🔭 astrophysics

Cross-Comparison of Galaxies Detected in the CSST Spectroscopic Survey and the SKA HI Survey

This paper presents a forward-modeling framework using the L-Galaxies 2020 model and Millennium-II simulation to forecast and cross-compare galaxy samples from the CSST spectroscopic and SKA HI surveys, thereby assessing their synergy and characterizing the relationships between cold gas, stellar content, and halo properties.

Original authors: Yingfeng Liu, Furen Deng, Wenxiang Pei, Haitao Miao, Qi Xiong, Shuanghao Shu, Xingchen Zhou, Qi Guo, Yan Gong, Yougang Wang, Xuelei Chen

Published 2026-05-05

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yingfeng Liu, Furen Deng, Wenxiang Pei, Haitao Miao, Qi Xiong, Shuanghao Shu, Xingchen Zhou, Qi Guo, Yan Gong, Yougang Wang, Xuelei Chen

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe as a giant, bustling city that has been growing and changing for billions of years. Astronomers want to understand how this city works: how the buildings (galaxies) are built, what fuel they use, and how they are connected.

This paper is like a simulation game where scientists build a "fake universe" to test two powerful new telescopes that haven't fully started working yet: the CSST (a Chinese space telescope) and the SKA (a massive radio telescope array).

Here is the story of what they did, explained simply:

1. Building the Fake Universe (The Simulation)

Before the real telescopes start taking pictures, the scientists needed to know what they should see. They used a super-computer to run a simulation called L-Galaxies.

Think of this as a video game engine that follows the rules of physics. It starts with a dark, empty box and fills it with dark matter, gas, and stars.
They didn't just make a static picture; they built a time machine (called a "lightcone"). This lets them watch the universe evolve from the past to the present, just like looking down a long hallway where the walls show different eras of history.
In this fake universe, they tracked two specific things for every galaxy:
1. The "Gas Tank" (HI): The amount of neutral hydrogen gas, which is the raw fuel for making new stars.
2. The "Engine Light" (Emission Lines): The bright glow from new stars being born, which the CSST can see.

2. The Two Different Cameras

The scientists simulated what these two telescopes would see, acting like two different photographers with different lenses:

The Radio Photographer (SKA): This telescope looks for the "Gas Tank." It listens for a specific radio signal (the 21cm line) emitted by cold hydrogen gas.
- The Challenge: It's hard to hear a whisper in a noisy room. The scientists had to simulate the "static" (noise) of the radio telescope and use a smart computer program (called SOFIA 2) to find the real signals hidden in the noise. They found that this telescope mostly sees the "heavy" galaxies with big gas tanks.
The Optical Photographer (CSST): This telescope looks for the "Engine Light." It takes pictures of the bright colors (like H-alpha and Oxygen) coming from young, hot stars.
- The Advantage: This camera is very sensitive. It can see galaxies that are too faint or too gas-poor for the radio telescope to detect. It essentially sees the "births" happening in the city.

3. Comparing the Photos (The Cross-Check)

Now, the scientists overlaid the two lists of galaxies to see how they matched up.

The Overlap: Most of the galaxies the radio telescope sees (the big gas tanks) are also seen by the optical telescope. They are the "rich" galaxies with lots of gas and lots of new stars.
The Difference: The optical telescope sees many more galaxies. It can spot the "poor" galaxies that have very little gas left. The radio telescope misses these because their gas tanks are too small to make a loud enough signal.
The Connection: They found a strong link: Galaxies with more gas tend to have more new stars. It's like a car with a full tank of gas driving faster.

4. The "Stacking" Trick (Hearing the Whisper)

What about the tiny galaxies that the radio telescope can't hear individually? They are too quiet.

The Analogy: Imagine trying to hear a single person whispering in a stadium. You can't. But if you ask 1,000 people to whisper at the exact same time, you might hear a collective hum.
The Method: The scientists took the "quiet" galaxies found by the optical telescope and mathematically "stacked" their radio signals on top of each other.
The Result: Even though individual small galaxies were invisible to the radio telescope, the stacked signal revealed that they do have gas, just not enough to be seen alone. This allowed them to measure the average fuel tank size of the "invisible" galaxies.

5. The Map of the City (Clustering)

Finally, they looked at how these galaxies are arranged in space.

They measured how "clumpy" the galaxies are. Do they hang out in groups, or are they spread out?
They found that the gas-rich galaxies (the ones the radio telescope sees) tend to live in slightly less crowded neighborhoods compared to the most massive galaxies. This helps scientists understand how the "city" of the universe is structured.

The Bottom Line

This paper didn't discover a new planet or a new type of star. Instead, it was a dress rehearsal.

It proved that when the real CSST and SKA telescopes start working in the late 2020s, they will work beautifully together.
The optical telescope will find the "where" and "when" (location and star formation), and the radio telescope will tell us the "how much" (gas content).
By combining them, we will get a complete picture of how galaxies grow, fueled by their gas, and how they evolve over time.

Technical Summary: Cross-Comparison of Galaxies Detected in the CSST Spectroscopic Survey and the SKA HI Survey

Problem Statement
Modern astrophysics increasingly relies on cross-comparing observations across different wavelength regimes to mitigate selection biases and survey-specific systematics, thereby providing joint constraints on cosmology and galaxy formation. While optical spectroscopic surveys trace star-forming regions and neutral hydrogen (HI) surveys trace the cold gas reservoirs, a realistic connection between theoretical predictions and forthcoming multi-wavelength data is required. Specifically, there is a need to forecast the synergy between the Chinese Space Station Survey Telescope (CSST), which will conduct slitless spectroscopy to detect emission-line galaxies (ELGs), and the Square Kilometre Array (SKA-Mid), which will map the cosmic distribution of HI via the 21 cm line. This work addresses the challenge of constructing realistic mock observations to predict the overlap, correlations, and statistical properties of galaxy samples detectable by both instruments in the redshift range $0.2 \lesssim z \lesssim 1.6$ (specifically focusing on $0.235 < z < 0.495$ ).

