The second Venus flyby of BepiColombo mission reveals stable atmosphere over decades

Venus and Earth formed in the same region of the solar nebula and have similar sizes, yet their climates diverged dramatically. Understanding the divergent evolution of Earth and Venus—especially in the context of Earth-like exoplanets—is a major challenge in planetary science, with implications for feedback mechanisms in Earth’s climate system. Tracking Venus’s thermal structure and cloud formation, particularly in the transition region between the troposphere and mesosphere (about 60–70 km altitude), is crucial for climate monitoring because of its strong variability with latitude and local time. The present study aims to provide new mid-infrared (mid-IR) spectral observations of the Venusian mesosphere using BepiColombo’s MERTIS instrument during its second Venus flyby (FB2, August 10, 2021). The objective is to retrieve temperature profiles and cloud properties and compare them with historical datasets (e.g., Venera-15/PMV) and models (VIRA) to assess the stability of Venus’s atmosphere over decades and to demonstrate the capability of an instrument optimized for Mercury to measure Venus’s cooler atmosphere.

Past remote sensing of Venus’s temperature profiles and cloud features includes the Venera-15 FS-1/4 Fourier spectrometer (PMV) in 1983, which provided spectra from 6.0–36.5 µm (275–1656 cm⁻¹) at 6.3 cm⁻¹ resolution. ESA’s Venus Express (2006–2014) delivered extensive global datasets, notably through VIRTIS for thermal imaging/spectroscopy and VeRa for radio science. More recent data came from JAXA’s Akatsuki mission (2016–2018), with radio occultation and thermal imaging providing high-precision mesospheric temperature profiles and insights into waves and dynamics. Mid-IR nadir spectra longward of 5 µm comparable to PMV had not been acquired since 1983. BepiColombo’s MERTIS now provides the first hyperspectral mid-IR observations in four decades, enabling direct comparisons with PMV and validation against VIRA as well as complementary radio occultation results from PVO, Magellan, Venus Express, and Akatsuki.

Observations: During the second Venus flyby (FB2) on August 10, 2021, BepiColombo approached to a closest altitude of 552 km at 13:51 UTC. MERTIS observed Venus using its space baffle (normally used for deep-space calibration) because the planetary baffle was blocked in cruise configuration. Observations spanned −4 to +23 minutes around closest approach, at distances of ~13,000 to ~900 km, covering local times 6–18 h and latitudes mostly 8–14°N. Approximately 2,200 spectrometer-channel measurements and ~300 radiometer measurements were acquired; >900,000 individual spectra were recorded at full spatial resolution (100 pixels across track).

Data selection and averaging: Some spectra showed very low radiances (e.g., 6.8–8 µm and near 13.5 µm) due to partial deep-space viewing, yielding non-physical brightness temperatures; these were removed, leaving ~94,000 usable spectra. To minimize brightness temperature (TB) variability due to observation angle (θobs), spectra were binned and the most populated interval (20–30°) was selected, yielding ~13,000 spectra at local times ~10.5–13.0 h. Observations at angles >75° were excluded due to plane-parallel model limitations. Averaging across selected spectra reduced noise, yielding a robust zonal average spectrum comparable to PMV.

Calibration and geometry: MERTIS’s calibration uses two internal blackbodies (one near instrument temperature, one at 700 K) and deep-space views. For Venus, special handling accounted for low brightness temperatures (220–260 K), with deep-space pre/post-flyby data used to adjust image offset and stripe noise. During the flyby, the ‘cold’ blackbody was 282.4–286.9 K; periodic calibration every 60 s showed <0.03 K difference between sensor and spectrally retrieved temperatures. Spacecraft thermal changes from night-to-day viewing increased baseplate by 3.2 K and detector housing by 4.2 K; microbolometer chip temperature remained stable within 23 mK. Geometric registration used NAIF/SPICE kernels validated with Moon flyby data; PDS4-calibrated and registered products were generated.

