1. Introduction

During an intense evening thunderstorm in Trussville, Alabama on 22 July 2008, two nearby lightning flashes were captured on digital video and audio by the lead author. Although the direct lightning strokes were not visible to the camera, the associated flashes of light were recorded. The audio signals of the thunder were recorded by the camera approximately 2.3 and 1.1 s, respectively, after each flash. However, “click” sounds were heard by the author and recorded by the camera, almost simultaneously with the lightning flashes. Therefore, it may not be attributed to the propagation of sound waves directly from the lightning flash. The sound is likely due to the propagation of electromagnetic waves from the lightning strike causing a vibration in, or a coronal discharge from, an object very near the camera.

In this note, background information on the electromagnetic pulse (EMP) produced by lightning will be presented in section 2. Data and methodology are examined in section 3. The lightning strikes and the associated sounds on 22 July 2008 are considered in section 4, using spectrogram analysis of the sound, and still images from the video. Section 5 contains a discussion of the data and conclusions.

2. Background

Audible sounds, occurring simultaneously with various phenomena producing electromagnetic radiation, have been reported extensively in scientific literature. Sounds occurring simultaneously with meteors entering the earth’s atmosphere were reported over 200 yr ago (Blagdon 1784), and have been discussed since by many others (e.g., Udden 1917; Romig and Lamar 1963; Keay 1980; Price and Blum 2000). Sounds have also been associated with the aurora (e.g., Silverman and Tuan 1973) and lightning (Romig 1963).

In the cases of meteors, the aurora, and lightning at some distance from the observer, sound waves propagating directly from the associated phenomena cannot explain the observed almost simultaneous sounds, since the speed of sound in the atmosphere is only about 340 m s⁻¹. For example, the light from a typical aurora is produced near a height of 100 km MSL (Stern and Peredo 2001), and any sound waves generated from that height would take almost 300 s to reach the earth’s surface. The instantaneous nature of the sound implies that it must be associated with electromagnetic waves generated by the aurora, traveling at or near the speed of light. Of course, one cannot hear electromagnetic waves, only the effects the waves may have on nearby objects (Hrynyshyn 2001). Romig and Lamar (1963) suggested that sound may be produced by coronal discharges of electricity from conductors (even plant leaves) subjected to a large potential gradient produced by a rapidly changing electric field associated with electromagnetic waves. The basic physics behind this theory is also supported by Silverman and Tuan (1973), and Vaivads (2002) points out that other sharp protruding objects, including trees and hair, may produce such a discharge.

The range of sound frequencies audible to humans is generally in the range of 20 Hz–20 kHz. Keay (1980) proposed that an electrostatic field varying at audible frequencies may cause vibrations in metal or dielectric objects, producing sound energy. This transfer of electromagnetic energy to acoustic energy is known as transduction. Keay then performed numerous experiments on this transduction. He found that human hair vibrated at audible frequencies and produced sound energy when exposed to electromagnetic radiation, and that ordinary objects may vibrate in the presence of electromagnetic radiation, producing sound near the same frequency as the radiation (Keay and Ostwald 1991; Vinkovic et al. 2002; Hrynyshyn 2001). Many objects in the typical outdoor environment may act as transducers, vibrating because of electromagnetic energy and producing sound as a result. These may include pine needles, dry vegetation, and metallic objects (e.g., Hrynyshyn 2001).

