Acoustic Observations of the OSIRIS‐REx Sample Return Capsule Re‐Entry from Wendover Airport Open Access

Origins, Spectral Interpretation, Resource Identification, and Security‐Regolith Explorer (OSIRIS‐REx) spacecraft was launched in September 2016 and returned its sample return capsule (SRC) to Earth with samples from asteroid Bennu on 24 September 2023 (Ajluni et al., 2015; Beshore et al., 2015; Lauretta et al., 2017). The capsule re‐entered the Earth’s atmosphere at hypersonic speeds over the Pacific Ocean and eventually landed in the Utah Test and Training Range (UTTR). Although the primary mission was to collect the asteroid sample, the SRC re‐entry also provides the unique opportunity of studying the acoustic signature of the shock waves produced by the re‐entry (Revelle and Edwards, 2007; Yamamoto et al., 2011; Sansom et al., 2022). SRC acoustic signatures are comparable to those produced by meteoroids entering and traversing through the Earth’s atmosphere (Silber et al., 2018). Because meteoroids are unplanned natural events, comprehensive observational campaigns are difficult to plan (Silber et al., 2023). Thus, scheduled SRC re‐entries, such as the OSIRIS‐REx SRC, are great surrogates for the study of extraterrestrial objects entering Earth’s atmosphere. A collaborative effort involving 16 institutions and over 400 ground‐based sensors was executed to observe the signals generated by the re‐entry of the OSIRIS‐REx SRC (Silber et al., 2024). This article focuses on analysis of the ground‐based acoustic measurements acquired at Wendover Airport (United States) during the OSIRIS‐REx SRC re‐entry, the location where similar measurements were acquired during the Genesis and Stardust SRC re‐entries.

The Earth is bombarded with $∼105$ tons of extraterrestrial material annually (Plane, 2012). Although most particles are small and burn up in the Earth’s atmosphere, in rare instances they are large enough to land on the ground as meteorites. These larger objects could pose a threat on Earth or mimic signatures of other sources (Yoon et al., 2016; Pilger et al., 2021). These hypersonic entries can produce cylindrical shock waves (Revelle, 1976), and the high‐pressure air at the leading edge can cause fragmentation events (Ceplecha and Revelle, 2005; Trigo‐Rodríguez et al., 2021) that can produce quasi‐spherical shock waves (Ens et al., 2012; Silber et al., 2018; Revelle, 2009). These shock waves decay to low‐frequency sound (i.e., infrasound [< 20 Hz], Evans et al., 1972) that can be detected at large distances (Wilson et al., 2023, 2025). Although SRCs typically do not have planned fragmentation events, they do make excellent case studies for acoustic signatures produced from the shock wave generated by the re‐entry. However, OSIRIS‐REx marked only the fifth SRC re‐entry from interplanetary space since the conclusion of the Apollo missions (Silber et al., 2023).

The other four SRC re‐entries (Genesis, Stardust, Hayabusa 1, and Hayabusa 2) also returned to Earth with extraterrestrial samples. The capsules entered the atmosphere at shallow angles (≤12°) at hypersonic speeds, which was then slowed within the atmosphere prior to landing. Like OSIRIS‐REx, both Genesis and Stardust SRCs landed in the UTTR and were detected by an array of low‐frequency acoustic sensors (Chaparral Physics) deployed at Wendover Airport (ReVelle et al., 2005; Revelle and Edwards, 2007). The Genesis SRC collected samples of solar wind particles (Asmar et al., 2019), re‐entered the atmosphere on 8 September 2004, and landed with a speed of 86 m/s due to a failed main parachute (Jones, 2004). The Stardust SRC returned to Earth on 15 January 2006 after collecting comet samples (Brownlee et al., 2003). Stardust SRC was nearly identical to the OSIRIS‐REx SRC (Lauretta et al., 2017; Francis et al., 2024) but was smaller in size and mass relative to Genesis. The detected signature from the shock wave at Wendover Airport for both SRCs were N‐waves (Dumond et al., 1946), consistent with that of a ballistic shock (Silber et al., 2023). A more comprehensive discussion of these and the Hayabusa re‐entries (Yamamoto et al., 2011; Nishikawa et al., 2022; Sansom et al., 2022), specifically how they relate to OSIRIS‐REx, can be found in Silber et al. (2023).

