Analysis of Anthropogenic and Natural Noise from Multilevel Borehole Seismometers in an Urban Environment, Auckland, New Zealand

Two short-period seismometers were permanently installed at depths of 26 and 383 m beneath the Eden Park stadium in central Auckland in October 2008 and incorporated into the Auckland Volcano Seismic Network in 2011. These borehole seismometers were temporarily augmented by a surface sensor to characterize the site response at this location. Despite the borehole installations, seismic monitoring is challenging in this urban environment due to high anthropogenic noise that superimposes the Earth's signal. We analyze the power spectral density of continuous noise records over long and short time periods to quantify the reduction in noise with depth and the effect of temporal noise variations on the detection capabilities for earthquakes and volcanic tremor. We identify natural and anthropogenic noise sources that temporarily elevate noise levels by 10-15 dB using records from the Rugby World Cup matches held at the Eden Park stadium and by comparing records of daytime versus nighttime periods and windy versus calm days. Characterization of these noise sources shows that the frequency ranges of traffic and train noise at this site overlap (1-35 Hz and 8-35 Hz, respectively), however they exhibit distinct maxima at peak frequencies of 7 and 26 Hz. Modeling of the structure beneath the stadium shows that the noise spectrum generated by the nearby train excites frequencies within the topmost Waitemata sequences that get efficiently trapped in the low-velocity waveguide beneath the similar to 20 m thick basalt layer at the surface. Although the shallow borehole sensor generally shows a noise improvement of <= 5 dB in comparison to the surface sensor, it exhibits higher noise levels in the 8-35 Hz frequency range due to the trapped noise waves. Wind causes increased microseismic noise at 0.1-1 Hz that is coherent at all depth levels and shows the most pronounced time variation.