MeerKAT Absorption Line Survey (MALS)

• MALS is a multifaceted radio survey that investigates intervening and associated H I and OH absorption to trace cold gas evolution from 0<z<2.
• It employs wide-band spectroscopy, targeted deep continuum imaging, and preparatory optical spectroscopy to maximize absorber detection and characterization.
• The survey has produced significant public data releases cataloguing continuum sources and H I absorption features, establishing a rich legacy for cold-gas research.

MeerKAT Absorption Line Survey (MALS) is a MeerKAT continuum and spectral-line survey primarily designed to carry out a search for intervening and associated neutral hydrogen (H I) 21-cm and hydroxyl radical (OH) 18-cm absorption lines at $0AGN), and deep commensal continuum imaging, so it functions simultaneously as an absorption survey, a continuum survey, and a Galactic H I absorption survey (Gupta et al., 2017). Public MALS products already include Stokes $I$ continuum catalogues, wideband continuum catalogues, and a Galactic H I 21-cm absorption catalogue (Deka et al., 2023, Wagenveld et al., 2024, Gupta et al., 31 Mar 2025).

1. Scientific rationale and survey definition

The original survey definition placed MALS in the context of the mismatch between the strong evolution of the cosmic star-formation-rate density and the much weaker evolution inferred for the total neutral atomic gas density. The central premise was that the key evolving quantity is not simply the total H I reservoir, but the fraction of gas in the cold atomic and molecular phases from which stars actually form. MALS was therefore designed to measure the evolution of the cross-section, or incidence, of cold atomic and molecular gas in galaxies over $0Gupta et al., 2017).

In this framework, H I 21-cm absorption is a tracer of the cold neutral medium, while the OH main and satellite lines trace diffuse and dense molecular gas. The survey definition also distinguishes two absorber classes. Intervening absorbers arise in galaxies along the line of sight to a more distant radio source and are the primary sample for measuring cosmic cold-gas incidence. Associated absorbers arise in gas physically linked to the radio-loud AGN itself, including circumnuclear gas, host-galaxy interstellar medium, inflows, and outflows (Gupta et al., 2017).

The baseline survey-definition paper described a program of 740 L-band pointings and 370 UHF-band pointings, with 56 minutes per L-band pointing and 121 minutes per UHF-band pointing, for a total of 1655 hr including a 15% calibration overhead. The intended frequency coverage was 900–1670 MHz in L band and 580–1015 MHz in UHF band, giving access to H I over $0Gupta et al., 2017). Later MALS descriptions summarize the full survey as approximately 500 pointings at Galactic latitude $1.1^\circ < b < 84^\circ$, observed in MeerKAT L band and/or UHF band (Himes et al., 2024). This difference reflects survey evolution between the original design paper and later operational descriptions.

The survey-definition paper adopted the standard optically thin H I conversion

$$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$

and stated a target $5\sigma$ integrated optical-depth sensitivity of

$\int \tau\,dv = 0.045\ {\rm km\,s^{-1}},$

corresponding to sensitivity to cold H I with $N({\rm H\,I})\sim10^{19}\ {\rm cm^{-2}}$ for $I$0 K (Gupta et al., 2017). On that basis, MALS forecast roughly $I$1 intervening H I 21-cm absorbers, the first meaningful statistical sample of intervening OH absorbers, roughly $I$2 associated H I absorbers in AGN, and $I$3–$I$4 H I emitters (Gupta et al., 2017).

2. Bright-background targeting and preparatory spectroscopy

MALS depends on bright, compact continuum backgrounds. The original design emphasized that compact background radio structure matters because covering factor and absorption detectability are higher toward flat-spectrum radio quasars (FSRQs), and it proposed that L-band pointings should be centered on bright sources at $I$5, while UHF-band pointings should be centered on bright sources at $I$6 (Gupta et al., 2017). Later survey descriptions state that pointings are centered on bright radio AGN, specifically sources brighter than 200 mJy at 1 GHz, and note that MeerKAT’s field of view is 88 arcmin FWHM at 1 GHz, which makes deep commensal continuum science unavoidable rather than incidental (Himes et al., 2024).