This article reports ground‐based acoustic measurements during the OSIRIS‐REx SRC re‐entry using a variety of acoustic sensors. Chaparral Physics sensors, similar to those used during Genesis and Stardust, serve as the high‐fidelity reference measurements that other, more economical sensors are compared against. In addition, the high‐fidelity results are compared with the Genesis and Stardust SRC re‐entries due to the similarities between these events, including being recorded with the same type of sensors from nearly the same location. The remainder of the article includes an overview of the OSIRIS‐REx mission including specific details of the re‐entry and the broader geophysical observation campaign in the OSIRIS‐REx Overview section. The Methods section provides details of the methods used for these specific ground measurements at Wendover Airport. The Results and Discussion section presents the results and discussion of the acoustic data with particular focus on analysis of the primary N‐wave arrival. Finally, the article concludes in the Conclusions section with a summary of the primary conclusions drawn from the current analyses.

The OSIRIS‐REx SRC, which had a mass of ∼46 kg and a diameter of 0.81 m, entered the Earth’s atmosphere at a shallow angle of ∼8° (Francis et al., 2024). It entered the atmosphere at hypersonic speeds (roughly 12 km/s) at an altitude of 125 km over the Pacific Ocean just above San Francisco, California, at ∼14:42 UTC. Because it slowed down traversing through the atmosphere, the drogue parachute was supposed to open prior to the main parachute at ∼30 km altitude; however, it was not able to deploy due to faulty wiring (Silber et al., 2024). The main parachute was able to slow the capsule sufficiently to facilitate a relatively soft landing (i.e., impact speed was 5 m/s) at 14:52 UTC, a minute earlier than planned (Silber et al., 2024). The nominal capsule trajectory is shown in Figure 1 along with markers indicating Wendover Airport and the UTTR landing site. Because of the lack of telemetry during re‐entry, the exact SRC trajectory is unknown but that shown is an approximation based on Francis et al. (2024).

Approximately 80 investigators from 16 different institutions deployed a total of over 400 sensors, making it the most instrumented re‐entry in history (Silber et al., 2024). Most of the institutions deployed their sensors in or near Eureka, Nevada. Eureka was selected primarily because it was near the projected point of peak heating, which also corresponds to the point of greatest energy deposition. The current work focuses on ground‐based acoustic sensors deployed at Wendover Airport, which was located ∼215 km northeast of Eureka near the Utah/Nevada, United States, state line. The Wendover area was selected because of ease of access (i.e., roads and towns are sparse in this region). It would also be unique relative to the plethora of Eureka measurements, and based on acoustic propagation modeling there was a high probability of detection. Furthermore, Wendover Airport was also the location where an acoustic array was deployed during the first and second seismoacoustically measured SRC (i.e., Genesis and Stardust).

There were four different low‐frequency acoustic sensors deployed at Wendover Airport (see Fig. 2): (1) Chaparral Physics Model 64S, (2) Gem Infrasound loggers, (3) WERD ISSM23, and (4) Samsung Galaxy S23 running RedVox (Garcés et al., 2022). At least three of each type of sensor were deployed at the airport and arranged in triangular patterns to facilitate array‐based processing. Only the Chaparral Physics sensors had a single data acquisition system that recorded all their signals with the same clock. The other sensors had individual data acquisition systems with independent time stamps used for array processing. The Chaparral Physics array served as the high‐fidelity measurements that the more economical sensors are compared against. Consequently, each economical sensor had one sensor collocated with the center Chaparral Physics sensor for direct comparisons.