Instrument characteristics: MERTIS-TIS operates 7–14 µm, with resolution ~90 nm (≈18 cm⁻¹ near 7 µm and 5 cm⁻¹ near 14 µm), using an uncooled microbolometer, a TMA telescope, and a modified Offner spectrometer. Compared to PMV’s 6.0–36.5 µm range, MERTIS lacks much of the 15 µm CO₂ band core (lower limit ~720 cm⁻¹), making it sensitive primarily to 60–75 km altitudes rather than 55–90 km for PMV.

Radiative transfer and cloud model: A DISORT-based RTM accounts for gas absorption/emission and multiple scattering by H₂SO₄ cloud aerosols (four modes: M1, M2, M2′, M3) with log-normal size distributions (modal radii ~0.3, 1.0, 1.4, 3.65 µm; dispersions 1.56, 1.29, 1.23, 1.28). Refractive indices follow Palmer & Williams; Mie theory provides wavelength-dependent microphysical properties. Cloud optical depth is parameterized via altitude-independent mode factors MFj scaling altitude profiles, with a reference cloud top altitude zt defined where cumulative optical depth at 10 µm equals unity (initial model opacity 28.34; zt ≈66.64 km for MF1=1). Gas composition is dominated by CO₂ (96.5% VMR); H₂O and SO₂ are negligible in the MERTIS retrieval spectral range.

Retrieval technique: Temperature profiles and cloud mode factors MF1/2 are jointly retrieved by iteratively fitting simulated radiances to the averaged spectra in a least-squares sense. Temperature is retrieved versus altitude using Smith’s relaxation method with initial profiles from VIRA, continuously updating pressure via the barometric relation. Due to spectral limitations, MF2′/M3 cannot be disentangled robustly and are set to unity. The self-consistent workflow: (1) preliminary T1(z) using 720–1100 cm⁻¹; (2) retrieve MF1/2 and cloud top altitude zt using 830–1100 cm⁻¹; (3) update T2(z) using MF1/2, zt; (4) iterate steps 2–3; (5) generate final TB spectrum across the full range. Weighting functions quantify sensitivity with altitude; for MERTIS, Wmax indicates strongest sensitivity between ~60–75 km.

Comparative strategy: To validate MERTIS’s reduced spectral coverage, PMV retrievals were run over two ranges: A (300–1100 cm⁻¹) and B (MERTIS-like, 720–1100 cm⁻¹). Agreement in simulated vs measured spectra and retrieved temperature/cloud parameters between A and B assessed the reliability of using only the long-wavenumber wing of the 15 µm CO₂ band for lower-mesosphere retrievals.

• First hyperspectral mid-IR observations of Venus (>5 µm) since 1983: MERTIS-TIS acquired >900,000 spectra during FB2; after quality filtering, ~94,000 spectra remained for analysis.
• Observation geometry: Data concentrated near 10°N latitude (8–14°N), local times 6–18 h; retrievals emphasized 10.5–13.0 h and θobs=20–30° to minimize angular effects.
• Spectral fit quality: For MERTIS zonal averages at 10°N, simulated vs measured brightness temperatures agreed within <1 K typically and never exceeded 1.5 K.
• Temperature retrieval sensitivity: MERTIS sensitivity confined to ~60–75 km altitude due to spectral coverage (≥720 cm⁻¹), while PMV covers ~55–90 km.
• Cross-validation with PMV: PMV retrievals using full (A: 300–1100 cm⁻¹) and reduced (B: 720–1100 cm⁻¹) ranges yielded temperature profiles differing by ≤1 K between 61–76 km; larger deviations only outside that range where range B has limited sensitivity. Retrieved cloud parameters identical for A and B: MF1/2=0.93; cloud top zc(10 µm)=66.4 km.
• MERTIS-retrieved mesospheric profile: At 10°N, the MERTIS temperature profile matches VIRA (low latitudes) and PMV (30°N) as well as VIRTIS retrievals, with differences typically ~1 K relative to VIRA within the sensitive altitude range.
• Cloud properties from MERTIS FB2: MF1/2=0.96; cloud top altitude zt(10 µm)=66.5 km—typical values near the equator, consistent with PMV and VIRTIS.
• Instrument performance: Despite using the space baffle and operating near sensitivity limits (Venus TB ~220–260 K), regular internal blackbody checks indicated calibration stability (<0.03 K discrepancy).
• Atmospheric stability: Agreement of MERTIS results with Venera-15/PMV (1983) and VIRA indicates remarkable stability of the Venusian lower mesosphere and cloud-top region over multi-decade timescales, particularly near the equator.