Two optical spectroscopic campaigns supplied the source-selection backbone. “MALS-NOT: Identifying Radio-Bright Quasars for the MeerKAT Absorption Line Survey” selected southern radio sources using a single flux-density limit $I$7 mJy together with WISE color cuts

$I$8

and spectroscopically observed 99 candidates with the Nordic Optical Telescope. It identified 72 quasars, of which 64 have $I$9 and 48 have $0Krogager et al., 2018).

The later SALT–NOT campaign extended that effort to a larger, dust-unbiased, MIR-selected sample of powerful radio-bright AGN. It produced a sample containing 250 AGN with emission lines, 26 with no emission lines, and 27 without optical counterparts, with a median redshift of 1.8. The paper also summarized the full MALS central-AGN pool as about 650 AGN, giving the practical target base from which the survey pointings were assembled (Gupta et al., 2021). An important consequence is that MALS background sources are not merely bright; they are characterized by spectroscopic redshifts, optical classes, and MIR properties that are directly relevant for interpreting both associated and intervening absorption.

3. Data processing and public releases

The computational scale of MALS led to the development of ARTIP, the Automated Radio Telescope Imaging Pipeline, written in Python 3.6 and built extensively on CASA. ARTIP was designed to enable the geographically distributed MALS team to process large volumes of radio interferometric data collaboratively, and early MALS work demonstrated its use on both MeerKAT and uGMRT data for blind H I and OH absorption searches (Gupta et al., 2020). The pipeline structure separates calibration, spectral-line cube production, continuum imaging, and diagnostics, and later MALS releases continued to use ARTIP for survey data products (Deka et al., 2023).

Three public releases define the present survey state.

Release	Content	Key figures
DR1 (Deka et al., 2023)	Stokes $0	495,325 sources over 2289 deg$0
DR2 (Wagenveld et al., 2024)	Wideband L-band continuum catalogues from the same 391 pointings	4344 deg$010 $0
DR3 (Gupta et al., 31 Mar 2025)	Blind Galactic H I 21-cm absorption catalogue	19130 radio sight lines, 3640 H I absorption features

DR1 established the first large MALS continuum reference. It released Stokes $0Deka et al., 2023).

DR2 converted the same 391 L-band pointings into the first full wideband continuum release. It used multi-term multi-frequency synthesis over a usable bandwidth of 802.5 MHz and delivered wideband source catalogues, Gaussian-component catalogues, primary-beam-corrected Stokes $0Wagenveld et al., 2024).

DR3 turned the same survey geometry into a Galactic absorption survey. Using 390 pointings and 19130 unique sight lines, it detected 3640 Galactic H I absorption features, which the paper describes as the largest Galactic H I absorption line catalog to date. The central sight lines reach much better optical-depth sensitivity than the full source ensemble, so DR3 also formalized the distinction between high-quality central spectra and the much larger but shallower off-axis sample (Gupta et al., 31 Mar 2025).

4. Intervening and associated absorption science

Early MALS science verification already showed the survey’s wide-band absorption capability. The first MeerKAT UHF spectrum, obtained toward PKS1830−211, detected the known H I 21-cm and OH 18-cm main lines at $0Combes et al., 2021).

MALS science-verification work also demonstrated sensitivity to time-variable cold gas structure in known high-redshift absorbers. Toward the $0PKS 2355−106, MALS found the emergence of a new H I 21-cm absorption component at

$0

in addition to the previously known component at $0Srianand et al., 2022).

A different MALS case study, PKS1413+135 at $0Combes et al., 2022).

MALS has also begun to populate the rare class of high-redshift associated absorbers. “MALS discovery of a rare H I 21-cm absorber at $0Deka et al., 2023).