The return stroke of a typical cloud-to-ground (CG) lightning lasts only about 100 μs, and is associated with electric currents as high as 200 kA (Barrington-Leigh 2000; Uman 1987; Carpenter and Tu 1997). The current in a lightning return stroke changes very rapidly, possibly on the order of 10¹¹ A s⁻¹ (Krider 1986). Rapidly changing electric currents produce electromagnetic radiation. Therefore, lightning flashes produce an electromagnetic pulse of energy. The pulse typically lasts less than 1 ms (Price and Blum 2000; see Fig. 1), radiates a large part of its energy in the ELF and VLF frequency range, mainly from 1 to 20 kHz (e.g., Uman 1987; Price and Blum 2000; Bianchi and Meloni 2007), and has a peak in its frequency spectrum around 5 kHz (see Fig. 2). This frequency range is within the range of audible sound frequencies. Therefore, since objects may vibrate and produce sound near the frequency of the electromagnetic radiation they are exposed to, the electromagnetic pulse from lightning may, upon interaction with an object, produce a vibration in that object and the resulting audible sound. The object acts as the transducer, converting electromagnetic energy into sound energy. Coronal discharges, such as those proposed by Romig and Lamar (1963), may also produce sounds associated with lightning. Sound occurring simultaneously with (or even prior to) nearby lightning flashes, before the thunder reaches the observer, has been documented (e.g., Corliss 1989), and observed by many.

3. Data and methodology

The primary sources of data are the audio and video recorded by a Panasonic SDR-H18 digital video camera. The video data, taken at a sample rate of 30 Hz, produce one digital image every 33 ms. The timestamp for each digital image was assumed to be in the center of the 33-ms interval it corresponded to. The light intensity, as it varied with time in the video, was determined by using “RGB Analysis of Image Colors” software (Byers 2006) to find the average value of each of the three RGB (red, green, blue) components in each image, and then averaging those numbers to produce a normalized intensity [i.e., I = (R + G + B)/3]. Brighter images correspond to higher values of the normalized intensity.

The audio data, at an original sample rate of 48 kHz, was analyzed using spectrogram software. Spectrographs were produced, illustrating the intensity of sound energy as it varied with time and frequency. Numerical files were output, using a fast Fourier transform, that contain the sound level (in dB Hz⁻¹, with a frequency resolution 46.9 Hz) at each 11.1-ms time step. These sound levels were converted to sound power at each frequency range and time, allowing for computation of the total sound intensity (in dB) in any given frequency interval. To isolate the “click” sound produced by the lightning EMP, the sound intensity at frequencies greater than 5 kHz was calculated. To determine the time of the thunder, the total sound intensity at all frequencies between 0 and 12 kHz was calculated. Radar data from the Weather Surveillance Radar-1988 Doppler (WSR-88D) at Birmingham, Alabama (KBMX), was examined. Also, radio frequency (RF) source data from the National Aeronautics and Space Administration (NASA) North Alabama Lightning Mapping Array (LMA; Koshak et al. 2004) was used. Data from the Vaisala National Lightning Detection Network (NLDN) was also used to verify the times and locations of the lightning strikes.

4. Results and analysis

An intense thunderstorm was centered about 5 km northeast of the lead author’s location in Trussville, Alabama, around 0430 UTC 22 July 2008, as shown on radar (see Fig. 3). A digital video camera that also records audio was pointed eastward. The storm produced a large amount of lightning, as indicated by the NASA North Alabama LMA (Fig. 4). Over 2400 RF sources were recorded between 0400 and 0500 UTC in a 16 km × 16 km horizontal square centered near Trussville. The NLDN indicates that 35 CG lightning strikes occurred within 10 km of the lead author’s location in the 10-min period ending at 0435 UTC. At 0431:43 UTC, the first lightning flash being examined herein, (flash A) occurred, and the second one (flash B) occurred at 0433:30 UTC. Almost simultaneous with the visible flash of each lightning strike, and well before the thunder, the lead author heard at least one “clicking” sound. Analysis of the video and audio recorded by the camera confirms these sounds. (The video files may be viewed online at http://vortex.nsstc.uah.edu/~coleman/Thunder.)

a. Flash A

Flash A occurred at approximately 0431:43 UTC. There was a 2.3-s delay between the lightning flash and the onset of thunder. The speed of sound was calculated using (adapted from Nave 2006):