The Chaparral Physics array was composed of four Chaparral Physics Model 64S acoustic sensors (Merchant, 2022). These microphones have a flat frequency response between −3 dB points at 0.01 and 245 Hz. Each sensor had a dynamic range of 118 dB and a nominal sensitivity of 0.08 V/Pa at 1 Hz, though sensor‐specific calibrations were used for analysis. Each microphone was housed within a weatherproof case (1300 Case, Pelican), as shown in Figure 2a. All the array sensors were hardwired to a data acquisition device (PGS‐140 4‐channel, Pegasus) and sampled at the maximum sample rate of 1000 Hz. The array was powered by a 12 V marine deep cycle battery, which typically powered the array for one week without recharging. The battery and data acquisition device were housed within a separate weatherproof case (Storm Case iM2450, Pelican). See the supplemental material, available to this article, for a table summarizing the sensor dimensions, mass, and the mass of the accessories.

The Gem Infrasound Logger (Gem), shown in Figure 2b, was designed to be a small but powerful low‐frequency acoustic sensor at relatively low cost to enable broad use of acoustic sensing as well as deployment flexibility due to the integrated data acquisition. Consequently, it has seen widespread adoption with uses including monitoring of volcanoes (Rosenblatt et al., 2022), the atmosphere (Averbuch et al., 2022), and earthquakes (Scamfer and Anderson, 2023), as well as use on high‐altitude balloons (Brissaud et al., 2021). Each Gem was composed of a printed circuit board (PCB), an Arduino, a thermistor, a Global Positioning System (GPS) unit, and a secure digital (SD) card for data storage. These components facilitate onboard data acquisition, which allows the sensor to function as a solo unit. To record sound, the Gem uses a differential pressure transducer with a pneumatic filter on one side of the transducer diaphragm. The low‐pass filter causes a temporal delay in equalizing the ambient pressure, and the difference between the filtered side and the unfiltered side produces a differential pressure associated with the surrounding higher frequency acoustic waves (Anderson et al., 2018; Lindsey, 2023). Gems have a nominal frequency response of 0.05–25 Hz. They have an internal sample rate of 400 Hz, but an internal filter was applied prior to writing the data to memory at 100 Hz. In addition, the GPS location is sampled every 15 min. The manufacturer specified pressure resolution as 0.015 Pa and a GPS location resolution of 1.8 m. The differential pressure sensor itself has a sensitivity of $517μV/Pa$⁠. The Gems were powered by a 9 V battery and each housed within a compact acoustic windscreen box (Swaim et al., 2023).

Seven WERD sensors (Infrasonic Sensor Module, ISSM23) were arranged in two triangular arrays. The WERD sensors, shown in Figure 2c, had features similar to that of the Gem sensors. Like the Gem sensors, the ISSM23 can be deployed as a standalone sensor or in an array. The WERD sensors were housed within a manufacturer supplied weatherproof enclosure for outdoor operation within adverse weather conditions. The sensor is composed of an OrangePi 3 long‐term support (LTS) single‐board computer running Linux, a custom PCB, and a GPS unit. The two WERD arrays were powered by a 12 V battery each, in which a step‐down voltage converter was used to supply 5 V to each sensor through a single USB‐C connector. The acoustic signals were recorded using a differential pressure transducer in a configuration similar to that used for Gem sensors. The frequency response of the WERD sensors were nominally 0.1–200 Hz. The sample rate was set at 400 Hz (i.e., the maximum rate for ISSM23).

RedVox is a smartphone application for Android or iOS initially developed by researchers at the University of Hawaii and fully transitioned to industry in November 2023 (Garcés et al., 2022). It uses built‐in (or external) smartphone microphones to measure low‐frequency acoustic signals (Asmar et al., 2019; Liu et al., 2024). The application, which allows real‐time plotting, aims to enable mapping of the infrasound field all over the world. They can be deployed as arrays in which it uses time synchronization for precise timing. The application performance is limited to the specifications of the microphone. For the current study, four Samsung Galaxy S23s operating the RedVox application were collocated with each of the Chaparral sensors and data were recorded at 800 Hz. Each smartphone had a plastic cover over it for protection as shown in Figure 2d. Subsequently, these phones operating the RedVox application are referenced as RedVox sensors.