MALS has also been used to extend associated-absorption work to lower-power radio AGN at $0$1.1^\circ < b < 84^\circ$0 integrated-optical-depth threshold of 10.0 km s$1.1^\circ < b < 84^\circ$1, the overall detection rate was

$1.1^\circ < b < 84^\circ$2

The paper found no statistically significant dependence on excitation class, no compelling evidence for redshift evolution out to $1.1^\circ < b < 84^\circ$3, and line kinematics consistent with disturbed cold gas driven by lobe expansion or jet activity (Deka et al., 21 Apr 2026).

5. Galactic and nearby-environment absorption

DR3 showed that MALS is also a major Galactic absorption survey. The authors detected 3640 H I absorption features toward 19130 radio sources and argued that these lines form a homogeneous sample of H I clouds in the local interstellar medium. Two pieces of evidence were emphasized: the strong correlation between H I emission-line column density $1.1^\circ < b < 84^\circ$4 and visual extinction $1.1^\circ < b < 84^\circ$5, and the confinement of the absorption features to a narrow velocity range,

$1.1^\circ < b < 84^\circ$6

The paper further found that H I 21-cm optical depth is linearly correlated with $1.1^\circ < b < 84^\circ$7 and $1.1^\circ < b < 84^\circ$8 up to about $1.1^\circ < b < 84^\circ$9 mag, above which $$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$0 traces the total hydrogen content and $$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$1 and $$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$2 scale differently (Gupta et al., 31 Mar 2025).

A physically important DR3 result is the contrast between central sight lines with and without absorption. The slopes of the $$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$3 distributions differ at $$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$4, and the paper explicitly compares this to the difference between H$$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$5 detections and non-detections in damped Ly$$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$6 systems at $$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$7. The proposed interpretation is that turbulence-driven WNM-to-CNM conversion is the common governing factor for the presence of both H I 21-cm and H$$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$8 absorption (Gupta et al., 31 Mar 2025). Central versus off-axis comparisons also showed that the slope of rms fluctuations in optical-depth variations in quiescent LISM gas is shallower than earlier measurements in the Galactic disk, with inferred densities of 20–30 cm$$N({\rm H\,I}) = 1.823 \times 10^{18}\,\frac{T_s}{f_c}\int \tau(v)\,dv\ {\rm cm^{-2}},$$9, typical of CNM gas (Gupta et al., 31 Mar 2025).

MALS-relevant nearby-environment science has also extended beyond the Milky Way proper. A later paper used MALS data to detect H I 21-cm absorption from Magellanic Bridge gas toward J033242.97−724904.5 with a peak signal-to-noise ratio of 10. Together with earlier ATCA detections, the new line allowed the authors to compare cold atomic gas over 4–6 kpc. The gas-to-dust ratio along the new sight line was measured as

$5\sigma$0

which they argued is more consistent with Large Magellanic Cloud than Small Magellanic Cloud gas, supporting a direct LMC–SMC collision scenario for the Bridge (Morelli et al., 26 Sep 2025). This nearby application shows that the MALS observing strategy is sensitive not only to cosmological absorbers but also to weak H I absorption in the Local Group environment.

6. Continuum legacy and commensal science

Although conceived as an H I/OH absorption survey, MALS has developed into a substantial continuum-survey resource. DR1 already showed that 391 L-band pointings yielded hundreds of thousands of radio sources, in-band spectral indices, 140 ultra steep-spectrum sources as prospective high-$5\sigma$1 radio galaxies, 1308 long-term variable sources, and 122 transients (Deka et al., 2023). The later Fermi-unassociated-source study made this broader significance explicit by using MALS DR1 as the starting point for image-based searches for steep-spectrum radio counterparts to $5\sigma$2-ray sources. Within Fermi error ellipses, the authors identified nine pulsar candidates toward six Fermi sources and 41 steep-spectrum radio galaxy candidates, and argued that MALS can detect 80 percent of the known pulsar population, largely because its in-band spectral indices suppress false positives from non-contemporaneous multi-survey flux comparisons (Himes et al., 2024).

