Detecting infrasound, the pressure waves below about 20 Hz that fall beneath human hearing, requires specialized sensors, careful wind-noise management, and array-based methods that can pick faint signals out of a noisy atmosphere. The basic detection chain involves microbarometers or low-frequency measurement microphones arranged in spatial arrays, combined with signal-processing techniques that extract coherent wave arrivals from background noise. Analyzing those signals then means linking measured properties like frequency, amplitude, and arrival direction to specific sources, whether volcanic eruptions, severe storms, or rocket launches. The challenge is less about building a single sensitive sensor and more about suppressing the environmental noise that can easily overwhelm the signal you are after.
Why Infrasound Is Difficult to Measure
The biggest obstacle to reliable infrasound detection is not sensitivity but wind. At frequencies below 20 Hz, turbulent air pressure fluctuations caused by even moderate wind create noise levels that dwarf the infrasound signals you want to record. A field study comparing measurement microphones and microbarometers near wind power plants found that wind-induced noise was the dominant concern in every real-world measurement scenario, regardless of which sensor type was used.1SAGE Journals (Journal of Low Frequency Noise, Vibration and Active Control). Performance comparison of measurement microphones and microbarometers for sound pressure measurements near wind power plants This means that choosing the right sensor is only part of the problem. Without a strategy for reducing wind noise at the measurement site, even high-end equipment will record mostly turbulence rather than meaningful infrasound.
A second difficulty is that infrasound signals from distant sources arrive after traveling through an atmosphere whose temperature, wind speed, and layered structure all change with season, time of day, and weather patterns. These variations bend, reflect, and attenuate the waves in ways that can make a signal appear to come from a slightly different direction than the true source, or can cause it to vanish entirely during unfavorable atmospheric conditions. Both the detection step and the interpretation step need to account for this atmospheric complexity.
Choosing the Right Sensor
Two families of sensors dominate infrasound work. Microbarometers measure absolute or differential atmospheric pressure changes at very low frequencies, typically from a fraction of a hertz up to several hertz, with high sensitivity and good long-term stability. They are the standard instruments at the International Monitoring System (IMS) infrasound stations operated worldwide. Measurement microphones designed for low-frequency acoustics can also reach into the infrasound range, though their frequency response, self-noise, and susceptibility to wind differ from microbarometers.
In direct field comparisons, both sensor types can capture infrasound, but their behavior in wind diverges. The study near wind power plants tested the two systems simultaneously across multiple sites and then followed up with controlled experiments in an aeroacoustic wind tunnel to isolate each sensor’s wind sensitivity.1SAGE Journals (Journal of Low Frequency Noise, Vibration and Active Control). Performance comparison of measurement microphones and microbarometers for sound pressure measurements near wind power plants The practical takeaway is that whichever sensor you choose, you need to understand its wind-noise characteristics and build quality-control criteria around them. A reading that looks like a genuine infrasound signal can easily be turbulence unless you verify it against wind conditions at the time of recording.
For anyone setting up a detection system from scratch, microbarometers are the more proven choice for continuous monitoring at sub-hertz frequencies. Measurement microphones can work well for higher-frequency infrasound and situations where you also need standard audio-range measurements, but they generally require more careful wind shielding in outdoor environments.
Wind-Noise Reduction in the Field
Because wind is the primary enemy of clean infrasound data, the field has developed several physical noise-reduction methods. The two main approaches are spatial filters and wind barriers, and they work on different principles.
Spatial filters, often called “rosettes,” consist of arrays of inlet ports connected to a central sensor by pipe networks that spread out over a large area, sometimes tens of meters across. The idea is that turbulent pressure fluctuations, which are spatially incoherent over distances larger than their scale, cancel out when averaged across many ports, while the infrasound signal, which is coherent across the whole array, passes through. These rosette systems are used at many IMS stations and are effective, but they require a significant amount of flat, unobstructed land to install.
Wind barriers offer an alternative for sites where space is limited. Experimental work on a two-meter-high, 50%-porous hexagonal barrier coated with fine wire mesh showed that wind speed inside the barrier dropped by about 90% compared to ambient conditions. Above a certain corner frequency, the barrier reduced infrasonic noise by up to 20 to 25 dB. Below that corner frequency, the barrier still provided a small reduction of roughly 4 dB, whereas rosette spatial filters showed no reduction at all below their own corner frequency.2PubMed. Infrasonic wind-noise reduction by barriers and spatial filters The corner frequency itself depends on the size of the device and the wind speed, so a larger barrier or rosette pushes the useful range to lower frequencies.
