عنوان مقاله [English]
Horizontal to vertical spectral ratio technique on single station ambient noise data, is a well-known technique in study of site effect. Recently, this technique is introduced as a tool for identification of shear wave velocity profile of soil beside its normal usage.Many studies in recent years showed that the ellipticity of the fundamental mode of Rayleigh waves can be obtained by reducing the Love and Body waves effects from the H/V spectral ratio. Based on the relation of the H/V curves and the ellipticity of Rayleigh waves and dependency of ellipticity tothe shear-wave, this method can retrieve the S-velocity structure in a thick alluvial deposit. In this paper, HVTFA and RayDEC methods are used to retrieve the ellipticity curves for more than 140 single-station ambient noise measurements. The HVTFA technique based on time-frequency analysis with Continuous Wavelet Transform tries to reduce the SH-wave influence that is possible by identifying P-SV wavelets along the signal and computing the spectral ratio from these wavelets. It is assumed that the energetic points in time-frequency representation of the vertical signal is related to a single Rayleigh wave wavelet. The average over all wavelets defines as ellipticity. Based on random decrement technique, the RayDEC method uses the vertical component as a master trigger and stacks a large number of horizontal and vertical signals from three-component records of seismic noise to obtain ellipticity curves. The right flank of ellipticity curves (from the first peak of curves to the next trough) were used in inversion, because numerical studies show that the right flank is the most reliable part of ellipticity, and the energy of the Rayleigh-wave fundamental mode strongly dominates in these frequency ranges. In the following, ellipticity curves were classified based on the f0 peaks and the right flanks in two ways; visual observation of similarities and k-means clustering statistical approach.Inversions process performed using the Neighborhood Algorithm based on the partition of the parameter space into Voronoi cells. The Voronoi decomposition of the parameter space is the base of an approximation of the misfit function, which is progressively refined during the inversion. The method uses prior information (initial parameterizations) and try to optimize the computation at the different stages of inversion. The results of inversion show the existence of the thick alluvial deposits in the northern and eastern parts of the city. For the southern parts, the method shows higher velocity and lower depth of bedrock. These results are in agreement with geological situation of the region, existence of mountainous area at the southern and western parts, and extensive alluvial plains at northern and eastern parts.