Lyα Nebulae in HETDEX: The Largest Statistical Census Bridging Lyα Halos and Blobs across Cosmic Noon - IOPscience

Lyα Nebulae in HETDEX: The Largest Statistical Census Bridging Lyα Halos and Blobs across Cosmic Noon - IOPscience
The American Astronomical Society (AAS)
, established in 1899 and based in Washington, DC, is the major organization of professional astronomers in North America. Its membership of about 7,000 individuals also includes physicists, mathematicians, geologists, engineers, and others whose research and educational interests lie within the broad spectrum of subjects comprising contemporary astronomy. The mission of the AAS is to enhance and share humanity's scientific understanding of the universe.
The following article is
Open access
Ly
Nebulae in HETDEX: The Largest Statistical Census Bridging Ly
Halos and Blobs across Cosmic Noon
Erin Mentuch Cooper
Karl Gebhardt
Dustin Davis
Robin Ciardullo
Chris Byrohl
Chenxu Liu
(刘辰旭)
Maya H. Debski
Óscar A. Chávez Ortiz
Maximilian Fabricius
Daniel J. Farrow
Published 2026 March 11
© 2026. The Author(s). Published by the American Astronomical Society.
The Astrophysical Journal
Number 1
Citation
Erin Mentuch Cooper
et al
2026
ApJ
1000
38
DOI
10.3847/1538-4357/ae44f3
Article
PDF
Article
ePub
You need an eReader or compatible software to experience
the benefits of the ePub3 file format
Erin Mentuch Cooper
AFFILIATIONS
Department of Astronomy, The University of Texas at Austin, 2515 Speedway Boulevard, Austin, TX 78712, USA; erin.hetdex@gmail.com
EMAIL
erin.hetdex@gmail.com
Karl Gebhardt
AFFILIATIONS
Department of Astronomy, The University of Texas at Austin, 2515 Speedway Boulevard, Austin, TX 78712, USA; erin.hetdex@gmail.com
Dustin Davis
AFFILIATIONS
Department of Astronomy, The University of Texas at Austin, 2515 Speedway Boulevard, Austin, TX 78712, USA; erin.hetdex@gmail.com
Robin Ciardullo
AFFILIATIONS
Department of Astronomy & Astrophysics, The Pennsylvania State University, University Park, PA 16802, USA
Institute for Gravitation and the Cosmos, The Pennsylvania State University, University Park, PA 16802, USA
Chris Byrohl
AFFILIATIONS
Kavli Institute for the Physics and Mathematics of the Universe (WPI), University of Tokyo, Kashiwa, Chiba 277-8583, Japan
Institut für Theoretische Astrophysik, ZAH, Universität Heidelberg, Albert-Ueberle-Str. 2, 69120 Heidelberg, Germany
Chenxu Liu
(刘辰旭)
AFFILIATIONS
South-Western Institute for Astronomy Research, Key Laboratory of Survey Science of Yunnan Province, Yunnan University, Kunming, Yunnan 650500, People’s Republic of China
Maya H. Debski
AFFILIATIONS
Department of Astronomy, The University of Texas at Austin, 2515 Speedway Boulevard, Austin, TX 78712, USA; erin.hetdex@gmail.com
Óscar A. Chávez Ortiz
AFFILIATIONS
Department of Astronomy, The University of Texas at Austin, 2515 Speedway Boulevard, Austin, TX 78712, USA; erin.hetdex@gmail.com
Maximilian Fabricius
AFFILIATIONS
Max Planck Institute for Extraterrestrial Physics, Giessenbachstr. 1, 85748 Garching, Germany
Universitäts-Sternwarte München, Fakultät für Physik, Ludwig-Maximilians-Universität München, Scheinerstrasse 1, 81679 München, Germany
Daniel J. Farrow
AFFILIATIONS
Centre of Excellence for AI, Data Science and Modelling (DAIM), University of Hull, Cottingham Road, Hull, HU6 7RX, UK
E. A. Milne Centre for Astrophysics, University of Hull, Cottingham Road, Hull, HU6 7RX, UK
Steven L. Finkelstein
AFFILIATIONS
Department of Astronomy, The University of Texas at Austin, 2515 Speedway Boulevard, Austin, TX 78712, USA; erin.hetdex@gmail.com
Caryl Gronwall
AFFILIATIONS
Department of Astronomy & Astrophysics, The Pennsylvania State University, University Park, PA 16802, USA
Institute for Gravitation and the Cosmos, The Pennsylvania State University, University Park, PA 16802, USA
Gary J. Hill
AFFILIATIONS
Department of Astronomy, The University of Texas at Austin, 2515 Speedway Boulevard, Austin, TX 78712, USA; erin.hetdex@gmail.com
McDonald Observatory, The University of Texas at Austin, 2515 Speedway Boulevard, Austin, TX 78712, USA
Maja Lujan Niemeyer
AFFILIATIONS