In practice, many monitoring stations combine both methods. A barrier around the sensor reduces local turbulence, while a pipe-rosette network provides additional spatial averaging. For a smaller-scale or DIY setup, even a modest wind screen around the microbarometer inlet can make a noticeable difference at the higher end of the infrasound band.
Array Detection and Locating the Source
A single infrasound sensor can tell you that pressure is fluctuating, but it cannot reliably tell you where the signal came from or whether the fluctuation is a coherent wave or just noise. This is why infrasound detection almost always relies on arrays: groups of sensors spaced apart by distances ranging from tens of meters to a few kilometers. By comparing the arrival times of a coherent wave across the array elements, you can determine the direction the wave came from (back-azimuth) and the speed at which it crossed the array (trace velocity). These two parameters together help identify and locate sources.
The workhorse algorithm for this is the Progressive Multichannel Correlation (PMCC) method, which scans the array data for time windows and frequency bands where signals are correlated across multiple sensors. Researchers analyzing infrasound from the 2016 Gyeongju earthquake in South Korea applied PMCC across seven arrays at distances from roughly 180 to 470 km from the epicenter and were able to identify infrasound sources not only near the epicenter but also in non-epicentral regions where ground shaking coupled into the atmosphere.3Geophysical Journal International. Observations and seismoacoustic simulations of earthquake-generated infrasound waves in non-epicentral regions By combining back-azimuth measurements with models of how fast infrasound travels through the atmosphere (celerity models), they could map the wave back to its origin.
For anyone working with array data, the key outputs at this stage are detection bulletins: lists of time windows where a coherent signal was found, along with the estimated back-azimuth, trace velocity, and dominant frequency. These bulletins are the raw material for all subsequent analysis.
Signal Processing Techniques
Once you have a detection, the next step is characterizing the signal. Standard spectral analysis using Fourier transforms can reveal the dominant frequencies and their amplitudes, but infrasound signals are often non-stationary, meaning their frequency content changes over time. A volcanic eruption might produce a signal that shifts in frequency as the eruption intensity changes, and a bolide’s sonic boom has a very different time-frequency signature than the sustained hum of ocean microbaroms.
Wavelet-based methods handle this non-stationarity well. Research on infrasound from wind turbines demonstrated that the synchrosqueezed wavelet transform is effective for detecting and isolating the dominant frequency components that carry the most energy in an infrasound recording.4PubMed. Application of wavelet synchrosqueezed transforms to the analysis of infrasound signals generated by wind turbines Unlike a simple spectrogram, this technique sharpens the time-frequency representation, making it easier to distinguish closely spaced frequency peaks and to track how individual components evolve. This is useful for any source that produces tonal or quasi-periodic infrasound, whether it is a wind turbine, a volcanic tremor, or an industrial facility.
Beyond wavelets, cross-correlation and beamforming across the array remain essential. Beamforming steers the array response toward different directions and trace velocities, producing maps that show where energy is concentrated in azimuth-velocity space. When combined with time-frequency analysis, you get a detailed picture of what arrived, when, from where, and at what frequency, which is usually enough to start identifying the source.
How the Atmosphere Shapes What You Can Detect
Infrasound does not travel in a straight line from source to receiver. It refracts through the atmosphere, bending back toward the ground when it encounters layers where the effective sound speed increases. The stratosphere, roughly 30 to 60 km above the surface, is the most important layer for long-range infrasound propagation. Winds and temperature gradients in the stratosphere create waveguides that efficiently duct infrasound energy back down to the ground, sometimes allowing detection at distances of thousands of kilometers.5Pure and Applied Geophysics. The State of the Stratosphere Throughout the Seasons: How Well Can Atmospheric Models Explain Infrasound Observations at Regional Distances?
The direction of favorable ducting depends heavily on the season. In the Northern Hemisphere winter, strong westerly stratospheric winds favor propagation from west to east, meaning a station east of a source is more likely to detect it. In summer, the winds reverse and favor eastward-to-westward propagation. This seasonal pattern is so pronounced that it determines whether entire categories of signals are detectable at a given station during a given month. Researchers studying how well atmospheric models predict real infrasound observations found that the stratospheric state is the primary driver of detection capability at regional distances.5Pure and Applied Geophysics. The State of the Stratosphere Throughout the Seasons: How Well Can Atmospheric Models Explain Infrasound Observations at Regional Distances?
