Mechanism-dependent scaling of GNSS peak ground displacement for rapid earthquake magnitude estimation

Peak ground displacement (PGD) derived from high-rate Global Navigation Satellite System (GNSS) observations has become a key metric for rapid moment magnitude estimation in earthquake early warning systems. Existing empirical PGD magnitude scaling relationships, however, differ in datasets and processing approaches, and the influence of focal mechanisms remains insufficiently quantified, limiting their robustness and general applicability. Here, we systematically evaluate four widely used PGD scaling relationships based on a global database containing high-frequency GNSS displacement records from 45 earthquakes (M_W6.0-9.1). We assess the impact of focal mechanisms (normal, strike-slip, and reverse faulting) on magnitude estimation and develop refined, mechanism-dependent scaling relationships. The four conventional models exhibit substantial variability, with root mean square errors (RMSEs) of 0.34-0.48, reflecting differences in datasets and modeling assumptions. Incorporating focal mechanism information reduces the overall RMSE to 0.31, highlighting the first-order control of source characteristics on PGD scaling behavior. Independent validation using 12 global earthquakes from 2004 to 2024 demonstrates further improvement, with an overall RMSE of 0.22 and errors of 0.46, 0.14, and 0.12 for normal, strike-slip, and reverse events, respectively. These findings demonstrate that faulting mechanism exerts a significant control on PGD scaling, and that accounting for mechanism dependence significantly improves the accuracy and robustness of GNSS-based magnitude estimation for earthquake early warning applications.