Max-Planck-Institut für Astrophysik, Karl-Schwarzschild-Str. 1, 85741 Garching, Germany
Brianna McKay
AFFILIATIONS
Department of Astronomy, University of Washington, Seattle, 3910 15th Avenue NE, Room C319, Seattle, WA 98195-0002, USA
Shiro Mukae
AFFILIATIONS
Department of Astronomy, The University of Texas at Austin, 2515 Speedway Boulevard, Austin, TX 78712, USA; erin.hetdex@gmail.com
MIRAI Technology Institute, Shiseido Co., Ltd., 1-2-11, Takashima, Nishi-ku, Yokohama, Kanagawa 222-0011, Japan
Masami Ouchi
AFFILIATIONS
Kavli Institute for the Physics and Mathematics of the Universe (WPI), University of Tokyo, Kashiwa, Chiba 277-8583, Japan
National Astronomical Observatory of Japan, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan
Institute for Cosmic Ray Research, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8582, Japan
Department of Astronomical Science, SOKENDAI (The Graduate University for Advanced Studies), Osawa 2-21-1, Mitaka, Tokyo 181-8588, Japan
Huub Röttgering
AFFILIATIONS
Leiden Observatory, Leiden University, P.O. Box 9513, 2300 RA Leiden, The Netherlands
Donald P. Schneider
AFFILIATIONS
Department of Astronomy & Astrophysics, The Pennsylvania State University, University Park, PA 16802, USA
Institute for Gravitation and the Cosmos, The Pennsylvania State University, University Park, PA 16802, USA
Sarah Tuttle
AFFILIATIONS
Department of Astronomy, University of Washington, Seattle, 3910 15th Avenue NE, Room C319, Seattle, WA 98195-0002, USA
Lutz Wisotzki
AFFILIATIONS
Leibniz-Institut for Astrophysik Potsdam (AIP), An der Sternwarte 16, 14482 Potsdam, Germany
Gregory Zeimann
AFFILIATIONS
Hobby Eberly Telescope, University of Texas, Austin, Austin, TX 78712, USA
Sai Zhai
AFFILIATIONS
Leiden Observatory, Leiden University, P.O. Box 9513, 2300 RA Leiden, The Netherlands
Article metrics
1349
Total downloads
Video abstract views
Share this article
Dates
Received
2025 October 31
Revised
2026 January 19
Accepted
2026 February 4
Published
2026 March 11
Unified Astronomy Thesaurus concepts
Emission line galaxies
Lyman-alpha galaxies
High-redshift galaxies
Emission nebulae
Journal RSS
Create or edit your corridor alerts
What are corridors?
0004-637X/1000/1/38
Abstract
The Hobby–Eberly Dark Energy Experiment (HETDEX) is an untargeted ∼540 deg
spectroscopic survey of Ly
emission in the 1.9 <
< 3.5 Universe. In surface brightness, this survey reaches 1
Ly
sensitivities of ∼2–5 × 10
−18
erg s
−1
cm
, allowing large samples of extended Ly
nebulae (LAN) to be studied. We selected a sample of 70,691 Ly
-emitting galaxies (LAEs) with an emission-line signal-to-noise ratio greater than 6 and modeled the Ly
emission as a point-source component with an optional exponential envelope. Half (∼47.5%) of the LAE sample (33,612 objects) exhibits significant extended emission and is best fit by the two-component model. The fraction of resolved sources increases with Ly
flux and luminosity. Their isophotal areas range from 10 to 130 arcsec
(median 15 arcsec
), with integrated Ly
fluxes from 6 to 2000 × 10
−17
erg s
−1
cm
−2
(median 20 × 10
−17
erg s
−1
cm
−2
). Comparison between point-spread function-weighted and isophotal flux measurements shows that the HETDEX pipeline underestimates the total Ly
flux by ∼30% on average, reflecting the substantial halo contribution in extended sources. Approximately 420 LANs are found per deg
over 79.5 deg
of noncontiguous sky. About 12% of resolved sources show active galactic nuclei signatures and are bright in Ly
and continuum. The remaining 88% span a wide range of morphologies and often lack continuum counterparts. Exponential scale lengths show no strong correlation with Ly
flux or luminosity (median 11.6 ± 1.9 kpc). Only 2.9% of the full S/N > 6 LAE population with ancillary data have radio counterparts, but 64% of those are found to be extended, with the radio fraction increasing with Ly
size. We present a catalog of all modeled sources, with their positions, redshifts, luminosities, and structural parameters for over 70,000 LAEs consisting of 33,000 spatially extended LAN. The catalog can be found at
and in the online version of this paper.