The sensors were deployed near the northeast corner of Wendover Airport in Wendover, Utah, as shown in Figure 2e. The minimum perpendicular distance from the nominal trajectory to the sensors was ∼70 km. Three of the Chaparral sensors were deployed in an equilateral triangle with a 112 m aperture, whereas the fourth was placed in the center of the triangle as illustrated in Figure 2f. Table 1 lists the coordinates of the Chaparral Physics sensors along with the other sensors deployed at the airport. The four RedVox sensors (R1–R4) were collocated with the four Chaparral Physics sensors. The Gem (G1–G3) and WERD (W1–W7) sensors were arranged in three separate triangular configurations (see the supplemental material). The three Gem sensors had an aperture of 49 m, an average spacing between sensors of 44 m, and the westernmost Gem sensor (G3) was collocated with the center Chaparral Physics sensor (C1). Four of the WERD sensors (W1–W4) had an aperture of 51 m, average sensor spacing of 34 m, and the center sensor (W1) collocated with C1. The other three WERD sensors (W5–W7) had an aperture of 51 m, an averaging spacing of 42 m, and were arranged around the southwest Chaparral Physics sensor (C4). The sensors were deployed ∼18 hr prior to the capsule re‐entry and collected data until ∼3 hr after the event. To minimize noise contamination, the team monitored the re‐entry from a separate location.

On the morning of the OSIRIS‐REx SRC re‐entry, at ∼14:47 UTC (08:47 MDT local time), the deployment team heard a sonic boom at their viewing location that was ∼5 km west of Wendover Airport. The sound heard was described as not being too loud but sufficient to easily distinguish it above the background noise. No secondary arrival of the sonic boom was heard by the team, and there was no visual confirmation of the capsule re‐entry. This was expected due to the re‐entry occurring during daytime and the deployment area being relatively far from the nominal trajectory and landing site. The wind was calm (<0.3 m/s) during the re‐entry, but the wind got stronger later in the morning. All sensors were recovered three hours after the event. The following section compares the acoustic observations using different sensors as well as providing some comparisons to the previous Genesis and Stardust SRC re‐entries.

The four‐element array of Chaparral sensors will serve as the high‐fidelity reference for comparisons. The time traces for each of these sensors are shown in Figure 3, which each has a clear N‐wave immediately followed by a broadband coda. The peak‐to‐peak amplitude between the four sensors had an average of 4.07 ± 0.17 Pa. The N‐wave signal period from its initial rise (first zero‐crossing) until the third zero‐crossing (full cycle) had an average duration of 0.196 ± 0.005 s. These are similar in form to the N‐wave signals observed during the Genesis and Stardust SRC re‐entries (ReVelle et al., 2005; Revelle and Edwards, 2007), which were measured at Wendover Airport. However, the amplitudes (⁠$pp2p$⁠) are different for each re‐entry; the peak‐to‐peak amplitude (⁠$pp2p$⁠) for each re‐entry is listed in Table 2. Although Stardust had similar mass and diameter to OSISIS‐REx, Genesis was more massive and had a larger diameter (see Silber et al., 2023, for a more detailed comparison). All three SRC re‐entries had similar entries (entry speeds and angles), but they each had different trajectories that varied their positions relative to Wendover airport (see Fig. 1). Thus, the larger amplitude during the Genesis re‐entry is consistent with it being larger and having a trajectory that passed closer to Wendover Airport. However, the weaker signal from Stardust relative to OSIRIS‐REx highlights that the acoustic signature is sensitive to the local flight dynamics.

