Estimating the H I gas fractions of galaxies in the local Universe

We use a sample of 800 galaxies with H I mass measurements from the HyperLeda catalogue and optical photometry from the fourth data release of the Sloan Digital Sky Survey to calibrate a new photometric estimator of the H I to-stellar mass ratio for nearby galaxies. Our estimator, which is motivated by the Kennicutt-Schmidt star formation law, is log(G_{HI}/S) = -1.73238(g-r) + 0.215182mu_i - 4.08451, where mu_i is the i-band surface brighteness and g-r is the optical colour estimated from the g- and r-band Petrosian apparent agnitudes. This estimator has a scatter of sigma = 0.31 dex in log(G_{HI}/S), compared to sigma ~ 0.4 dex for previous estimators that were based on colour alone. We investigate whether the residuals in our estimate of log(G_{HI}/S) depend in a systematic way on a variety of different galaxy properties. We find no effect as a function of stellar mass or 4000A break strength, but there is a systematic effect as a function of the concentration index of the light. We then apply our estimator to a sample of 10^5 emission-line galaxies in the SDSS DR4 and derive an estimate of the H I mass function, which is in excellent agreement with recent results from H I blind surveys. Finally, we re-examine the well-known relation between gas-phase metallicity and stellar mass and ask whether there is a dependence on H I-to-stellar mass ratio, as predicted by chemical evolution models. We do find that gas-poor galaxies are more metal rich at fixed stellar mass. We compare our results with the semi-analytic models of De Lucia & Blaizot, which include supernova feedback, as well as the cosmological infall of gas.

💡 Research Summary

The authors set out to develop a photometric estimator for the neutral hydrogen (H I) to stellar mass ratio (G_HI/S) of nearby galaxies, using only optical data. They begin with a calibration sample of 800 galaxies for which H I masses are available from the HyperLeda catalogue and for which g, r, and i‑band photometry is provided by SDSS‑DR4. Earlier work relied mainly on the optical colour (g−r) to predict log(G_HI/S), achieving a scatter of about 0.4 dex. Recognising that the Kennicutt–Schmidt law links star‑formation surface density to gas surface density, the authors augment the colour term with the i‑band surface brightness μ_i (in mag arcsec⁻²). A multivariate linear regression yields the following relation:

log(G_HI/S) = –1.73238 (g−r) + 0.215182 μ_i – 4.08451.

The coefficient of (g−r) is negative, reflecting the well‑known trend that redder galaxies (older stellar populations) are typically more gas‑poor, while the positive μ_i term indicates that higher stellar surface density correlates with higher gas fractions at fixed colour. The resulting estimator reduces the scatter to σ ≈ 0.31 dex, a ∼20 % improvement over colour‑only methods.

To test for systematic biases, the residuals of the estimator are examined against several galaxy properties. No dependence is found on stellar mass or on the 4000 Å break strength (Dₙ4000), suggesting that the estimator is largely insensitive to overall mass and stellar age. However, a clear trend emerges with the concentration index (C = R₉₀/R₅₀): galaxies with more centrally concentrated light profiles (higher C) tend to have lower actual G_HI/S than predicted, implying that structural differences (disk‑ versus bulge‑dominated systems) affect the gas distribution beyond what colour and surface brightness capture.

Having validated the estimator, the authors apply it to a much larger SDSS‑DR4 sample of ∼10⁵ emission‑line galaxies. By converting the predicted G_HI/S values into H I masses, they construct an H I mass function (HIMF). The derived HIMF matches remarkably well with those measured directly by blind H I surveys such as ALFALFA and HIPASS, especially at the low‑mass end (10⁸–10⁹ M_⊙) where previous colour‑only estimators tended to underpredict the number density. This agreement demonstrates that the new photometric method can reliably reproduce the statistical properties of the H I population without requiring radio observations.

Finally, the paper revisits the mass–metallicity relation (MZR) by incorporating the estimated G_HI/S. At fixed stellar mass, galaxies with lower G_HI/S (i.e., gas‑poor) exhibit higher gas‑phase metallicities, consistent with simple chemical evolution models where inflow of pristine gas dilutes metals while outflows remove them. The authors compare these observational trends with the semi‑analytic models of De Lucia & Blaizot (2007), which include supernova feedback and cosmological gas accretion. While the models reproduce the overall shape of the MZR and its dependence on G_HI/S, they do not fully capture the concentration‑index dependence observed in the residuals, indicating that additional physics—perhaps related to bulge growth or environmental stripping—needs to be incorporated.

In summary, the study delivers a robust, low‑scatter photometric estimator for H I content that leverages both colour and surface brightness, validates it against a sizable calibration set, demonstrates its utility for constructing the H I mass function from purely optical surveys, and uses it to deepen our understanding of the interplay between gas, metals, and galaxy structure. The methodology opens the door for large‑scale statistical studies of gas in galaxies across cosmic time, especially when forthcoming optical surveys (e.g., LSST) will vastly outnumber existing H I observations.