Jafar Shoja-Taheri, Saeid Naserieh, and Hadi Ghofrani
M _l and M _w scales in the Iranian Plateau based on the strong-motion records
Bulletin of the Seismological Society of America (April 2007), 97(2):661-669

Abstract:
The availability of a large amount of strong-motion data recorded by the National Strong Motion Network of Iran (NSMNI) has motivated this study to develop relations for routine determination of M _L and M _W from digital horizontal components of the strong-motion records. The dataset comprises 861 two-component horizontal acceleration time series recorded for 125 earthquakes with magnitudes of 4.5 and larger. The M _L scale is based on the horizontal synthesized Wood-Anderson seismograms. We have applied the Monte Carlo technique to evaluate distance correction curves for use in determining the local magnitude, M _L , in Iran and in its northern, eastern, and Zagros subregions. Results indicate that the distance correction curves show trilinear behavior for geometrical spreading. The resulting coefficients evaluated for Iran as a whole are: R ₁ = 96+ or -5 km; R ₂ = 131+ or -5 km; n ₁ = 1.01+ or -0.02; n ₂ = -0.14+ or -0.1; n ₃ = 0.14+ or -0.03; k = 0.00020+ or -0.00008. The distances less than R ₁ correspond to attenuation of the direct waves. Between R ₁ and R ₂ is the distance where the multiply reflected and refracted shear waves from Moho dominate the arrivals. n ₁ , n ₂ , and n ₃ are the coefficients of geometrical spreading for distances from the source to R ₁ , R ₁ to R ₂ , and beyond R ₂ . k is the coefficient of inelastic attenuation. For estimating the M _W scale from the strong-motion data, we used the method proposed by Andrews (1986). To find the best correlation between the moment magnitudes measured from the strong-motion data and those measured from teleseismic data, we examined several time windows (e.g., whole trace, S-wave coda, and source time durations). The regressions show that the M _W estimates from different time windows are all equally well correlated with the corresponding reported values with nearly identical standard deviations. Finally, relations between the estimates of local and moment magnitudes for the regions show that for earthquakes with magnitudes larger than about 6.0, the M _L scale gradually becomes saturated and, therefore, it gives smaller values than those obtained by the M _W scale. However, for smaller earthquakes, the M _L scale overestimates the M _W scale. This discrepancy occurs mainly because the frequency contents of the waveforms employed in these scales are different.