FEASTS and MHONGOOSE: HI Column Density Distribution at $z=0$ for $N_\mathrm{HI}>10^{17.8}\, \mathrm{cm}^{-2}$

This paper presents the first high-resolution ($\sim$1 kpc) measurement of the HI column density distribution function at $z=0$ down to $\log N_\mathrm{HI} = 17.8$, revealing a high covering fraction of neutral gas within 1 kpc of Milky-Way-like galaxies and providing new constraints on baryonic flows and galaxy evolution.

The Gist

The Big Picture: Mapping the "Cosmic Rain"

Imagine the universe is a giant, invisible ocean. Galaxies are like islands floating in this ocean. To grow and build stars, these islands need to drink from the ocean. The "water" they drink is Hydrogen gas (specifically, atomic hydrogen, or HI).

For a long time, astronomers could only see the "deep water" right next to the islands—the thick, dense clouds where stars are actually being born. But they couldn't see the "shallow water" or the misty spray further out, which is crucial because that's where the gas comes from before it gets sucked into the galaxy.

This paper is the first time we've made a high-definition map of that "shallow water" and "mist" right around our current cosmic neighborhood (the present-day universe).

The Tools: A Giant Ear and a High-Resolution Camera

To do this, the scientists used two powerful "ears" (radio telescopes) to listen to the faint hum of hydrogen gas:

1. FAST (The Giant Ear): Located in China, this is the world's largest single radio dish. It's incredibly sensitive and can hear very faint, diffuse gas far away from galaxies. Think of it as a super-sensitive microphone that can hear a whisper from a mile away, but it's a bit blurry.
2. MeerKAT (The High-Res Camera): Located in South Africa, this is an array of many smaller dishes working together. It sees fine details (like a high-resolution camera) but can't hear the faintest whispers as well as FAST.

The Magic Trick: The team combined data from both. They took the "blurry but loud" picture from FAST and the "sharp but quiet" picture from MeerKAT. By merging them, they created a single image that is both sharp and sensitive. It's like taking a high-definition photo of a foggy night and using a super-sensitive microphone to fill in the parts the camera missed.

The Discovery: How Much Gas is Out There?

The team looked at 70 nearby galaxies and measured the "column density" of the gas.

• Analogy: Imagine looking at a forest. "Column density" is like counting how many tree trunks are stacked up in a straight line from the ground to the sky.
  • High Density: A thick forest where you can't see through (where stars are born).
  • Low Density: A sparse scattering of trees or just a few twigs (the gas waiting to be collected).

What they found:
They successfully mapped gas all the way down to very low densities (100 times fainter than previous maps could see). They found that the distribution of this gas follows a predictable pattern, like a bell curve that gets cut off at the top.

The Time Travel Comparison: Then vs. Now

Astronomers often look at distant galaxies to see what the universe was like billions of years ago (high redshift). They found that:

• In the past (3 billion years ago): The universe was full of "foggy" gas clouds everywhere. It was easier to find gas far away from galaxies.
• Today (z=0): The universe is "clearer." There is less of this diffuse gas floating around. The gas has mostly been pulled into galaxies or turned into stars.

The Analogy: Imagine a party.

• In the past (High Redshift): The party was chaotic. People (gas) were everywhere, even in the hallways and outside the building.
• Today (Low Redshift): The party has settled down. Most people are inside the main room (the galaxy), and the hallways are mostly empty. The gas has been "consumed" by the galaxies.

The Mystery: The "Missing" Gas Clouds

Here is the most surprising part. The scientists looked at gas clouds detected by looking at distant quasars (bright lighthouses in the background). When they tried to find the galaxy that "owns" that gas cloud, they often couldn't find it, or the gas seemed to be floating in empty space.

• The Paper's Finding: When they looked at the actual maps of nearby galaxies, they realized that the gas clouds detected by the "lighthouses" are actually much closer to galaxies than we thought.
• The Analogy: Imagine you hear a noise in a dark forest and assume a bear is 100 yards away. But when you turn on a flashlight, you realize the bear was actually right next to you, just hidden in the shadows.
• The Problem: Our current computer simulations (like TNG50) predict that gas should be spread out over huge distances. But the real data shows the gas is clumped much tighter around the galaxies. This suggests our computer models of how galaxies eat gas need an update.