Methodology
The authors employ a forward-modeling framework based on the Millennium-II (MS-II) N-body simulation and the L-Galaxies 2020 semi-analytic model (SAM). The methodology proceeds through the following steps:

Simulation and Physical Modeling:
- The L-Galaxies 2020 model is run on the MS-II simulation (rescaled to Planck cosmology) to generate galaxy populations with physical properties.
- Cold gas is partitioned self-consistently into atomic (HI) and molecular ( $H_2$ ) components using a shielding model dependent on gas surface density and metallicity.
- Star formation rates (SFR) are derived from the $H_2$ surface density.
- Emission-line luminosities ( $H\alpha$ , $H\beta$ , $[OIII]$) are computed in a post-processing step using the BC03 stellar population synthesis model and radiative transfer calculations (CLOUDY) to determine line ratios relative to $H\alpha$ . AGN contributions are not explicitly included.
Lightcone Construction:
- Using the Mock Map Facility (MoMaF) approach, static simulation snapshots are converted into a continuous lightcone extending to $z=0.5$ .
- The simulation box is replicated, rotated, and translated to cover the desired solid angle, with random tiling applied to mitigate repetition artifacts while managing finite-volume effects.
- Redshift-space distortions (RSD) are incorporated by adding the line-of-sight peculiar velocity to the cosmological redshift.
Mock Observations:
- SKA-Mid HI Survey: A realistic HI data cube is generated for a 2000-hour observation covering $\sim 20$ deg $^2$ (frequency 950–1150 MHz). HI fluxes are converted from masses, and spatial/morphological distributions are assigned by matching simulated galaxies to observed HI templates from the THINGS and HALOGAS surveys. Thermal noise and instrumental beam effects (FWHM = 7") are added. The Source Finding Application 2 (SOFIA 2) is used to perform blind source finding, with parameters optimized via grid search to maximize detection reliability.
- CSST Spectroscopic Survey: A catalog of ELGs is generated by applying the CSST selection function and noise model. Detection is defined by an effective signal-to-noise ratio ( $SNR_{eff}$ ) $> 5$ for at least one strong emission line ( $H\alpha$ , $H\beta$ , or $[OIII]$).
Analysis:
- The resulting catalogs are cross-matched to identify galaxies detectable by both surveys.
- Statistical analyses include correlations between HI mass, stellar mass, halo mass, and SFR; the Baryonic Tully-Fisher Relation (BTFR); stacking analysis of HI signals for CSST-selected galaxies below the individual SKA detection threshold; and the measurement of the optical-HI cross-correlation power spectrum to derive galaxy bias.

Key Contributions and Results

Catalog Generation: The study produces a realistic mock catalog of 38,527 HI detections (with a global reliability of 95.7%) and a corresponding CSST ELG catalog. The SKA detection limit is found to be approximately 6 Jy Hz.
Detection Characteristics:
- The CSST spectroscopic survey is more sensitive than the SKA HI survey, detecting galaxies with lower HI masses and lower halo masses.
- Most SKA-detectable galaxies are also detectable by CSST, but there is a population of high-HI-mass, low-halo-mass galaxies detectable by SKA but not CSST.
- CSST detection correlates tightly with SFR, whereas SKA detection correlates with HI mass.
Scaling Relations:
- A positive correlation is found between SFR and HI mass, though with significant scatter.
- The Baryonic Tully-Fisher Relation (BTFR) is confirmed for the HI-detected sample, showing a strong correlation between total baryonic mass and maximum rotation velocity.
Stacking Analysis:
- Stacking the HI signal for galaxies selected by CSST (which may not be individually detected by SKA) successfully recovers the mean HI content for galaxies with stellar masses $M_* \lesssim 10^9 M_\odot$ , a regime below the individual SKA detection threshold.
- The stacked HI masses broadly follow the intrinsic relation, though uncertainties increase at the lowest stellar masses due to noise dominance.
Clustering and Bias:
- The cross-correlation power spectrum between the optical (CSST) and HI (SKA) catalogs is measured.
- The linear bias ( $b_{HI}$ ) of the HI-selected galaxies is derived. The results indicate that HI galaxies trace relatively low-density regions (low bias) and that the bias increases with redshift, consistent with the evolution of host halo masses.

Significance
The paper claims that these results provide useful insights into the cold gas and stellar content of galaxies at low redshifts. By establishing a forward-modeling framework that incorporates realistic instrumental effects and selection functions, the work enables:

Synergy Assessment: A quantitative evaluation of the complementarity between CSST and SKA-Mid, demonstrating how optical selection can probe HI content in galaxies otherwise undetectable in radio.
Methodological Validation: The successful application of SOFIA 2 and MoMaF to generate realistic mock catalogs serves as a testbed for future analysis pipelines.
Cosmological Constraints: The measurement of the optical-HI cross-correlation power spectrum and the derived bias of HI galaxies offer a pathway to constrain galaxy formation models and cosmological parameters using future multi-wavelength survey data.

The authors maintain a modest tone, noting that the omission of AGN contributions in the emission-line model may lead to an underestimation of line luminosities at the bright end, and that the stacking results in the lowest mass bins are less well-constrained due to noise. The work is presented as a preparatory step to interpret the upcoming data from CSST and SKA.