Observational Evidence for Wind-Driven Low-Pass Filtering of Infrasound at Short Range

This study presents the first direct observational evidence that tropospheric winds, independent of temperature inversions, impose azimuth-dependent low-pass filtering on infrasound from controlled explosions at short ranges, causing upwind paths to exhibit systematically longer periods and different celerity compared to downwind paths.

The Gist

The Big Idea: The Atmosphere is a Shifty Sound Filter

Imagine you are shouting at a friend across a field. Usually, you expect your voice to travel the same way every time, right? But what if the wind suddenly changed? If the wind is blowing with you, your voice might carry clearly and quickly. If the wind is blowing against you, your voice might get muffled, distorted, or take a weird path.

This paper is about scientists testing exactly that, but with infrasound (very low-frequency sound waves that humans can't hear) generated by large explosions. They wanted to see if the weather could change how these sound waves travel, even over short distances (less than 15 miles).

The Experiment: Two Explosions, Two Different Days

The researchers set up a "sound trap" using 31 microphones spread out in a circle around a test site in New Mexico. They detonated two identical 10-ton chemical explosions (about the size of a small building exploding):

1. May 2024: A windy spring day.
2. October 2024: A calm autumn day.

Because the explosions were identical and the microphones were in the same spots, the scientists expected the sound recordings to look the same both times. They were wrong.

What They Found: The "Split Personality" of the Sound

1. The October (Calm) Day:
On this day, the air was relatively still. The sound waves traveled out in a smooth, predictable circle. No matter which direction the microphone was facing, the sound arrived with the same timing and "shape." It was like dropping a stone in a calm pond; the ripples spread out evenly.

2. The May (Windy) Day:
On this day, there was a strong jet stream of wind blowing from the East. The results were dramatic and split into two distinct groups:

• Downwind (With the wind): The microphones facing the direction the wind was blowing heard the sound exactly as expected—fast and sharp.
• Upwind (Against the wind): The microphones facing into the wind heard something very different. The sound arrived slower and sounded longer and deeper (like a slow, low rumble instead of a sharp crack).

The Mechanism: Wind as a "Low-Pass Filter"

The paper explains this using a concept called low-pass filtering. Think of a sound wave like a complex song with high-pitched notes (short periods) and low-pitched notes (long periods).

• The Headwind Effect: When the sound tried to travel against the strong wind, the wind acted like a sieve or a filter. It pushed the "high-pitched" (short, sharp) parts of the sound wave upward and away from the ground, scattering them into the sky.
• The Result: Only the "low-pitched" (long, deep) parts of the sound managed to stay near the ground and reach the microphones.

The scientists call this wind-driven low-pass filtering. The wind didn't just slow the sound down; it physically removed the high-frequency parts of the explosion's "signature," leaving behind a longer, slower sound.

Why This Matters (According to the Paper)

The paper makes a crucial point: You cannot understand an explosion just by listening to the sound; you must know the weather.

• The Trap: If a scientist hears a long, slow sound, they might think the explosion was huge or happened a long time ago. But in this study, the explosion was exactly the same size as the one in October. The "longer" sound in May was an illusion created by the wind.
• The Lesson: Even at very short distances (just a few kilometers), the atmosphere acts like a dynamic lens. It can bend, focus, or filter sound depending on the wind. To accurately figure out what caused a sound (its size or timing), you need a perfect, real-time map of the wind and temperature at that exact moment.

Summary Analogy

Imagine the explosion is a basketball and the sound waves are the bounces.

• In October (Calm), the ball bounces the same distance every time, no matter which way you throw it.
• In May (Windy), if you throw the ball with the wind, it bounces normally. But if you throw it against the wind, the wind catches the ball, lifts it up, and only lets the heavy, slow parts of the bounce reach the ground. The ball didn't change; the wind changed how the ball behaved.

The Bottom Line: The atmosphere is not just empty space; it is an active participant that can completely rewrite the story of a sound wave, even just a few miles away from the source.

Technical Summary

Technical Summary: Observational Evidence for Wind-Driven Low-Pass Filtering of Infrasound at Short Range

Problem Statement
Accurate characterization of large explosions via infrasound monitoring is fundamentally limited by uncertainties in atmospheric propagation. While the atmosphere is known to act as a dynamic acoustic waveguide, the specific mechanisms by which local atmospheric structure (<50 km) modifies low-frequency acoustic signals remain poorly constrained. Previous research has largely focused on long-range propagation or relied on climatological models, leaving a gap in understanding how local, time-of-event atmospheric conditions—specifically wind fields and turbulence—impact signal spectral content and travel times at proximal ranges. A critical challenge is disentangling source terms from propagation effects to reliably estimate explosion yields, particularly when signal period is used as a proxy for yield.

