In situ recording of Mars soundscape

The study investigates Mars’ acoustic environment, previously unmeasured above 20 Hz, to test predictions that low-pressure CO2 atmospheres yield frequency-dependent sound speed, strong high-frequency attenuation, and turbulence signatures at very small spatial and temporal scales. With Mars surface pressures around 0.6 kPa and CO2-dominated air, theory suggested lower sound speeds (~240 m s−1), reduced acoustic impedance (~20 dB weaker sounds than Earth for equivalent sources), and substantial attenuation. Prior to Perseverance, absence of in situ acoustic data at relevant pressures and temperatures limited validation of these models. By recording atmospheric and artificial sounds at high sampling rates (up to 100 kHz), the study aims to characterize pressure fluctuations, detect turbulence regimes at microscale, directly measure sound speed dispersion with frequency, and quantify acoustic attenuation with distance.

Previous work modeled acoustic propagation in CO2 atmospheres but disagreed substantially due to lack of validation at Mars-like pressures and temperatures (for example, Petculescu and Lueptow 2007; Bass and Chambers 2001; Williams 2001). Terrestrial atmospheric studies and Mars meteorological measurements (for example, InSight and rover sensors) provided lower-frequency pressure and wind data, but not in the audible range. Theoretical frameworks predict: (1) reduced acoustic impedance in thin atmospheres, (2) frequency-dependent sound speed due to molecular relaxation in polyatomic gases (CO2), and (3) enhanced attenuation from classical (viscous/thermal) and molecular (rotational/vibrational) processes. However, experimental constraints on attenuation and dispersion under Mars conditions were missing, especially in open environments where boundary effects are minimized.

• Instruments and datasets: SuperCam carries an electret microphone capable of sampling air pressure fluctuations from 20 Hz to 12.5 kHz at 25 kHz sampling, or up to 50 kHz at 100 kHz sampling. An additional EDL microphone records up to 20 kHz (uncalibrated). The dataset spans Sol 1–216, totaling about 4 h 40 min for SuperCam and 56 min for EDL, including atmospheric turbulence, LIBS sparks, and rover/mechanical sounds (MOXIE, Ingenuity, rover drive). Environmental data (pressure, wind, temperature) are from MEDA (1–2 Hz sampling), and modeled temperatures from the Mars Climate Database (MCD v5.3).
• Acoustic processing: Microphone voltages are converted to pressure using gain-dependent sensitivities (0.6–21.6 V Pa−1). Instrument response (bandpass 100 Hz–10 kHz) is used to correct spectra outside this range. Power spectral densities (PSDs) are computed via Welch’s estimator; spectrograms use Hanning windows (2 s typical).
• LIBS acoustic analysis (time-of-flight): Laser-induced breakdown sparks generate shock N-wave acoustic pulses (~300 μs primary pulse; total <5 ms with echoes). Recordings at 100 kHz over 60 ms per pulse are precisely triggered to measure propagation times with <10 μs uncertainty. Distances to targets (autofocus) are known to ±0.5%. Time-of-arrival is determined when the signal exceeds background by a factor of 30. For sound speed derivation, only Mars rock targets between ~2–6 m are used (excluding Ti calibration and >6 m low SNR cases), yielding 109 targets across Sols 1–216. Daytime measurements above 2 kHz probe the f > fr regime (vibrational modes unrelaxed), using γ = 1.4 for temperature-derived comparisons.
• Ingenuity Doppler analysis: SuperCam recorded several flights; flight 4 (Sol 69) was used for Doppler-based sound speed below relaxation frequency (f < fr). The blade passage frequency (BPF) near 84 Hz is tracked by Gaussian fits every 0.5 s to derive frequency shifts versus range rate; quiet periods (t > 60 s) are used to fit the Doppler model for emitted frequency and c.
• Attenuation with distance: Using LIBS as a point source, the spherical spreading (∝1/r) is combined with laser irradiance variation (r−0.698) and exponential atmospheric absorption exp(−αr). Frequency content is split into three bands corresponding to LIBS lobes: 3–6 kHz, 6–11 kHz, and 11–15 kHz. For each band, amplitudes versus distance are fit to retrieve α with 95% confidence intervals. Comparisons are made with attenuation models (Bass and Chambers; Williams) and Earth reference (Bass et al. 1984).
• Environmental correlations: Atmospheric acoustic intensities are correlated with MEDA wind speeds; combined PSDs from microphone and MEDA pressure/wind over overlapping times assess regime transitions from shear-dominated to dissipative regimes (transition between ~1–20 Hz in example).

