Acoustics of infrasound from wind turbines using cross-spectra

Résumé

Measured wind turbine (WT) infrasound often contains mostly wind noise, not just the acoustic WT signature.  With microphones placed on opposite sides of a residence, the random infrasonic wind noise is measured to be incoherent above 0.2 Hz, while the WT signals from the blade-pylon interaction above 0.5 Hz are coherent.  This allows separation of the total true acoustic WT infrasound power (pressure_amplitude²) from the total power.  An estimate of the acoustic outside-to-inside transmissibility function of a residence can also be obtained, if an interior microphone is included.

We show that the averaged cross-spectrum |Gxy| between the two outside microphones removes the dominant pseudo-noise wind-induced power, leaving the true acoustic WT power, which is mainly above 0.5 Hz.  WT acoustic pulses are quite low in pressure amplitude and our data shows that total infrasound power can be up to 20 dB above the power from the nearest WT.  A wind farm of 100 units increases the total acoustic power by 6-10 dB, so much of the measured signal must be wind noise.  This paper discusses microphone arrangements and processing, with some examples, to separate the true acoustic WT signal power from random wind noise.