The study addresses the need for updated, spectrally resolved mid-IR measurements of Venus’s mesosphere to assess long-term atmospheric stability and to validate models used for forthcoming missions. By demonstrating that MERTIS, an instrument optimized for hot Mercury surfaces, can retrieve reliable temperature profiles and cloud parameters between 60–75 km on Venus, the work bridges a four-decade observational gap. The close agreement of MERTIS-derived profiles with VIRA, PMV, and VIRTIS confirms the robustness of the lower-mesosphere thermal structure and cloud-top characteristics at low latitudes and supports the hypothesis of a stable equatorial mesosphere over decades. These findings complement radio occultation results from Venus Express and Akatsuki, which are sensitive to finer-scale structures, together providing a coherent picture of Venus’s mesospheric state. The results also validate the use of the long-wavenumber wing of the 15 µm CO₂ band for temperature retrievals when full band coverage is unavailable, an important methodological insight for instruments with constrained spectral ranges. The demonstrated capability of an uncooled microbolometer-based hyperspectral system to measure relatively cool atmospheric scenes has broader relevance, suggesting cost- and lifetime-effective approaches for future long-duration Venus atmospheric monitoring. Finally, the validated temperature structure underpins atmospheric corrections and radiative transfer models critical for upcoming missions (EnVision, VERITAS) and their instruments (e.g., VEM).

MERTIS observations during BepiColombo’s second Venus flyby delivered the first hyperspectral mid-IR measurements of Venus’s mesosphere since 1983 and enabled retrievals of temperature profiles and cloud parameters between ~60 and 75 km. The retrieved profiles and cloud properties agree closely with VIRA and historical PMV and VIRTIS results, indicating a stable equatorial mesosphere and cloud-top structure over decades. Methodologically, the study shows that reliable atmospheric retrievals are possible using only the long-wavenumber wing of the 15 µm CO₂ band, even with an instrument not optimized for Venus and operating via a non-standard viewing port. These results both validate models used for atmospheric corrections in upcoming Venus missions and highlight the utility of uncooled microbolometer-based hyperspectral instruments for planetary atmospheres. Future work should: (1) analyze FB1 data to extend coverage from 50°S to 85°N and derive two-dimensional temperature fields; (2) further quantify stray light impacts for varied geometries; (3) integrate findings with Akatsuki and future EnVision/VERITAS datasets; and (4) pursue inclusion of microbolometer-based mid-IR spectrometers on future Venus missions for long-term mesospheric monitoring.

• Spectral coverage: MERTIS does not include much of the 15 µm CO₂ band core (lower limit ~720 cm⁻¹), limiting temperature sensitivity primarily to 60–75 km and reducing sensitivity above ~77 km and below ~60 km.
• Observation geometry and coverage: FB2 data are concentrated near 10°N and dayside local times; global latitude and local-time coverage is limited, constraining broader climatological assessments.
• Viewing port and stray light: Use of the space baffle (not optimized for planetary viewing) introduces potential stray-light uncertainties; no in-flight calibration of the space port was available for FB2, necessitating data clustering to mitigate errors.
• Data quality: Individual spectra were noisy under Venus conditions, preventing detailed per-spectrum error analysis; results rely on averaging.
• Angular constraints: Plane-parallel assumption breaks down for θobs>75°, restricting usable data; PMV comparisons required differing observation angles.
• Retrieval degeneracies: The algorithm cannot disentangle M2′ and M3 from M1/M2 variations; MF2′ and MF3 were fixed to unity.
• Dayside-only MERTIS data: Thermal emissions longward of 7 µm are not affected by solar radiation, but nightside mid-IR spectral sampling by MERTIS during FB2 was not available for comparison.