• First in situ acoustic characterization on Mars from 20 Hz to 50 kHz, extending pressure fluctuation measurements to 1,000 times smaller scales than previously observed, and revealing a dissipative turbulence regime spanning five orders of magnitude in energy.
• Turbulence and regime transition: Acoustic PSDs (>20 Hz) correlate with MEDA winds. Frequency-domain analysis shows a clear slope change relative to low-frequency pressure/wind PSDs, indicating a transition from a probable shear-dominated regime to a dissipation regime between ~1 and 20 Hz during the Sol 38b example. Reported slopes include approximate values: pressure slope ~−1.1, wind slope ~−0.95, and acoustic slope ~−4.9 (indicative of dissipation).
• Frequency-dependent speed of sound (dispersion): Due to CO2 vibrational relaxation, two regimes of sound speed are observed, differing by ~10 m s−1 across the relaxation frequency fr ≈ 240 Hz at Mars surface conditions (~0.6 kPa, ~240 K).
  • High-frequency regime (f > fr): From LIBS time-of-flight, daytime sound speeds between 246 and 257 m s−1 were measured. Diurnal variability shows maximum values between 11:00–14:00 LTST, minima ~18:00, and dispersion during ~20 min LIBS sequences of up to 1.5% at noon, decreasing to ~0.5% by 18:00, consistent with reduced turbulence at dusk. These values agree with temperature-derived speeds (γ = 1.4) using MEDA temperatures at 0.85 m and 1.45 m and MCD surface/2 m temperatures.
  • Low-frequency regime (f < fr): From Ingenuity flight 4 Doppler at BPF ≈ 84.43 Hz, c = 237.7 ± 3 m s−1 is obtained. Accounting for ~2.5 m s−1 LOS wind (toward helicopter), the true sound speed is ~240 m s−1. With MEDA temperatures 232–240 K at 1.45 m and γ = 1.2857 (below fr), temperature-derived speeds of 238.8–242.9 m s−1 are consistent with the Doppler result. Overall, a dispersion of ~10 m s−1 across the audible range is highlighted.
• Strong acoustic attenuation above 2 kHz:
  • Retrieved atmospheric attenuation coefficients α (95% CI) for LIBS bands: 3–6 kHz: α = 0.21 ± 0.04 m−1; 6–11 kHz: α = 0.34 ± 0.05 m−1; 11–15 kHz: α = 0.43 ± 0.05 m−1. Higher frequencies attenuate more strongly.
  • Example at 8 kHz relative to 1 m: attenuation is −9 dB at 2 m and −40 dB at 8 m; at 5 m, atmospheric absorption dominates over geometric spreading. For comparison, on Earth at 8 kHz with α ≈ 0.01 m−1, attenuation is ~−6 dB at 2 m and ~−20 dB at 8 m; −40 dB would require ~65 m.
  • Model comparison: In situ attenuation tends toward Bass and Chambers behavior (plateau <6 kHz, increasing at higher frequencies) and does not exhibit the attenuation gap predicted by Williams. The 2–6 kHz band still shows higher α than Bass and Chambers, suggesting differences in relaxation strength under Mars conditions.
• Acoustic impedance difference: At Mars surface (ρ ≈ 0.02 kg m−3, c ≈ 238 m s−1), Z ≈ 4.76 kg m−2 s−1 vs. Earth Z ≈ 413 kg m−2 s−1, explaining ~20 dB weaker sounds on Mars for the same source.
• Artificial sources characterized: Ingenuity BPF at 84 Hz and harmonic at 168 Hz observed; MOXIE compressor tones with commanded speed changes (50–58.3 Hz) detected; rover drive sounds show frictional bands (520–700 Hz, 1.2–1.4 kHz, 1.6–1.9 kHz) and transient clanks (structural resonances).

