Effects of Processing and Ambient Noise Distribution on Seismic Interferometry using Auto-Correlation Stacks

The influence of processing and ambient noise distribution on seismic interferometry using auto-correlation stacks is investigated. We first study the effects of automatic gain control (AGC) and spectral whitening on the daily and monthly auto-correlation stacks. It was found that although the AGC equalizes the amplitudes over longer time lag windows, it does not dramatically affect the results over time lag windows shorter than the AGC window applied. The effects of spectral whitening are more pronounced where the daily and monthly auto-correlation stacks for the PmP window become less coherent when spectral whitening is not applied. The influence of band-pass filtering is next investigated, where the results are robust to the effects of the high cut filter applied from 0.55 to 0.85 Hz. However, the results are less robust if the low-cut filter is set lower than about 0.3 Hz. Similar to not applying spectral whitening, this could result from lower frequency microseisms contaminating the body waves on the auto-correlation stacks. Synthetic modeling was then performed to investigate the effects of the ambient noise distribution on the auto-correlation stacks. For sources along the top boundary, it was found that the zone of stationary phase of the auto-correlation stacks was very localized to energy coming from nearby the receiver with energy from sources at greater distances destructively interfering. For noise energy arriving from below the model, such as far from regional and teleseismic sources, it was found that the angular zone of stationary phase was much broader. Assuming that the time duration for the temporal stacking is sufficient to allow for a smoothly varying range of incident arriving noise energy, then a broad region of stationary phase could explain the observation of both PmP and SmS phases on the auto-correlation stacks.