DR2 demonstrated that MALS continuum products are also precise enough for cosmological number-count work. Using the 971,980-source wideband catalogue, the DR2 paper modeled completeness, false detections, and a declination-dependent source-density systematic, then measured the cosmic radio dipole. Its preferred result at a 400 $5\sigma$3Jy flux-density cut was

$5\sigma$4

with direction

$5\sigma$5

which the authors described as consistent with the cosmic microwave background dipole in both direction and amplitude (Wagenveld et al., 2024). The same paper effectively established MALS as a bridge survey between shallow all-sky cm-wave catalogues and very small ultra-deep fields.

Morphological radio-source studies have likewise benefited from the MALS continuum footprint. A 2025 study reported 36 new hybrid morphology radio sources from MALS radio maps, described as the largest collection of such sources in the southern sky to date. The sample spans $5\sigma$6, has a median 1.4 GHz luminosity of $5\sigma$7 W Hz$5\sigma$8, and includes nine sources near the centers of galaxy clusters, one with radio jets extending over 811 kpc (Manik et al., 25 Jun 2025). In combination with the Fermi, DR2, and DR1 continuum papers, this establishes MALS as a survey whose continuum products are scientifically significant in their own right.

7. Limitations, evolution, and prospects

Several MALS papers emphasize that the survey is still an evolving data system rather than a finalized static dataset. DR1 spectral indices are derived from only two L-band spectral windows, centered at 1006.0 and 1380.9 MHz, not from full-band spectral modeling, and catalogue-level compactness can fail in individual cases, so visual validation remains necessary in some applications (Himes et al., 2024). The Fermi counterpart work also showed that pulsar detectability in imaging is limited by variability and scintillation over the 56-minute MALS integrations, so raw continuum sensitivity does not translate directly into complete pulsar recovery (Himes et al., 2024).

The continuum releases have likewise required explicit treatment of systematics. Early MALS homogeneity work on ten pointings showed that deep source counts are sensitive to primary-beam behavior, flux-recovery bias, resolved-source incompleteness, and bright central-source artifacts, but also concluded that these effects can be modeled well enough for corrected number counts down to 100–200 $5\sigma$9Jy (Wagenveld et al., 2023). DR2 then identified a declination-dependent source-density systematic of up to 5%, traced operationally through the restoring-beam major axis, and corrected it before the radio-dipole analysis (Wagenveld et al., 2024). These results imply that MALS continuum science is strongest when accompanied by release-specific completeness and quality modeling.

The survey’s implemented geometry has also evolved relative to its original design forecasts. The original baseline paper described 1110 pointings split between L and UHF bands (Gupta et al., 2017), while later operational descriptions refer to approximately 500 pointings (Himes et al., 2024), and current public continuum releases are based on 391 L-band pointings (Deka et al., 2023, Wagenveld et al., 2024). This suggests a transition from design-stage forecast to staged release practice rather than a single immutable survey specification.

Even so, the trajectory is clear. Later MALS discussions state that future releases will include wideband and full-Stokes images for each targeted field in both L and UHF bands, with particular interest in Stokes $\int \tau\,dv = 0.045\ {\rm km\,s^{-1}},$0 for pulsar candidate identification (Himes et al., 2024). The same discussion anticipated DR2 wideband continuum images for the 391 L-band pointings with rms noise as low as 10 $\int \tau\,dv = 0.045\ {\rm km\,s^{-1}},$1Jy beam$\int \tau\,dv = 0.045\ {\rm km\,s^{-1}},$2 near the pointing centers and 971,980 sources over 4344 deg$\int \tau\,dv = 0.045\ {\rm km\,s^{-1}},$3, which DR2 then delivered (Himes et al., 2024, Wagenveld et al., 2024). In that sense, MALS has already moved from survey concept to multi-release archive, but its final value still depends on the full absorber catalogues, the remaining UHF and full-Stokes products, and the eventual merger of its absorption-line and continuum legacies into a unified cold-gas survey resource.