In Figure 4a, the power spectral density (PSD) from the Chaparral array center sensor (C1) during the N‐wave arrival is plotted as well as the 2 hr before and after the N‐wave arrival for reference. Note that the other sensors are not shown due to redundancy. Furthermore, this location had a collocated sensor from each sensor type. Each PSD was produced using the Welch method (Welch, 1967) with 100 s windows and 88.9% overlap on a 200 s segment centered on 14:47:16 UTC. A band‐pass filter was applied from 0.01 to 500 Hz. The background spectra were significantly more elevated after the event relative to the background before the entry, which is attributed to increasing wind speed. The entry occurred in the morning when the wind speeds were very low (<0.3 m/s). Later in the morning, as the temperature increased, the wind speed had a corresponding increase (∼2.7 m/s) that was consistent with the increased background noise. During the entry, there is a broadband peak with the maximum at 4.98 ± 0.148 Hz and spectral ripples (or spectral leakage) at higher frequencies (up to 50 Hz). The ∼5 Hz broadband peak with the spectral ripples was not observed during the 2 hr before or after the N‐wave arrival, which is evidence that it was associated with the dominant N‐wave signal.

The current PSD from the OSIRIS‐REx SRC re-entry is compared with the PSDs from the Genesis (ReVelle et al., 2005) and Stardust (Revelle and Edwards, 2007) SRC re‐entries in Figure 4b. All three recordings were acquired at Wendover Airport and produced a broadband peak with spectral ripples. Note that units were not provided for Stardust (Revelle and Edwards, 2007), so the Stardust amplitude cannot be compared. The fundamental peak frequency (⁠$f0$⁠) for Stardust and OSIRIS‐REx are comparable when Genesis is lower. The fundamental peak frequencies as well as the associated temporal duration for each N‐wave signal for all three SRC re‐entries are included in Table 2.

Beamforming using the Chaparral array was performed using frequency–wavenumber (f‐k) analysis combined with time difference of arrival to identify periods of high coherence between the microphones and the corresponding direction of arrival (Rost and Thomas, 2002). Because the array had a 112 m aperture and a 1000 Hz sample rate, the angular resolution of the array was ∼0.2°. The f‐k analysis was performed using 0.5 s windows with 25% overlap with an angle resolution of 0.5°; these results along with the corresponding back‐azimuth angle (BAZ) are provided in Figure 5a. The largest coherence (i.e., F‐ratio from analysis of variance, Melton and Bailey, 1957) coincides with the N‐wave arrival and is marked in Figure 5a with a star. The corresponding BAZ of this arrival was 167.5° measured clockwise relative to north (i.e., from the southeast). This direction is nominally perpendicular to the SRC trajectory, in which cylindrical shock waves are known to have a perpendicular propagation from their source (Wilson et al., 2025).

The f‐k analysis also identified secondary arrivals of high coherence within the acoustic coda. A list of these strong coherent signals within the coda (i.e., points marked with red dots in Fig. 5a) is provided in Table 3. The first strong peak in coherence (i.e., large F‐ratio) occurs 10.0 s after the primary N‐wave arrival. This secondary arrival had a BAZ of 338°. Given that this BAZ is nearly 180° opposite the N‐wave arrival, this was suspected to be a reflection of the primary N‐wave. Using the local speed of sound (339 m/s), the 10 s delay corresponds to a travel distance of 3.4 km with the reflecting object being half of that distance (1.7 km). A map showing the topography of the area surrounding the Chaparral array is provided in Figure 5b, which includes arrows indicating direction of arrival for signals (red indicates primary N‐wave; green indicates secondary signals listed in Table 3), as well as dashed circles based on half the travel time (i.e., assuming reflections) for the strong secondary arrivals. This shows that the arrival at 10 s was likely due to a reflection from the mountain range to the northwest of the deployment location. This conclusion is supported by previous studies in which low‐frequency sounds reflect due to topography (Albert and Bowman, 2019; Bird et al., 2022; Khodr et al., 2022). In addition, both Genesis (Revelle et al., 2005) and Stardust (Revelle and Edwards, 2007) re‐entries detected reflections from the mountains surrounding Wendover Airport. Similarly, the arrivals at dt ≈ 25 s with a BAZ of 313° was also likely due to reflections from the nearby mountain range. Conversely, the arrivals at dt ≈ 21 s with a BAZ of 185° has no obvious topographical feature (e.g., mountain range) apart from an old abandoned military facility in that area. Thus, it is unclear whether this was a reflection or a nondirect propagation path.

