Joint Optical-HI mock catalogs and prospects for upcoming HI surveys

Atomic hydrogen (HI) regulates star formation as cold gas fuels star formation. It represents a key phase of matter in the baryon cycle involving accretion, feedback, outflows, and gas recycling. Redshifted $21$ cm line emission originating from galaxies serves as a key tracer for investigating HI gas and its dynamics in the interstellar medium (ISM) and circumgalactic medium (CGM), and enables the study of galaxy evolution. Nonetheless, direct detections of HI are currently limited to $z \leq 0.4$ due to the inherently weak $21$ cm emission line. Ongoing and upcoming large radio surveys aim to detect $21$ cm emission from galaxies up to $z \gtrsim 1$ with unprecedented sensitivity. In current work, we present a novel approach for creating optical-HI joint mock catalogs for upcoming SKA precursor surveys: MIGHTEE-HI and LADUMA with MeerKAT and WALLABY with ASKAP. Incorporation of optical properties along with HI in our mock catalogs makes these a powerful tool for making predictions for upcoming surveys and provides a benchmark for exploring the HI science (e.g., conditional HIMF and optical-to-HI scaling relations) expected from these surveys. As a case study, we show the use of the joint catalogs for predicting the expected outcome of stacking detection for average HI mass in galaxies that are below the threshold for direct detection. We show that combining stacking observations with the number of direct detections puts a strong constraint on the HI mass function, especially in the regime where the number of direct detections is small, as often happens near the farther edge of HI surveys. This intermediate step may be used to set priors for the full determination of the HI mass function.

💡 Research Summary

The paper presents a novel method for constructing joint optical–HI mock catalogs tailored to upcoming SKA precursor surveys—MIGHTEE‑HI, LADUMA (MeerKAT) and WALLABY (ASKAP). Recognising that direct 21 cm detections are currently limited to z ≲ 0.4, the authors aim to provide realistic predictions for surveys that will push the detection horizon to z ≈ 1. Their approach departs from the usual reliance on full N‑body simulations; instead, they build the catalogs directly from observed statistical distributions.

First, they adopt the ALFALFA‑SDSS (A100‑SDSS) sample as a template. This dataset contains ~31 500 HI detections with matched SDSS photometry, providing a well‑characterised relationship between HI mass, optical luminosity, and colour. Using the observed HI mass function (Schechter form) they randomly assign HI masses to galaxies within a simulated survey cone. The Baryonic Tully‑Fisher relation supplies the 20 % line width (W20), while galaxy inclinations are drawn from a uniform cos i distribution. Disk sizes follow the recent D_HI–M_HI scaling of Wang et al. (2025). With these physical parameters they compute the signal‑to‑noise ratio (SNR) using the Arecibo sensitivity formula; sources with SNR > 6 are flagged as direct detections.

The mock catalog is validated against the real ALFALFA data. The predicted number of detections (~24 000) matches the RS02 HI mass function expectation (≈22 200) and reproduces the colour–magnitude bimodality seen in SDSS. The HI‑luminosity scaling also aligns with observations, confirming that the simple probabilistic framework can faithfully reproduce the key statistical properties of the local HI‑optical galaxy population.

Having demonstrated fidelity, the authors extend the method to the specifications of MIGHTEE‑HI and WALLABY. Survey‑specific parameters (integration time, beam size, frequency coverage) are incorporated to compute detection thresholds as a function of redshift. Direct detections are expected to drop dramatically beyond z ≈ 0.5, making stacking essential. By cross‑matching the mock galaxies with deep optical redshift surveys (e.g., DESI, COSMOS‑Web), they simulate HI stacking: the spectra of sub‑threshold galaxies are co‑added to recover an average HI mass.

Crucially, they show that combining the counts of direct detections with the stacked average mass provides a powerful Bayesian constraint on the HI mass function (HIMF). Even when only a handful of direct detections exist at the high‑z edge, the stacked signal sharply narrows the posterior on the low‑mass slope (α) and normalization (φ*). This synergy is demonstrated through mock likelihood analyses, indicating a reduction of HIMF parameter uncertainties by >30 % compared with using either method alone.

The paper acknowledges limitations: the mocks lack spatial clustering, which could affect environmental studies, and the optical colour‑luminosity assignment is based on a simple empirical model that may not capture rare populations (e.g., extreme low‑metallicity or strong AGN hosts). Nevertheless, for large‑area surveys where average statistics dominate, the omission of clustering is argued to be negligible.

In conclusion, the authors provide a computationally efficient, observation‑driven pipeline for generating joint optical‑HI mock catalogs. These catalogs enable realistic forecasts of detection rates, stacking signal‑to‑noise, and, most importantly, the joint exploitation of direct and stacked measurements to tightly constrain the evolution of the HI mass function. The work therefore offers a valuable tool for the planning and scientific exploitation of upcoming SKA precursor surveys and sets the stage for the full SKA era.