Why Does This Matter?

1. Fueling the Engine: Galaxies need gas to make stars. If we don't understand where the "shallow" gas is, we don't understand how galaxies feed themselves.
2. Fixing the Simulations: The fact that real gas is closer to galaxies than our simulations predict tells us we are missing something in our physics models. Maybe the gas is being heated up or pushed away by stars in ways we didn't account for.
3. The "Covering" Factor: The paper calculates that if you were to look at a random spot in the sky near a galaxy like the Milky Way, there is a very small chance (about 0.6%) that you would see a cloud of this specific type of gas. It's rare to find this "mist" right next to us, even though it's everywhere in the universe.

Summary in One Sentence

This paper used the world's most sensitive radio telescopes to create the sharpest map yet of the invisible hydrogen gas surrounding nearby galaxies, revealing that the gas is much more tightly packed around galaxies today than it was in the past, and that our computer models of the universe need to be tweaked to match this new reality.

Technical Summary

1. Problem Statement

The distribution of neutral hydrogen (HI) in the universe is a critical tracer of galaxy evolution, gas accretion, and feedback processes. While the HI column density distribution function, $f(N_{HI})$, has been well-studied at high redshifts ($z \sim 3$) using quasar absorption lines (Lyman Limit Systems, LLSs), the local universe ($z=0$) lacks a comprehensive measurement extending to low column densities.

• The Gap: Previous $z=0$ studies (e.g., Zwaan et al. 2005, "Z05") were limited by sensitivity to high column densities ($\log N_{HI} \gtrsim 19.8$), missing the crucial LLS regime ($10^{17} - 10^{20} \text{ cm}^{-2}$) where gas accretion and circumgalactic medium (CGM) interactions occur.
• The Challenge: Mapping this regime requires combining the high sensitivity of single-dish telescopes (to detect diffuse, low-density gas) with the high spatial resolution of interferometers (to resolve structures and avoid confusion). Furthermore, identifying host galaxies for absorption systems at $z=0$ has been historically difficult due to the faintness of low-mass galaxies and the complex nature of multiphase gas.

2. Methodology

Data Sources

The study utilizes two major, high-sensitivity 21-cm emission surveys:

1. FEASTS (FAST Extended Atlas for Selected Targets Survey): Observations using the Five-hundred-meter Aperture Spherical radio Telescope (FAST) targeting 55 HI-rich local galaxies.
2. MHONGOOSE (MeerKAT HI Observations of Nearby Galactic Objects): Observations using the MeerKAT telescope targeting 30 nearby star-forming galaxies.

• Final Sample: A combined sample of 70 galaxies (40 from FEASTS + interferometric data, 30 from MHONGOOSE), covering HI masses ($M_{HI}$) from $10^{7}$to$10^{10.7} M_\odot$and stellar masses ($M_*$) from$10^{7.2}$to$10^{11} M_\odot$.

Data Processing and Image Construction

A novel pipeline was developed to create uniform, high-resolution HI column density maps:

• Combination Strategy: The authors combined FAST data (high sensitivity, large scale) with interferometric data (high resolution, small scale).
  • Traditional Method: Standard feathering (Fourier space combination).
  • New Method (Fiducial): To address systematic errors caused by the non-Gaussian, multi-beam nature of FAST, the authors developed a "FAST observing simulator." This tool mimics the specific beam responses of FAST on the interferometric data to accurately extract the "diffuse HI" component. The diffuse HI is then corrected for beam variations and deconvolved (using Maximum Likelihood) to produce a clean, Gaussian-like Point Spread Function (PSF).
• Resolution and Sensitivity: The final maps achieve a physical resolution of $\sim 1$ kpc and a column density limit of $\log N_{HI} \approx 17.8 \text{ cm}^{-2}$ (100 times deeper than previous $z=0$ studies).
• Deblending: A 3D watershed algorithm was used to separate overlapping galaxies in interacting pairs.