The results validate long-standing theoretical predictions for sound propagation in a low-pressure, CO2-dominated atmosphere and provide critical ground truth. The observed frequency-dependent sound speed, with two distinct regimes separated near 240 Hz, confirms the role of CO2 vibrational relaxation at Mars conditions and refines expectations for acoustic dispersion. The measured attenuation coefficients quantify the strong absorption of high-frequency sound and support models that include vibrational relaxation effects (consistent with Bass and Chambers), thereby constraining acoustic propagation models essential for interpreting acoustic signals on Mars and similar atmospheres (for example, Venus). The acoustic PSDs and their correlation with wind demonstrate sensitivity to boundary-layer dynamics at unprecedented temporal resolution (10–1,000 times higher than prior datasets), revealing the dissipative regime and its transition from shear-dominated behavior above a few Hz. These insights can inform and constrain planetary boundary layer parameterizations and turbulence closure assumptions (including LES), particularly regarding unresolved energy fractions and dissipation rates tied to heat diffusion. Beyond environmental science, acoustic monitoring of rover systems (for example, MOXIE, mobility) offers diagnostics for instrument health. Overall, the findings address the initial questions by: (1) quantifying turbulence signatures and dissipative scaling in the audible band, (2) directly measuring sound speed dispersion in situ, and (3) retrieving atmospheric attenuation with distance, thereby reducing uncertainties in Mars acoustic modeling.

This work presents the first in situ measurements of Mars’ acoustic environment from 20 Hz to 50 kHz, establishing ground truth for sound speed dispersion and frequency-dependent attenuation in a CO2-dominated, low-pressure atmosphere. Key contributions include: (1) direct measurements of sound speed in two regimes separated by the CO2 vibrational relaxation frequency (~240 Hz), with a ~10 m s−1 contrast; (2) quantification of strong high-frequency attenuation (α ≈ 0.21–0.43 m−1 over 3–15 kHz), supporting models that include vibrational relaxation; and (3) high-temporal-resolution turbulence observations revealing the dissipative regime and its transition above a few Hz. These results enhance modeling of acoustic processes and boundary-layer dynamics on Mars and provide a framework for future geophysical applications. Future work should acquire more sound speed and attenuation measurements across local times and seasons, further characterize the transition to dissipation and associated energy cascades, and quantify dissipation rates, improving parameterizations in Mars PBL models. Continued acoustic monitoring can also support diagnostics of rover subsystems’ health.

• Frequency coverage limits: LIBS-based attenuation measurements do not extend below ~2–3 kHz, leaving the large discrepancies between models in the sub-kHz to ~1 kHz range unconstrained.
• Signal-to-noise constraints: For sound speed derivation, targets beyond ~6 m were excluded due to low SNR, potentially biasing statistics toward closer ranges and specific surface conditions; Ti calibration targets were excluded for speed (bias from rover-induced heating/turbulence).
• Instrumental/processing considerations: The SuperCam microphone response requires corrections outside 100 Hz–10 kHz; EDL microphone data are uncalibrated (arbitrary units). Electromagnetic interferences (for example, from laser warm-up) are mitigated via bandpass filtering but may leave residuals.
• Environmental and geometric variability: Attenuation retrieval assumes spherical spreading and a deterministic laser irradiance-distance relation; local surface roughness, echoes, and rover structures can introduce variability. Day-to-day atmospheric state (dust, temperature gradients) and wind can affect measurements.
• Seasonal and spatial sampling: Data span Sols 1–216 of one mission location; broader seasonal coverage and other terrains/heights are needed to generalize results.