The Gem time traces (see the supplemental material) showed N‐wave arrivals with a smaller time difference relative to the Chaparral sensors due to the smaller aperture of the Gem array. The Gem signals appear to have less fluctuations within the N‐wave relative to the Chaparral sensors, which is due in part to the lower sample rate of the Gems (Gem: 100 Hz; Chaparral Physics: 1000 Hz). The average peak‐to‐peak amplitude from the Gem sensors was 3.66 ± 0.04 Pa. Although the Gem sensors record at 100 Hz, multiple filters are applied to the signal resulting in a roll off above ∼20 Hz (Anderson et al., 2018). Thus, for comparison, the Chaparral signals were band‐pass filtered between 0.1 and 20 Hz to mimic the internal filtering of the Gem sensors. The collocated Gem (G3) and Chaparral (C1) sensors are compared using the filtered Chaparral signal in Figure 6a, which shows a slightly lower peak‐to‐peak amplitude though overall good agreement through the N‐wave and coda. The peak‐to‐peak amplitude of the filtered Chaparral signals was 3.87 Pa, which was 5% above the Gem sensors.

The PSDs from the collocated Gem (G3) and filtered Chaparral (C1) sensors are provided in Figure 6b. The spectra were produced using 100 s windows with 88.9% overlap. Once again, the band‐pass filter (0.1–20 Hz) was applied to the Chaparral signal. The fundamental peak frequency from the Gem sensors was 5.1 Hz, slightly higher than C1. The Gem PSD amplitude was slightly lower, which is consistent with the lower amplitude in the time trace. The overall structure of both sensors is in excellent agreement with the only significant variation being above 20 Hz (i.e., the filtered frequency range). Thus, the Gem sensor cannot resolve higher frequencies, but they are in excellent agreement with Chaparral sensors over their design frequency band (1–20 Hz).

Figure 7 shows the f‐k analysis results using the Gem array. The primary arrival is denoted by the strong coherent peak in F‐ratio at t = 0 s. BAZ for this signal was 167°, which was only half degree different from that reported by the Chaparral array. Given the configuration and sample rate of the Gem array, the nominal array resolution for the Gems was 2°. Thus, the 0.5° difference between the Gem and Chapparal arrays was within the accuracy of the measurement given the array geometry and sample rate. Also of note, all the signals with high coherence in the coda had nominal BAZs of ∼185°. This contrasts with the Chaparral array that had multiple strong arrivals from the opposite direction in the coda. It is important to note that this analysis only identifies the BAZ associated with the maximum coherence, and in the coda there were likely signals being received from multiple paths that this analysis cannot differentiate between.

The time trace for the seven WERD sensors (see the supplemental material) shows that the N‐wave arrivals are readily apparent as well as the coda following the N‐wave. The time difference between N‐waves is shorter relative to the Chaparral sensors, which reflects the smaller apertures. The average peak‐to‐peak amplitude between the seven WERD sensors was 4.07 ± 0.08 Pa. Figure 8a compares the WERD sensor (W1) that was collocated with the center Chaparral sensor (C1). The Chaparral signal was band‐pass filtered (0.1–80 Hz) based on the WERD sensor design. The peak‐to‐peak amplitudes are in excellent agreement (WERD = 4.07 Pa; filtered Chaparral = 4.03 Pa) as well as the coda.

Figure 8b compares the PSD from the WERD sensor (W1) collocated with the Chaparral sensor (C1), and again the Chaparral signal was band‐pass filtered (0.1–80 Hz). The fundamental peak frequency for the WERD was 5.01 Hz, which is similar to Chaparral (4.98 Hz). The only significant deviation was below ∼1 Hz, but that WERD sensor (W7) was noisier than most of the WERD sensors. Thus, Figure 8b also shows a second WERD sensor (W7), which is in good agreement with the Chaparral sensor below 1 Hz. This shows that the deviation in W1 was related to self‐noise of that particular sensor. These deviations are consistent with the performance of the Gem sensors. Overall, there is excellent agreement between WERD and Chaparral sensors.