View Expanded

where c_a is the speed of sound in air, γ = c_p/c_υ = 1.4 (where c_p and c_υ are the specific heats of air at constant pressure and constant volume, respectively), R = 8314 J kmol⁻¹ K⁻¹ is the universal gas constant, T_υ is the virtual temperature of the air, and M = 29 kg kmol⁻¹ is the molecular weight of dry air. The surface observation at 0430 UTC from nearby Birmingham, Alabama (KBHM) indicated a temperature of 305 K and a dewpoint of 291 K, implying a water vapor mixing ratio r_υ = 0.013 (at p = 1000 hPa). Houze (1993) states that, to an approximation,

View Expanded

where T is the air temperature. So, in this case, T_υ = 307.4 K, and using (1), c_a = 351 m s⁻¹. Combining the speed of sound with the 2.3-s delay, it is estimated that the closest part of the lightning flash was about 807 m away from the camera.

A sequence of images taken by the video camera across a 2-s time period beginning at 0431:42 UTC are shown in Fig. 5. Note that not all images are shown; many of the images without any lightning flash were removed, while some of these were shown for comparison. Figure 5 shows three distinct flashes of light associated with flash A. The first, a rather weak one, occurs in frame number 29 (at t = 950 ms), followed by two brighter flashes in frame 39 (t = 1283 ms) and in frame 43 (t = 1417 ms), apparently indicating an initial CG stroke followed by two return strokes.

The spectrograph of the sound associated with a 5-s video, beginning at 0431:42 UTC (so that t = 0 is consistent with the images shown in Fig. 5) is shown in Fig. 6. Note that there is a substantial amount of background noise below 6400 Hz, caused by katydids and other miscellaneous noise not associated with the lightning or the thunder. There is a fairly constant, narrowband acoustic source near a frequency of 2500 Hz, apparently associated with crickets. Also note the three short bursts of sound energy, across a wide range of frequencies (consistent with clicks), at t = 1200, 1380, and 1530 ms. These are associated with the clicks heard by the author almost simultaneous with the light from the lightning flashes. It is also interesting to note that the sound associated with the clicks tends to decrease most quickly at higher frequencies, with lower-frequency sound lasting slightly longer. The much more intense lower-frequency sound energy beginning around t = 2550 ms is associated with the author’s voice on the video, and the thunder begins around t = 3500 ms.

In Fig. 7, the intensity of the light in each frame of the 5-s video (as described in section 3) is plotted along with the intensity of sound energy with frequency between 5 and 12 kHz. This frequency range helps to eliminate most of the background noise, yet captures most of the click sound energy. The first stroke of light associated with flash A occurs around t = 950 ms, while the first click, indicated by an increase in sound level in the band 5 kHz < ν < 12 kHz from around −45 to −31 dB, begins at t = 1210 ms, only 260 ms after the first stroke. The second stroke of light begins at t = 1249 ms, and the third begins at t = 1415 ms. Additional clicks were associated with each of these strokes, with the second beginning at t = 1388 ms, and the third beginning at t = 1543 ms. Therefore, the delays between the flashes of light (associated with the strokes) and the audible clicks for the second and third strokes are 139 and 128 ms, respectively.

Since the delay between the first stroke of light and the thunder is 2300 ms, the increases in sound level within 260 ms of the light flashes cannot be associated with thunder, or the propagation of sound energy directly from the lightning strike to the camera. However, the pattern of 3 clicks, associated with sound level increases of up to 14 dB, closely matching the pattern of 3 flashes of light, strongly indicates that the 2 sets of phenomena are correlated. Therefore, the audible clicks must be associated with the lightning EMP, specifically with sound transduction in nearby objects or coronal discharge caused by the EMP, as discussed in section 2. The time delays between light strokes and audible clicks of 260, 139, and 128 ms, indicate that the sources of the sound energy are only 45–91 m away from the camera.

