Page 16 - January 2006

atmospheric absorption. At the same time, the atmospheric pressure decreases exponentially with a halving distance of about 5 km. As the sound waves reach higher altitudes, the ratio of the sound pressure, Δp, to the ambient pressure, p, increases and non-linear effects become more important. One such effect is the creation of harmonics or a shift of energy from low frequency to high frequency. Since high frequencies are absorbed more rapidly, this has the effect of distorting waveforms and increasing overall attenuation of sound.
These combined effects make propagation predictions quite complex. Add in the effect of ground bounces and tur- bulence and the overall physics is challenging. The individ- ual elements of propagation have been studied for decades and are well understood. Recently there have been quite successful attempts to combine all the effects that impact signal amplitudes and frequency content into a single algo- rithm. Although that effort is a work in progress, the soft- ware package InfraMAP32 does a reasonable job including all relevant physics as well as modern meteorological mod- els. The ability to correct for atmospheric effects has now reached the point where researchers feel confident using measurements of received signals at different directions and distances from a source to infer the conditions of the inter- vening atmosphere using acoustic tomography. This ability promises to provide a tool to continuously measure the wind speeds of the upper atmosphere in near real time glob- ally.
Limitations imposed by wind noise
Wind is the dominant source of noise recorded by
infrasound stations and can readily overwhelm signals of
33
interest . A quick glance at a raw infrasound record clearly
reveals the strong diurnal wind pattern that characterizes most locations on the Earth (Fig. 5). The large pressure fluctuations created by the wind obscure the weak signals from distant infrasound sources. This is especially striking
given that, where possible, most infrasound stations have intentionally been sited in low wind locations. In fact, a recent study of the wind speeds observed at infrasound sta- tions showed that most stations have mean annual wind speeds of roughly 3 m/s or less (a “light breeze” on the Beaufort scale).
Many strategies have been explored to reduce the impact of wind-generated infrasound noise. Pipe-rosettes, as described above and by References 2 and 34 are perhaps the most common form of wind filter used today. A variant uses porous hose, rather than non-porous pipe, to construct the rosette. Regardless of the materials used, these strategies all attempt to exploit the incoherence of wind-induced pressure fluctuations over length scales of tens of meters. Another effective wind mitigation strategy is the wind fence35 that sur- rounds a microbarometer in much the same way a foam windscreen surrounds a microphone.
New instruments and approaches are also being devel- oped to provide alternatives to mechanical wind suppression techniques. For example, a distributed sensor36 consisting of numerous individual microphones allows signals to be summed electronically rather than mechanically (such as in the summing manifolds of pipe rosettes). The clear advan- tage of this approach is that these data can be summed adap- tively to optimally reduce the noise and preserve the signal. The recently developed optical fiber infrasound sensor37 is another approach to electronic, and thus instantaneous, averaging of wind noise. In this instrument optical fibers wrapped around a sealed, compliant tube measure the total deformation of the tube, sensing the coherent deformations of the tube induced by acoustic waves and averaging out the incoherent deformations along the length of the tube that are due to wind fluctuations. Several optical fiber infra- sound sensors (OFIS) distributed in a fan configuration have been used to characterize the signal as well as remove inco- herent noise.
Current research on the physical mechanisms responsi-
Fig. 5. A one month time series showing root-mean-square (RMS) infrasound variations (bottom) and RMS wind speed (top) for October 2004 for an infrasound station near Lake Titicaca in Bolivia. A clear diurnal variation in the wind speed, peaking at about 5-6 m/s, is obvious and is strongly correlated to the pressure variations.
14 Acoustics Today, January 2006