Closer to the ground, temperature inversions and low-level wind jets can create additional waveguides that trap infrasound in the lower atmosphere.6INTER-NOISE and NOISE-CON Congress and Conference Proceedings. Efficient modeling of atmospheric infrasound propagation with the Semi-Analytical-Finite-Element (SAFE) method These tropospheric ducts are less predictable than the stratospheric ones but can explain detections that atmospheric models based on stratospheric winds alone would miss. For anyone interpreting infrasound data, the practical lesson is that you always need an atmospheric model, or at least knowledge of the current wind and temperature profile, to make sense of what you are seeing. A missing signal does not necessarily mean the source was silent; it may mean the atmosphere was not cooperating.
Natural Sources and What Their Signals Reveal
One of the most rewarding aspects of infrasound monitoring is the sheer variety of natural phenomena that produce detectable signals. Each source type has a characteristic frequency range and signal pattern that, with experience, becomes recognizable in the data.
Volcanoes
Volcanic eruptions are among the strongest and most scientifically useful infrasound sources. During the June 2021 paroxysmal eruptions of Mt. Etna, researchers used infrasound recordings to calculate the flow velocity of material exiting the vent, obtaining values between 50 and 125 meters per second. They then fed those velocities into a plume model to estimate how high the eruption column rose.7PubMed Central. Assessment of eruption source parameters using infrasound and plume modelling: a case study from the 2021 eruption of Mt. Etna, Italy This kind of near-real-time eruption characterization from infrasound alone is valuable for aviation safety and civil protection, especially when visibility is poor or the volcano is remote.
Severe Storms and Tornadoes
Tornado-producing storms emit infrasound in the 0.5 to 10 Hz range, and these signals may appear up to two hours before a tornado forms on the ground. Because atmospheric attenuation at these frequencies is very low, the signals can travel hundreds of kilometers from the storm.8The Journal of the Acoustical Society of America. Comparison of infrasound emissions observed during a tornado with potential fluid mechanisms The exact mechanism producing the infrasound is still debated; the once-popular idea of a radially vibrating vortex has been shown to be non-physical, and researchers continue exploring alternative fluid-dynamic explanations. Still, the observational pattern is robust enough that infrasound has been investigated as a complementary tornado warning tool.
Bolides and Meteors
When a large meteor enters the atmosphere, it produces a shock wave that registers as infrasound at stations worldwide. A global study using over 70 bolides detected simultaneously by satellites and recorded as more than 140 infrasonic waveforms established a quantitative relationship between the observed infrasonic period and the energy yield of the event. That period-yield relationship turned out to be remarkably similar to the one originally derived from nuclear test data.9Journal of Atmospheric and Solar-Terrestrial Physics. Infrasound production by bolides: A global statistical study This means infrasound can independently estimate the energy of a bolide entry even when optical observations are unavailable, which matters for planetary defense monitoring.
Ocean Microbaroms
The most persistent natural infrasound signal on Earth comes from the ocean. Microbaroms, peaking at around 0.2 Hz, are generated by the nonlinear interaction of wind-driven ocean waves. They dominate the coherent ambient noise background at infrasound stations worldwide.10Geophysical Research Letters. Global Microbarom Patterns: A First Confirmation of the Theory for Source and Propagation For signal analysts, microbaroms are both a nuisance and a tool. They are noise you must account for when looking for transient signals, but their strength and directional patterns serve as a continuous probe of the atmosphere. If your microbarom detections match predictions from ocean-wave models and atmospheric wind models, your detection system is working correctly. Discrepancies may point to errors in the atmospheric model rather than in your instruments.
Earthquakes
Large earthquakes generate infrasound through multiple mechanisms. Ground shaking can couple directly into the atmosphere, producing what are called ground-coupled air waves. In addition, seismic waves traveling through the earth can excite secondary infrasound sources when they reach mountain ranges, coastlines, or other topographic features. Analysis of infrasound from the 2001 Arequipa earthquake in Peru, a magnitude 8.4 event, revealed both types of signal. The ground-coupled air waves at the IS08 station in Bolivia allowed researchers to track the earthquake’s rupture propagation from northwest to southeast along the fault at about 3.3 km per second, while additional infrasound arrivals were attributed to secondary sources along high mountain ranges that diffracted acoustic energy toward the station.11Geophysical Research Letters. Ground‐coupled air waves and diffracted infrasound from the Arequipa earthquake of June 23, 2001
Anthropogenic Sources
Human-made infrasound is abundant and, in many cases, useful for calibrating and validating detection networks. Rocket launches are a particularly well-characterized source. A comprehensive study collected ground-truth information for 1,001 rocket launches from 27 spaceports worldwide between 2009 and mid-2020 and was able to identify infrasound signatures from up to 73% of those launches on IMS stations.12Geophysical Research Letters. 1001 Rocket Launches for Space Missions and Their Infrasonic Signature Because the time, location, and type of each launch are known precisely, these events serve as ground-truth references for testing atmospheric propagation models and assessing station performance. If your station reliably picks up known rocket launches, you can be more confident in its ability to detect unknown events.
