Optimizing the empirical Gaussian filter parameter in the multiple filter technique to accurately determine surface-wave group velocities

The multiple filter technique (MFT) is widely employed to measure surface wave group velocities in surface wave and ambient noise tomography which are used to image deep crustal and upper mantle structures. The accuracy of the MFT is highly dependent on the rational selection of the Gaussian filter parameter \alpha . The conventional \alpha selection scheme only considers the epicentral distance (Δ), and ignores the influence of the central period ( T_n ), which causes non-negligible measurement errors, particularly at long periods. In this study, we optimized the \alpha selection scheme by integrating both Δ and T_n , which was then validated using synthetic and observed Rayleigh wave data. The results indicate that \alpha must be greater than the attenuation coefficient \beta to ensure valid filter windows in the MFT. For a fixed T_n , the optimal \alpha increased with increasing Δ. In addition, reliable long-period (>30 s) group velocities cannot be obtained when Δ≤1000 km. For epicentral distances of 2000–8000 km, the optimal \alpha (4–30 s) was substantially larger than that for long periods (31–100 s). We determined that 30 s was the most reasonable period division, which was shorter than the division used in previous studies. Compared with the conventional \alpha selection scheme, the optimized empirical \alpha values reduced measurement errors in the group velocity dispersion curves and improved the accuracies of surface wave and ambient noise tomography, thus providing a more reliable observational basis for imaging deep crustal and upper mantle structures and constraining lithospheric tectonic evolution and geodynamic processes.