In situ recording of Mars soundscape

Abstract

Before the Perseverance rover landing, the acoustic environment of Mars was unknown. Models predicted that: (1) atmospheric turbulence changes at centimetre scales or smaller at the point where molecular viscosity converts kinetic energy into heat¹, (2) the speed of sound varies at the surface with frequency^2,3 and (3) high-frequency waves are strongly attenuated with distance in CO₂ (refs. ^2-4). However, theoretical models were uncertain because of a lack of experimental data at low pressure and the difficulty to characterize turbulence or attenuation in a closed environment. Here, using Perseverance microphone recordings, we present the first characterization of the acoustic environment on Mars and pressure fluctuations in the audible range and beyond, from 20 Hz to 50 kHz. We find that atmospheric sounds extend measurements of pressure variations down to 1,000 times smaller scales than ever observed before, showing a dissipative regime extending over five orders of magnitude in energy. Using point sources of sound (Ingenuity rotorcraft, laser-induced sparks), we highlight two distinct values for the speed of sound that are about 10 m s^-1 apart below and above 240 Hz, a unique characteristic of low-pressure CO₂-dominated atmosphere. We also provide the acoustic attenuation with distance above 2 kHz, allowing us to explain the large contribution of the CO₂ vibrational relaxation in the audible range. These results establish a ground truth for the modelling of acoustic processes, which is critical for studies in atmospheres such as those of Mars and Venus.

Fig. 1. Variety of sounds recorded by SuperCam.

Atmospheric spectra spread over the light blue area; turbulence increases in the direction of the arrow. LIBS acoustic spectra spread over the light red area. Ingenuity tones are recorded at 84 Hz and 168 Hz (purple). The black spectrum is the quietest recording so far below 1 kHz. SuperCam’s microphone is located on the rover mast (green).

Fig. 2. Sound recordings and correlation with atmospheric data.

Recording of Sol 38b. a, On top, the y axis of the time series ranges from −0.2 to 0.2 Pa. The spectrogram (bottom) shows bursts that extend to 300 Hz. Overlaid, with the y axis on the right, are wind speeds from MEDA booms. b, The PSD calculated for SuperCam’s microphone (in Pa² Hz⁻¹ for 167 s) and for MEDA pressure (in Pa² Hz⁻¹ for 51 min around the microphone acquisition time) and MEDA wind data (in (m s⁻¹)² Hz⁻¹). The wind PSD is artificially offset by 10⁻² in the y axis.

Fig. 3. Sound speed variations.

a, Sound speeds as a function of local time from LIBS time-of-flight data in purple. Other sound speeds are calculated at the three heights from the MEDA temperatures and at the surface and at 2-m altitude from MCD simulations; for these conversions, the adiabatic index above f_R is used. Error bars for microphone data: standard deviation of the sound speeds during each laser burst (vertical); total duration of the burst (horizontal). Error bars for the MEDA data: standard deviation of 1-h bins between Sols 37 and 216. b, Sound speeds are calculated at three heights from the MEDA temperatures during Ingenuity’s fourth flight; the adiabatic index below f_R is used. The sound speed estimated from the Ingenuity Doppler effect is in purple. Error bars: 95% confidence interval of the Doppler shift fit.

Fig. 4. Sound attenuation with distance.

a, Sound amplitude as a function of target distance r from LIBS acoustic data between 6 kHz and 11 kHz. The second vertical axis on the right is for sound pressure level in dB. Signal intensities are in dB relative to 20 μPa. Error bars: standard deviation of the acoustic amplitudes during each laser burst. b, Comparison of the attenuation models for Mars^, (computed at 240 K and 740 Pa) and Earth (293 K and 30% relative humidity). The experimental points correspond to this study. Error bars: 95% confidence interval of the fit performed in Fig. 4a (vertical) and width of each frequency range (horizontal).

Extended Data Fig. 1. Artificial sounds recorded by Perseverance.

For each panel: time is along the x axis; on top time series data in mPa (see hereafter for the range when it exists); at the bottom PSD in Pa² Hz⁻¹ (a colour scale is shown next to panel a; ranges are different for each panel but not indicated). a, Unfiltered spectrogram of a segment (240–270 s) of the 16-min EDL microphone recording during the rover drive (Sol 16). Time series data are not calibrated in pressure. b, Spectrogram of Ingenuity’s fourth flight (Sol 69). Time series data range from −10 mPa to +10 mPa. c, Unfiltered spectrogram of MOXIE oxygen production (Sol 81). Time series data range from −2.5 mPa to +2.5 mPa. d, Series of ten LIBS shots on target Hedgehog (Sol 37), labelled 1 to 10. Time series data range from −500 mPa to 500 mPa.

Extended Data Fig. 2. Recording of laser-induced shock wave.

Target Hedgehog (point #1, shot #1). The inset focuses on the time when the acoustic wave is recorded.

Extended Data Fig. 3. Recording of the BPF of Ingenuity’s fourth flight.

a, PSD for the entire recording (purple). The rover’s thermal pumps are shown at 195 Hz and 198.75 Hz. PSD for Sol 38a is the reference of a quiet day. b, Top, variations of the received frequency along the trajectory of the flight (diamond symbol). During the first 60 s, the recording is perturbed by high winds. In red, the Doppler effect is modelled with f = 84.44 Hz at the source and c = 237.7 m s⁻¹. Bottom, range between the rover and Ingenuity (solid line, left y axis) and rotorcraft speed (dotted line, right y axis).