Statistical Construction

• Weighting: Galaxies were weighted by the HI Mass Function (HIMF) from Guo et al. (2023) to ensure the sample represents the cosmic HI density.
• Metric Calculation: The column density distribution function $f(N_{HI})$ was calculated by counting pixel areas in logarithmic bins, corrected for distance uncertainties and stochastic sampling of galaxy morphologies.
• Error Analysis: Systematic uncertainties were quantified via bootstrapping (galaxy sampling), distance errors ($\pm 30\%$), and comparisons between old/new imaging procedures.

3. Key Contributions

1. First Deep $z=0$ $f(N_{HI})$: The first measurement of the HI column density distribution function at $z=0$ extending down to $\log N_{HI} = 17.8$ with $\sim 1$ kpc resolution.
2. Advanced Imaging Pipeline: Introduction of a robust method to combine single-dish and interferometric data that corrects for multi-beam artifacts, significantly improving the reliability of low-surface-brightness HI detection.
3. Impact Parameter Analysis: A statistical derivation of the probability distribution of impact parameters ($b$) for HI detection, comparing emission-based maps with absorption-based LLS observations.
4. Simulation Benchmarking: Direct comparison of observational $f(N_{HI})$ and $b$-distributions with the TNG50 cosmological simulation.

4. Key Results

The HI Column Density Distribution ($f(N_{HI})$)

• Shape: The distribution is well-fitted by a Schechter function in the range $17.8 < \log N_{HI} < 21$.
• Comparison with $z=3$:
  • For $19.2 < \log N_{HI} < 21$, the$z=0$distribution is **0.1–0.4 dex lower** than at$z \sim 3$.
  • For $17.8 < \log N_{HI} < 19.2$, the distributions become comparable, suggesting weak evolution in the low-column-density regime.
• Comparison with Simulations: The observed $f(N_{HI})$ matches the OWLS simulation predictions well for $\log N_{HI} < 20.2$ but deviates at higher densities where ISM physics is complex.

Length Incidence and Covering Fraction

• Length Incidence ($dN/dX$): Extrapolating the $z=0$ data to $\log N_{HI} > 17.5$ yields a length incidence of $\ell \approx 0.3$.
• Discrepancy: When extrapolating $z=0$ observations to higher redshifts using predicted HIMFs, the incidence is 1.48 times higher than observed LLSs at $0.5 < z < 3$.
• Covering Fraction ($f_{cov}$): This discrepancy implies a covering fraction of $\sim 0.7$ for HI clouds within 1-kpc pixels at $z=0$. For Milky-Way-like galaxies within the virial radius, the covering fraction for $\log N_{HI} > 17.8$ is estimated at $\sim 0.006$.

Impact Parameters ($b$)

• Emission vs. Absorption: The impact parameters for HI detected in emission (this study) are significantly smaller than those for LLS absorbers detected in quasar spectra at $z < 0.5$. Only 1 out of 32 LLSs falls within the 50% cumulative probability contour of the emission-based distribution.
• Implication: This suggests a major difficulty in associating absorption systems with their true host galaxies; absorbers may trace gas in satellite halos, tidal tails, or the IGM rather than the central galaxy.
• TNG50 Comparison: The TNG50 simulation overestimates the impact parameters for low-column-density HI ($\log N_{HI} < 20$) around Milky-Way-like galaxies, predicting gas extending 30–50% further than observed.

5. Significance

• Constraints on Galaxy Evolution: The results provide new constraints on the baryonic flows (inflows/outflows) and the evolution of the CGM. The modest decline in $f(N_{HI})$ from $z=3$ to $z=0$ suggests a balance between decreasing UV background (increasing neutral fraction) and heating from feedback/collisional ionization (decreasing neutral fraction).
• Simulation Calibration: The discrepancy in impact parameters between observation and TNG50 highlights specific failures in current cosmological simulations regarding the spatial extent of cool gas and the identification of host galaxies for absorbers.
• Methodological Advance: The study demonstrates that emission-line imaging can effectively probe the LLS regime at $z=0$, offering a complementary and potentially more complete view of the HI universe than absorption-line studies alone.
• Future Probes: The derived relation between impact parameter, column density, and HI radius ($r_{001}$) offers a new method to estimate the HI mass of galaxies associated with absorption systems in future surveys.