How Earth's sky can distort a galaxy's stellar history

A galaxy's spectrum contains clues to the ages and chemical composition of its stars. But a telescope on Earth also collects light from the sky, including emission from the atmosphere. Removing that foreground is essential: subtracting too much or too little can alter the spectral features used to reconstruct how a galaxy formed its stars.

Camila de Sá-Freitas, Fuyan Bian and Patrícia Sánchez-Blázquez investigated this problem using observations of NGC 1097 and NGC 1433. Their study asks a practical question: how much can the choice of sky measurement change the stellar populations inferred from the same galaxy?

Measuring the foreground separately

The observations came from MUSE, the Multi Unit Spectroscopic Explorer at the Very Large Telescope. MUSE records spectra across a field of view, allowing researchers to map stellar properties rather than measure just one combined spectrum.

Nearby galaxies can fill that field, leaving no suitable empty area for measuring the sky. Astronomers therefore take separate exposures pointed away from the galaxy. These sky frames must represent the foreground present during the galaxy exposure, even though they are not recorded simultaneously.

The team examined two situations. NGC 1097 was observed over roughly two consecutive hours, during which the measured sky flux varied by up to about 18%. The researchers processed its eight galaxy exposures five ways: once using the usual nearby-in-time sky frames, then four more times using each individual sky frame for all eight exposures. They changed only the sky-subtraction step.

NGC 1433 offered a different comparison. Its observations were taken on two nights with different transparency, Moon illumination and airmass—the amount of atmosphere along the telescope's line of sight. The team compared results from each night's four exposures with those from all eight combined.

Bright regions held up better

The researchers fitted spectra between 4,800 and 5,800 angstroms using the same analysis settings and stellar-model library throughout. They estimated stellar age, metallicity—the abundance of elements heavier than helium—and alpha-enhancement, a measure of the relative abundance of a particular group of elements.

For NGC 1097, regions more than about ten times brighter than the sky in the study's flux comparison produced very similar stellar properties across the different reductions. This held despite the variation among sky frames. Fainter regions were more sensitive, and the intermediate sky exposures generally agreed better with the default reduction than the first or last exposure.

The spatial comparison reinforced that result. Inside the bright nuclear disc, a central stellar structure, inferred properties and star-formation histories changed relatively little. Outside it, different sky choices produced substantially different histories. These differences concern the reconstruction, not actual changes in the galaxy during the observations.

NGC 1433 showed a similar brightness dependence. Its bright nuclear disc gave consistent results between nights. In faint regions, particularly those less than twice as bright as the sky, metallicity was much more vulnerable. Some estimates shifted from negative to positive values on the study's metallicity scale.

That matters because chemical composition helps inform explanations of a galaxy's growth. Metal-poor populations can be associated with incoming gas from interactions, whereas metal-rich populations can indicate locally enriched star formation. The study does not establish either history for these regions; it shows that observing conditions can influence evidence used to distinguish them.

A troublesome spectral overlap

The team identified a specific source of contamination. An atmospheric nitrogen emission feature near 5,200 angstroms overlaps the redshifted magnesium absorption region known as Mgb for galaxies with redshifts between 0.0014 and 0.0077. Mgb helps constrain metallicity and alpha-enhancement. Residual sky light there can therefore interfere directly with useful stellar diagnostics.

Other affected features included an oxygen sky line near an iron diagnostic and the hydrogen H-beta feature used to assess stellar ages. Their relative impact was generally greater outside the bright nuclear discs.

A warning, not a universal threshold

These are two case studies, not a calibrated prescription for all MUSE observations. The analysis used one stellar library and a restricted wavelength range. The authors did not calculate uncertainties for the derived stellar-population parameters, and combining four rather than eight exposures can also change the spatial binning required to reach the target signal-to-noise ratio.

The default reduction was a comparison baseline, not independently established ground truth. Nevertheless, the controlled NGC 1097 reductions demonstrate that sky-frame choice alone can alter inferred properties. The practical lesson is to assess sky subtraction against the brightness and scientific purpose of the target, rather than assume that acceptable observing conditions guarantee equally reliable stellar histories everywhere.