Methodology
The study leverages two controlled experimental campaigns, "Bouncing Bilby 1" (BB1) and "Bouncing Bilby 2" (BB2), conducted at the New Mexico Institute of Mining and Technology – Energetic Materials Research and Testing Center (EMRTC).

• Source: Both campaigns utilized identical 10-ton TNT-equivalent surface chemical explosions (May 8, 2024, and October 16, 2024).
• Instrumentation: A dense network of 31 single-sensor infrasound stations (Gem microbarometers) was deployed in a consistent geometry around the source, extending from 2.7 km to 22.3 km. The network included two orthogonal azimuthal clusters (~22° and ~253°) designed to test propagation anisotropy.
• Atmospheric Characterization: Three radiosonde soundings were launched during each campaign to provide vertical profiles of temperature, wind, and effective sound speed ($c_{eff}$).
• Data Processing: Raw waveforms were filtered (0.1–45 Hz) to isolate the N-wave. Key metrics extracted included signal period (derived from zero-crossings of the principal lobe) and celerity (apparent horizontal propagation speed, $v_{app} = r/\Delta t$).

Key Results
Despite identical source characteristics, the observed wavefields exhibited a stark seasonal dichotomy:

1. October (Isotropic Baseline): Signals followed a near-unimodal trend in both period and celerity across all azimuths. The lack of strong directional dependence suggested propagation through a relatively quiescent lower atmosphere with weak winds.
2. May (Anisotropic Wavefield): Signals displayed a pronounced azimuthal bifurcation.
   • Downwind (~22°): Signals exhibited faster celerities and shorter periods, closely matching the October baseline.
   • Upwind (~253°): Signals showed systematically slower celerities and progressively longer periods with distance.
3. Atmospheric Drivers: Radiosonde data revealed that in May, a coherent easterly jet (>40 m/s) created strong tailwinds along the ~22° azimuth and opposing headwinds along the ~253° azimuth. In October, winds were weak and disorganized.
4. Mechanism Identification: The study attributes the upwind signal modification to wind-driven low-pass filtering. The strong headwind reduced the effective sound speed ($c_{eff}$), forcing upward refraction of acoustic energy. This refraction induced frequency-dependent attenuation, preferentially removing high-frequency (short-period) content. The study explicitly notes that while low-level temperature inversions were present, they could not explain the directional asymmetry; the effect was driven solely by the wind field.
5. Role of Turbulence: Residual scatter within azimuthal groups was attributed to fine-scale turbulence (shear layers, gravity waves) interacting with the finite-amplitude N-wave, causing additional waveform broadening.

Key Contributions

• Direct Observational Evidence: This study provides the first direct observational evidence that tropospheric winds alone can impose azimuth-dependent low-pass filtering on infrasound at local ranges (≤23 km), independent of strong temperature inversions.
• Local Range Benchmark: It establishes a benchmark dataset demonstrating that atmospheric state specifications are indispensable for interpreting infrasound signals even at distances of only a few kilometers, challenging the assumption that local propagation is isotropic or source-dominated.
• Decoupling Source and Path: By using repeatable sources and dense azimuthal coverage, the study successfully separated source characteristics from path-dependent variability, revealing that signal period is not a universal proxy for yield but is highly state-dependent.

Significance and Claims
The paper claims that the structure of the atmosphere can significantly modify the spectral characteristics of low-frequency acoustic waves at very short ranges. The findings highlight that:

• Yield Estimation Risks: Relying on signal period to estimate explosion yield without accounting for local wind-driven filtering can lead to systematic biases (e.g., overestimating yield due to artificially lengthened periods in upwind directions).
• Modeling Requirements: Predictive propagation models and monitoring practices must incorporate time-of-event atmospheric specifications rather than relying solely on climatological averages or sparse observations.
• Anisotropy Detection: Dense, multi-azimuthal networks are essential to resolve atmospheric propagation effects that would otherwise be misinterpreted as source variability or noise.

The authors conclude that while the shock-to-acoustic transition occurs within ~1 km, the subsequent propagation in the weak-shock to linear regime is highly sensitive to the combined influence of background wind structure and turbulence, necessitating rigorous atmospheric characterization for accurate geophysical monitoring.