% bootstrap confidence interval: 1.26–2.20; based on 145 records), whereas for inslab events the GMPE overestimates PGA with a median ratio of 0.69 (95% CI: 0.55–0.84; 160 records). Extreme local ratios occur at a few stations, where HVSR analyses reveal amplification factors of 4-6 at frequencies >10 Hz, which contribute to but do not fully explain the observed discrepancies. These findings underscore the need for caution when applying forearc-calibrated GMPEs in the TMVB and have direct implications for seismic hazard assessment and engineering practice in central Mexico.