Other anthropogenic infrasound sources include industrial facilities, wind turbines, mining blasts, and military ordnance. Wind turbines, in particular, produce quasi-periodic infrasound tied to blade-pass frequency, which makes them useful for testing time-frequency analysis methods but can also be a persistent noise source for nearby monitoring stations.
Animal Infrasound and Bioacoustic Monitoring
Not all infrasound comes from geophysical or industrial sources. Elephants communicate using calls that extend well below 20 Hz, and these vocalizations can travel long distances. Research into elephant low-frequency communication has examined how the lungs, larynx, and vocal tract generate the calls, how the composition of expired air and ambient temperature affect the signal, and how the elephant ear is adapted to detect low-frequency sound and localize its source.13PubMed. Long-distance, low-frequency elephant communication For conservation biologists, detecting and analyzing elephant infrasound with arrays similar to those used in geophysics offers a way to monitor herds remotely, track movement, and detect distress calls associated with poaching.
Other animals known to produce or respond to infrasound include whales, whose low-frequency calls propagate across ocean basins, and certain birds that may use infrasound for navigation. The detection principles are the same as for geophysical infrasound: sensitive sensors, wind-noise management, and array processing to separate the biological signal from the ambient background.
Infrasound and Human Health
Public interest in infrasound detection sometimes stems from health concerns rather than geophysical curiosity. People living near wind farms, industrial sites, or transportation corridors sometimes report symptoms they attribute to low-frequency noise. A narrative review of the biological and medical literature on infrasound found that the evidence shows effects from controlled exposures in laboratory settings and animal models, but also identified significant gaps in the mechanistic understanding of how infrasound affects human physiology.14PubMed Central. Infrasound in Biology and Medicine: Insights into Mechanisms, Health Outcomes and Research Perspectives – A Narrative Review The review characterized infrasound’s biological effects as “dualistic,” meaning that at certain frequencies and intensities, infrasound may have therapeutic applications, while at others it can cause discomfort or physiological changes.
For someone trying to measure infrasound at home to investigate a suspected noise problem, the same principles apply on a smaller scale. A microbarometer or low-frequency microphone with a verified flat response below 20 Hz is essential; standard sound-level meters and smartphone apps typically roll off well above the infrasound range and will miss the frequencies of concern. Wind shielding is still critical, even indoors, because air currents from HVAC systems can produce pressure fluctuations that mimic infrasound signals. And recording for extended periods with timestamped data lets you correlate measured levels with the times when symptoms occur, which is far more useful than a single snapshot measurement.
Practical Steps for Setting Up a Detection System
If you are building or configuring an infrasound detection setup, whether for research, monitoring, or personal investigation, the workflow follows a consistent pattern regardless of scale.
- Select a sensor: A microbarometer is ideal for sub-hertz work; a calibrated low-frequency microphone works for frequencies closer to the audible boundary. Whichever you choose, verify its frequency response and self-noise specifications against the signals you expect to encounter.
- Manage wind noise: At minimum, use a windscreen over the sensor inlet. For outdoor stations, a porous wind barrier or pipe-rosette system dramatically improves data quality. Even a simple foam-covered enclosure helps indoors.
- Use multiple sensors: If budget and space allow, deploy at least three sensors in a triangular array spaced tens to hundreds of meters apart. This enables back-azimuth and trace-velocity estimation, which is essential for identifying sources and rejecting noise.
- Record continuously: Infrasound events are often transient and unpredictable. Continuous recording with GPS-synchronized timestamps ensures you do not miss events and allows post-hoc correlation with known sources like launches, earthquakes, or weather systems.
- Apply array processing: Run cross-correlation or beamforming algorithms on the multi-sensor data to extract coherent signals. Even basic cross-correlation across two channels can separate a real wave from uncorrelated turbulence.
- Use time-frequency analysis: Spectrograms, wavelet transforms, or synchrosqueezed transforms reveal how signal content evolves and help you classify the source.
- Incorporate atmospheric context: Check wind and temperature profiles for the time of your recording. Stratospheric wind models from agencies like ECMWF or GFS are freely available and tell you which propagation paths were favorable.
A well-run station produces data that can contribute to everything from volcano monitoring to climate research to noise-impact assessment. The global IMS network demonstrates what is possible at large scale, but even a single carefully sited array with good wind-noise management can produce scientifically meaningful infrasound observations.