Evaluation of Infrasonic Noise Reduction Filters

It is generally accepted that the study of acoustic signals in the atmosphere requires suppression of noise due to atmospheric turbulence. Most acoustic signals of interest arrive with longer spatial wavelengths than the noise that originates near the receiver. As a result, integration of acoustic energy across an area will attenuate spatially de-correlated noise and increase the signal-to-noise ratio. A second approach employs a barrier to reduce wind speeds and small-scale turbulence at the location of the sensor. We are investigating the utility of the various noise-reduction devices that have been proposed or are currently in use. Toward this end, we conduct noise suppression experiments at the Pinon Flat Infrasound test bed in Southern California. This test bed has proved to be ideal for this kind of research as wind speeds vary from near zero to > 15 m/s. The test bed is the site of the 8-element infrasound array IS57 and experiments involving the new fiber optic infrasound sensor. A second goal of our contract is to model the observations and propose remedies for observed shortcomings of the filters. In our first experiment at Pinon, we tested a wind fence against a 30-m-aperture array of microporous hoses. These data were collected concurrently with unsuppressed noise sampled by a reference port. This test revealed that a 50% porous 2-m wind fence coated with a fine wire mesh reduces infrasonic noise above 0.5 Hz by up to 30 dB. The porous wind fence is not effective at longer periods. Our second experiment compared the wind fence with an array of 144 open-air ports distributed across an area 70 m in diameter. This is the design favored by the Provisional Technical Secretariat of the Comprehensive Nuclear-Test-Ban Treaty Organization (PTS-CTBTO) for use at locations where wind speeds are commonly above 3 m/s. In this experiment we also compared these filters with an 18-m-aperture multiport filter (recommended by the PTS for use at relatively calm areas). After > 1 month of recording at wind speeds ranging from near zero to above 15 m/s, we observed significant (15 dB) resonance peaks in the 70m system at 2.65 and 7.95 Hz. These are the first two eigenfrequencies of an organ-pipe mode inside the nonporous pipes that connect the open ports to the microbarometer. The anti-aliasing filter near 10 Hz removes energy at the higher eigenfrequencies, which are predicted by resonance theory. The peaks become evident at wind speeds below 0.5 m/s. The fundamental resonance peak in the 18-m filter is observed at 9.5 Hz, beyond the band of interest to the monitoring community. Our noise experiment has verified the utility of the 70-m filter for suppression of long-period noise. Noise suppression depends on local wind speeds but will reach 15 to 20 dB between 0.1 and 1.0 Hz. Noise suppression is observed from 50 seconds to above 1 Hz. The 18-m filter is superior to the 70-m filter at frequencies above 0.5 Hz. How the mechanical noise suppression devices compare with the new fiber optic sensor and with a sensor buried in a shallow porous medium (as proposed by the Southern Methodist University (SMU) group) has not yet been determined.