Comparing Active and Passive Noise Cancellation Techniques for Wind Filtering in Infrasound Detection Applications

Comparing Active and Passive Noise Cancellation Techniques for Wind Filtering in Infrasound Detection Applications

Hakan Saplakoglu, Zachary Clayton, Chengzhi Wu Li, Michael Zink

Paros Center of Atmospheric Research

University of Massachusetts

Amherst, MA 01003

hsaplakoglu|zclayton|cwuli|mzink@umass.edu

Infrasound signals, defined as those below 20 Hz, are prevalent in both natural environments and man-made infrastructure. For example, infrasound signals are used to detect earth quakes, tornadoes, waves from big ocean stroms, and wake turbulence from airplanes. These signals can be detected, traced, and analyzed using infrasound barometers or microphones. In our research, we employ highly accurate (0.08 hPa) barometers to measure infrasound. Often, these signals are weak due to originating from distant sources, but the range of the barometers (620 hPa to 1100 hPa) allows for their detection. However, even minimal external noise can easily obscure these signals. The most significant source of noise in outdoor sensing environments is wind, a non-stationary noise source.

In this paper, we investigate techniques for canceling wind noise to preserve infrasound data that would otherwise be lost to interference. We examine two approaches for this purpose: active noise cancellation and passive noise cancellation.

For active noise cancellation, an ultrasonic anemometer is used to measure real-time wind speed at the same location where the infrasound barometer collects data. It has the advantage that the physical setup of this approach is rather simple and has only a small footprint. The disadvantage in comparison to passive noise cancellation is the requirement for complex post-data-collection processing algorithms. Our setup was utilized to compare various adaptive filtering techniques and evaluate their effectiveness. This is the first such approach that has been applied to infrasound measurements.

The evaluation of our measurements indicate a correlation between the noise signal from the anemometer and the barometer signal, with the latter containing both noise and the desired infrasound data. Our results demonstrate that this method effectively reduces wind noise in the barometer signal. However, we also found that it does not fully succeed in recovering lost infrasound signals. We plan to further investigate the reasons behind this limitation with the goal to improve the active noise cancellation approach for wind filtering.

Concerning passive noise cancellation, many different approaches can be applied to reduce wind noise. These include a wind barrier, or a static port to filter out dynamic air pressure. In our experiments, a large porous hose filter, also called a “spatial filter”, was directly connected to the pressure port of an infrasound barometer. This method does not require any post-data-collection calculation, as the data measured by the barometer is already filtered. The filters are connected to the pressure port of the barometer, where they then expand out in a varying number of “legs”, and lengths of those legs. While this is a well-known method for wind filtering, its large footprint (a spatial filter setup can require 700 - 1000 m2 of open space) makes its usage often impractical, especially in cities and urban areas. In addition, spatial filters require higher maintenance than the anemometer we used for the active noise cancellation approach.

We tested various configurations of these filters to evaluate their effectiveness, and demonstrated that this method is a successful way of reducing wind noise and recovering lost signals. We will continue to explore multiple new configurations to determine the most effective form of this filter.

We will present a comprehensive comparison of active and passive noise cancellation techniques for infrasound detection, highlighting the challenges involved in this area. Although both approaches yielded promising results, further research is necessary. This is especially true for active noise cancellation, which, despite its small footprint and low-maintenance advantages, currently underperforms compared to passive noise cancellation.

In future work, we will evaluate if combining active and passive noise cancellation will lead to further improvements.