Since the lightning flash was determined to be about 807 m away from the camera, but the first audible click occurred only 260 ms later, the direct propagation speed of energy from the lightning flash to the camera would have had to be at least c = 3000 m s⁻¹, which is almost an order of magnitude faster than the speed of sound. The supersonic shock wave associated with the thermal expansion of air near the lightning strike, which causes the thunder, only extends outward about 10 m from a typical lightning strike; beyond 10 m the energy is transmitted as a sound wave (Vavrek et al. 2008). Therefore, direct propagation of sound energy from the lightning strike to the camera cannot explain the audible click.

5. Discussion and conclusions

Sounds associated with natural sources of electromagnetic radiation, especially meteors entering the earth’s atmosphere and aurora, have been discussed extensively in the literature (e.g., Blagdon 1784; Udden 1917; Romig and Lamar 1963; Silverman and Tuan 1973; Keay 1980; Price and Blum 2000). Sounds associated with the electromagnetic radiation emitted by lightning flashes have received limited attention (e.g., Romig 1963; Corliss 1989), although they are surely observed by scientists and laymen. However, the physical processes used to explain the sounds associated with the electromagnetic radiation associated with meteors and aurora may be logically extended to sounds associated with the electromagnetic pulse emitted by lightning. As discussed in section 2, electromagnetic waves do not produce sound directly, but the effects the waves may have on nearby objects can produce sound. Some have proposed that sound may be produced by coronal discharges from conductors in the outdoor environment, including plant leaves and trees, due to the presence of electromagnetic radiation. Others have shown that ordinary objects, including pine needles, dry vegetation, and metallic objects, may act as transducers, vibrating because of the presence of electromagnetic radiation and producing sound frequencies near those of the radiation.

For the cases examined in section 4, a fairly unique dataset was collected in the vicinity of two close lightning strikes. The data from a video camera, including both video, allowing for light intensity measurements, and audio, allowing for audio measurements of any audible click sounds as well as thunder, were analyzed in detail. Two separate lightning flash events on 22 July 2008 near Trussville, Alabama, nearby lightning flashes (both within 1 km of the camera) produced audible clicks, observed by the author to occur almost simultaneously with the lightning flashes. Video and audio data confirmed that the audible clicks did indeed occur almost simultaneously with the lightning flashes, far too quickly to be associated with thunder.

In the first case, the delay between the first flash of light and the thunder was measured as 2300 ms, but the audible clicks occurred within 260 ms or less of the light flashes. Also, there was a pattern of three clicks, resembling the time pattern of three flashes of light associated with the lightning. In the second case, the delay between the primary lightning flash and the thunder was 1100 ms, while the audible click occurred only 84 ms after the flash. It was determined that the audible clicks must be associated with the electromagnetic pulse emitted by the lightning, which would travel at the speed of light and reach the camera and its surroundings almost instantaneously, as compared to the thunder, which traveled at the speed of sound.

The time delays between light flashes and audible clicks ranged from 84 to 260 ms, indicating that the sources of the sound energy were 29–91 m away from the camera, while the lightning flashes were 386–807 m from the camera. There were numerous objects within 30 m of the camera that could have caused coronal discharge and/or transduction of electromagnetic energy into sound energy, including trees, dry vegetation, and metallic objects such as automobiles, gutters, and fences. Interestingly, as noted in section 4 for both flashes, the sound level associated with the clicks decreased more rapidly at higher frequencies than at lower frequencies (see Fig. 8). In flash B, sound energy near 8 kHz lingered for over 300 ms (see Fig. 8), while sound energy near 15 kHz lingered only about 125 ms. This behavior is representative of the echo of the audible click produced by the lightning, as higher frequencies are attenuated more (Stokes 1845) and decay more quickly.

This study provides analysis of an interesting phenomenon, the production of kinetic energy in the form of vibration and sound, in objects exposed to electromagnetic energy. The production of sound by the electromagnetic pulse associated with lightning has been widely observed. However, this study successfully demonstrates and quantifies, through the use of video and audio data, two cases of the lightning EMP causing sound energy. The sound was created either through coronal discharge, or more likely through the transduction of electromagnetic energy at audible frequencies to vibrations at those same frequencies in objects.