Non-parametric spectral modelling of source parameters, path attenuation and site effects from broad-band waveforms of the Alborz earthquakes (2005-2017)

S-wave spectral amplitudes from 312 crustal earthquakes recorded at the Iranian National Broadband Seismic Network in the Alborz region between 2005 and 2017 are analysed in order to evaluate earthquake source parameters, path attenuation and site amplification functions using the non-parametric generalized inversion technique (GIT). We exploit a total number of 1117 seismograms with ML 3-5.6 in the frequency range 0.3-20 Hz. The evaluated non-parametric attenuation functions decay uniformly with distance for the entire frequency range and the estimated S-wave quality factor shows low Q values with relatively strong frequency dependence. We assume the omega-square source model to retrieve earthquake source parameters from the inverted source spectra. The obtained stress drops range from 0.02 to 16 MPa with a mean value of 1.1 MPa. Stress drop and radiated energy show fairly self-similar scaling with seismic moment over the available magnitude range; however, the magnitude range of this study is too narrow to draw a definite conclusion on source scaling characteristics. The obtained moment magnitude M-w and the local magnitude ML are linearly correlated and approximately equivalent in the range of M-w 3-4. For larger events, Mw generally underestimates ML by about 0.1-0.5 magnitude units. The estimated site amplification functions for horizontal component (GIT H) are nearly flat with no obvious pre-dominant frequency peaks for most stations, as expected for the sites of permanent broad-band seismic stations located on rock, though a few stations show amplification peaks from 1 to 8 Hz, with a maximum amplification of about a factor of 7 with respect to the reference site. The evaluated site responses for the vertical components present remarkable amplification or deamplification, leading to differences of the H/V amplitude levels in comparison with the GIT H amplification curves. The results of this study provide a valuable basis for predicting appropriate ground motions in a context of seismic hazard assessment.