On the Diffusivity of Noise Data Recorded (2-8 Hz) at Stations Located in North-Western Iran

Since the advent of seismic interferometry, the cross-correlation of received random noises has been very frequently used for the approximate assessment of the empirical Green’s function between the station pairs. Theoretically, the diffusivity of noise wavefield isthe key factor that contributes to the success of this idea in real applications. Diffusivity itself requires the fulfillment of two conditions of energy equipartitioning as well as a homogeneous distribution of the incoming noise. To meet these requirements, the attempts are made to select a longer study period to homogenize the incoming azimuths. This solution is mostly logical in the range of 0.1-0.3 Hz, because it is believed that in this frequency range, microseismic noise waves are received continuously by the stations. But at higher frequencies, seismic sources are mostly changing both spatially and temporally, and their spectral content is not stationary over time. Therefore, in high-frequency seismic interferometry, prolonging the study interval will not help much in improving the signal-to-noise ratio of the retrieved Green’s functions. For this reason, the main focus in such studies is on highly heterogeneous regions because multiple scattering in these regions may be able to azimuthally homogenize the wave field to some extent. According to the scientific studies, however, the homogeneity of the azimuthal distribution of noise is very unlikely to happen, even in the most heterogeneous areas. So to know which claim is more correct, we turned our attention to the northwestern region of Iran, where the severe heterogeneity of the medium for frequencies above 1 Hz has already been confirmed by recent studies. With the help of a Frequency-Dependent Polarization Analysis (FDPA), we first extracted those elliptically polarized waves whose degree of polarization (IDOP) was greater than 0.8. This constraint assures us that the received waves have maintained their coherency for a relatively long time before reaching the station. Our analysis on the azimuths of these polarized data fail to pass Kuaper and  tests, indicating that the azimuth distribution of the received signals at the stations located in north-western Iran is azimuthally inhomogeneous. At the stations near the topography, even strong directivity in coherent seismic noises was observed.