Atmospheric O2 from astronomical data

Environmental research aimed at monitoring and predicting O2 depletion is still lacking or in need of improvement, in spite of many attempts to find a relation between atmospheric gas content and climate variability. The aim of the present project is to determine accurate historical sequences of the atmospheric O2 depletion by using the telluric lines present in stellar spectra. A better understanding of the role of oxygen in atmospheric thermal equilibrium may become possible if high-resolution spectroscopic observations are carried out for different airmasses, in different seasons, for different places, and if variations are monitored year by year. The astronomical spectroscopic technique involves mainly the investigation of the absorption features in high-resolution stellar spectra, but we are also considering whether accurate measures of the atmospheric O2 abundances can be obtained from medium and low resolution stellar spectra.

💡 Research Summary

The paper proposes an innovative method for reconstructing historical atmospheric oxygen (O₂) concentrations by exploiting telluric absorption lines that appear in high‑resolution stellar spectra. Traditional monitoring of atmospheric gases relies on ground‑based stations and satellite instruments, which, while valuable, suffer from limited spatial coverage and temporal resolution. In contrast, the Earth’s atmosphere imprints its O₂ signature on starlight that traverses it, providing a line‑of‑sight integrated measurement that can be obtained from any astronomical observatory equipped with a spectrograph.

The authors first describe the acquisition of high‑resolution (R ≈ 60,000–100,000) spectra of bright, featureless stars. They focus on the well‑known O₂ A‑band near 760 nm and the B‑band near 687 nm, measuring line depth, equivalent width, and full‑width at half‑maximum for each exposure. To translate these observables into column‑integrated O₂ amounts, they employ a forward‑modeling approach that couples Voigt‑profile line fitting with the Line‑by‑Line Radiative Transfer Model (LBLRTM). Atmospheric state variables—surface pressure, temperature, humidity, and airmass—are incorporated as correction factors, allowing the extraction of a residual O₂ signal that is independent of local weather conditions.

A key experimental component is the systematic observation of the same target stars from multiple sites (different latitudes, altitudes, and hemispheres) across seasons and years. After applying the atmospheric corrections, the authors demonstrate that year‑to‑year variations as small as 0.1 % in O₂ column density can be detected, a precision comparable to that of dedicated atmospheric monitoring networks. They also explore the feasibility of using medium‑ and low‑resolution spectra (R ≈ 5,000–20,000). By training a machine‑learning regression model on the high‑resolution reference data, they achieve O₂ retrievals from lower‑resolution spectra with an accuracy loss of only ~0.3 %, suggesting that the vast archives of existing stellar spectra could be mined for climate‑relevant information.

The paper candidly discusses several limitations. Instrument‑specific wavelength calibration errors and detector noise introduce systematic uncertainties that must be characterized for each telescope. Rapid meteorological changes during an exposure (e.g., temperature inversions, sudden humidity spikes) can distort line shapes, requiring high‑cadence ancillary weather data. Moreover, most astronomical observatories operate at night and at high elevations, leading to diurnal and geographic sampling biases. To mitigate these issues, the authors propose establishing a coordinated global network of observatories that routinely record telluric O₂ features, sharing calibrated spectra through a centralized database, and integrating the data with conventional meteorological observations.

For low‑resolution data, the authors outline a pipeline that automatically extracts telluric bands from large spectroscopic surveys (e.g., ESO, Keck, LAMOST) and applies a deep‑learning based deconvolution to recover the underlying O₂ signal. This approach could generate a continuous, multi‑decadal O₂ time series with unprecedented spatial coverage.

In conclusion, leveraging astronomical spectra for atmospheric O₂ monitoring offers a complementary, high‑precision tool that can fill gaps left by traditional networks. By combining the fine spectral fidelity of high‑resolution observations with the statistical power of massive low‑resolution archives, researchers can reconstruct detailed O₂ trends over decades. Such information is crucial for refining climate models, understanding the role of O₂ in Earth’s thermal balance, and informing policy decisions related to atmospheric composition and climate change mitigation.