The f‐k analysis for the WERD sensors was done using WERD1 (W1–W4) and WERD2 (W5–W7) arrays separately as well as with them combined. Figure 9 shows the f‐k analysis results using all the WERD sensors (W1–W7). The strongest F‐ratio peak coincides with the N‐wave arrival with other weaker (though significant) coherent arrivals occurring within the coda. The BAZ of the primary N‐wave signal was 165°, which is slightly different from the Chaparral results. Using the individual arrays, the N‐wave BAZ for WERD1 and WERD2 were 163.5° and 162°, respectively. The nominal array resolution for the two WERD arrays was ∼0.5°. Thus, the 4° and 5.5° differences were larger than the limits of array configuration. Potential sources for this discrepancy include sensor timing errors (i.e., WERD sensors have independent data acquisition synchronized using time stamps) and accuracy of the separation distances between sensors (most likely error source). This is also consistent with the combined array (W1–W7) producing better results due to the associated decrease in the impact of such errors. In addition, note that the WERD sensors do have a strong coherent arrival ∼10 s after the N‐wave from the direction of the mountain range that produced the reflection observed on the Chaparral array.

Figure 10a compares the collocated RedVox (R1) and Chapparal (C1) sensors. The RedVox data are a normalized amplitude because they measure relative to maximum sensor deflection. At the expected N‐wave arrival, the RedVox sensors captured two bipolar waves. The total signal (i.e., both peaks) duration had a similar period to the observed N‐wave (∼0.2 s) with the first and second wavelets aligning with the positive and negative, respectively, portions of the N‐wave. The RedVox pulse is from the high‐passed frequency response and phase shifts at infrasound ranges (<20 Hz) (Asmar et al., 2019; Slad and Merchant, 2021; Takazawa et al., 2024). The effect of this can be seen in Figure 10b, which compares the spectral content between the collocated RedVox (R1) and Chapparal (C1) sensors. Note that both data sets were normalized before the spectral analysis. A 10 Hz high‐pass filter was also applied to the Chaparral data to mimic the RedVox performance. Unfortunately, the ∼5 Hz peak associated with the N‐wave is within this frequency band, though the attenuated peak is still seen in Figure 10b. The attenuated peak occurs at 7.2 Hz, which matches the high‐pass‐filtered Chapparal sensor (7.1 Hz). The agreement shows that this frequency shift in the peak is a product of the high‐pass filter. The spectral ripples are in good agreement between the two sensors. Beamforming was not performed on these data.

The current study reports on the ground-based acoustic measurements acquired at Wendover Airport during the OSIRIS‐REx SRC re-entry on 24 September 2023. All 18 operational sensors captured the N‐wave signal produced by the SRC re‐entry. The sensors included four Chaparral Physics (model 64) sensors, three Gem Infrasound loggers, seven WERD sensors (ISSM23), and four smartphones operating the RedVox application. The Chaparral Physics array served as the high‐fidelity reference sensors that the economical sensors (Gem, WERD, and RedVox) were compared against. Each economical sensor had one collocated at the center of the Chaparral array with the remaining sensors deployed in a triangular array pattern.

The Chaparral time traces showed an N‐wave signal followed by broadband coda. The N-wave signal had an average duration of 0.196 s and an average peak-to-peak amplitude of 4.07 Pa. The PSD showed a broadband fundamental peak at 4.98 Hz with spectral ripples up to ∼50 Hz during the SRC re-entry. Beamforming showed that the dominant signal had a BAZ of 167.5°, which was nominally perpendicular to the SRC trajectory. Secondary arrivals with high coherence were also detected and, for a subset of the arrivals, it indicates they had reflected from the nearby mountain range. The PSD from OSIRIS‐REx was also compared with the PSD during the Genesis and Stardust SRC re-entries (ReVelle et al., 2005; Revelle and Edwards, 2007), which all three produced a broadband fundamental peak with spectral ripples. Genesis had a fundamental peak at nominally half the frequency of OSIRIS‐REx (2.25 Hz; ReVelle et al., 2005), whereas Stardust was almost exactly the same (5.00 Hz, Revelle and Edwards, 2007).

For the economical sensors, the average peak‐to‐peak amplitudes of the N‐wave using the Gem and WERD sensors were 3.66 and 4.07 Pa, respectively. The discrepancy relative to the Chaparral sensors was primarily due to internal filtering used for Gem and WERD sensors, which was shown by comparing each sensor to filtered Chaparral data. The RedVox application captured a high‐pass‐filtered version of the N‐wave, but the amplitude could not be compared due to the application only reporting values relative to the maximum diaphragm deflection. PSDs from the Gem, WERD, and RedVox sensors identified fundamental peak frequencies at 5.1, 5.01, and 7.1 Hz. The RedVox fundamental peak frequency was impacted by its 10 Hz high‐pass filter, which was confirmed by filtering the Chaparral data. Beamforming the Gem and WERD arrays produces BAZs of 167° and 165°, respectively. The Gem sensors BAZ was within the limitations of the array configuration, whereas the WERD deviation was likely due to limitations associated with timing and/or separation distances.

Although all the sensors were effective at monitoring such geophysical events, each sensor comes with trade‐offs. The Chaparral sensors provide high‐fidelity results (precise calibration and higher frequency range), but they cost 5–10 times more than the other sensors and are less flexible in usage. The Chaparral array with its accessories was much heavier (∼60 kg) and required the entire array to be hardwired together. In contrast, the Gem sensors had a narrower frequency band that had some impact on the N‐wave amplitude but produced excellent beamforming results while being much easier to deploy (small size, low mass [<1 kg], and no wiring to single data acquisition/power). The WERD sensors are an intermediate option that has the primary advantage of being potentially the low‐cost option while having a slightly broader range than the Gems. However, the WERD sensors were operated from a single battery (not required) and as a result, for this particular deployment, it was heavier (∼25 kg) due to needing to run wires to a single battery. The RedVox sensors capture the N‐wave, though it was altered by its 10 Hz high‐pass filter. The biggest advantage of RedVox is that it is accessible to anyone with a smartphone, but signals are uncalibrated and attenuated below 10 Hz.

Details of this deployment as well as the larger collaborative observational campaign are available in Silber et al. (2024). OSU observational data are available upon request to the corresponding author. The supplemental material includes a table summarizing the sensor sizes, a schematic showing the relative layout of all sensors, and time traces during the N‐wave arrival for all sensors.

T. C. Wilson is the founder of Wilson Engineering Research and Development LLC (WERD), which manufactures the ISSM23 sensors. Although he did not take a lead role in the analysis of the WERD data, readers should be aware of this potential conflict of interest. The other authors acknowledge that there are no conflicts of interest recorded.

The authors would like to thank Elizabeth Silber and Daniel Bowman for organizing the Origins, Spectral Interpretation, Resource Identification, and Security-Regolith Explorer (OSIRIS‐REx) geophysical observational campaign, assistance in selecting the Wendover Airport area, and for providing guidance during the data analysis. The authors also thank Patrick Liu and Eric Lam from the Air Force Research Lab (AFRL/RYAA) for providing the RedVox data. The authors also thank the Wendover Airport staff for allowing the sensor deployment on the airport property. The authors also appreciate the insightful feedback provided by Danny Bowman and an anonymous reviewer on the initial draft of this article. This work was funded, in part, by the Gordon and Betty Moore Foundation under Grant GBMF11559 (doi: 10.37807/GBMF11559). M. Garces was supported in part by the Department of Energy National Nuclear Security Administration under Award Numbers DE‐NA0003920 (MTV) and DE‐NA0003921 (ETI).