Export citation and abstract
BibTeX
RIS
Previous
article in issue
Next
article in issue
NASA ADS Record
About Related Links
Original content from this work may be used under the terms of the
Creative Commons Attribution 4.0 licence
. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
1. Introduction
Lyα emission is one of the most prominent spectral features in the high-redshift Universe, observable from the ground as the UV Ly
line is redshifted into the optical. The line’s strength enables the detection of galaxies that are too faint to be seen in broadband continuum surveys, as ionizing photons from young stars and active nuclei are reprocessed into Ly
emission through hydrogen recombination. Not only does Ly
trace stellar and active galactic nuclei (AGN) light within galaxies, but it can also reveal ionized and resonantly scattered emission in the surrounding circumgalactic medium.
Spatially extended Ly
emission is observed across a wide range of physical scales and surface brightnesses. Lyα emitters (LAEs), galaxies detected through their strong Ly
emission, are often accompanied by faint Lyα halos (LAHs) on scales up to tens of proper kiloparsecs, kpc, observed across redshifts at ∼10
−19
erg s
−1
cm
sensitivities (C. C. Steidel et al.
2011
; M. Hayes et al.
2014
; L. Wisotzki et al.
2016
; F. Leclercq et al.
2017
; M. Lujan Niemeyer et al.
2022a
). LAHs might not only be a common feature of LAEs, but also ubiquitous for star-forming galaxies (M. Lujan Niemeyer et al.
2022b
). For massive systems, bright extended Ly
structures are detected on scales ranging from tens to hundreds of kiloparsecs—and in some cases approaching megaparsec scales—at surface-brightness levels of ∼10
−18
–10
−20
erg s
−1
cm
, revealing filamentary gas associated with Lyα blobs (LABs) and the cosmic web (e.g., Y. Matsuda et al.
2011
; S. Cantalupo et al.
2014
; J. F. Hennawi et al.
2015
; E. Borisova et al.
2016
; Z. Cai et al.
2017
; R. Bacon et al.
2021
; D. C. Martin et al.
2023
; D. Tornotti et al.
2025
).
A number of physical processes can source and give rise to extended Ly
structures, which we will collectively refer to as Ly
Nebulae (LANs) throughout this paper, encompassing both LAH and LAB nomenclature. Central star-forming galaxies and AGN produce a large number of Ly
photons following recombinations in their immediate vicinity. Subsequent resonant scatterings can illuminate the diffuse surroundings of these systems (Z. Zheng et al.
2010
; C. Byrohl et al.
2021
; C. Byrohl & D. Nelson
2023
). Satellite galaxies can provide an additional Ly
source, directly injected within the halo surroundings (L. Mas-Ribas et al.
2017
). In situ Ly
emission through recombinations and collisional excitations from the diffuse gas itself can be major contributors. Recombinations can be powered by escaping ionizing photons from the host halo galaxies and AGNs, as well as from the metagalactic ultraviolet background (S. Cantalupo et al.
2005
; J. A. Kollmeier et al.
2010
). Gravitational cooling, particularly through filamentary gas streams into the dark matter halos (A. Dekel et al.
2009
), can be a significant energy source with up to 50% of the cooling budget emitted through Ly
(M. A. Fardal et al.
2001
; M. Dijkstra & A. Loeb
2009
). Likely different physical processes contribute with different relative importance in different halo mass regimes. Large observational samples and their statistical analysis will disentangle these scenarios and their contributions.
Previous surveys have revealed populations of Ly
halos and nebulae using two complementary approaches. Narrowband imaging with Subaru/HSC, Very Large Telescope (VLT)/VIMOS, and DECam has identified thousands of emitters at discrete redshift slices (M. Ouchi et al.
2008
; D. Sobral et al.
2018
; M. Li et al.
2024
), including ∼300–500 extended Lyα nebulae in wide-area programs such as MAMMOTH-Subaru (M. Li et al.
2024
). The MAMMOTH-Subaru survey established the first statistical view of hundreds of LANs, showing that only a small fraction are AGN or radio associated, while most (∼80%) arise around UV-faint, likely dusty star-forming galaxies that may host obscured AGN (M. Li et al.
2024
). While these studies are pioneering in scale, narrowband imaging is limited to discrete redshift windows and relies on photometric emission-line selection.
Integral-field spectroscopy with instruments such as Multi-Unit Spectroscopic Explorer (MUSE; R. Bacon et al.
2015
) and Keck Cosmic Web Imager (KCWI; P. Morrissey et al.
2018
) enables 3D mapping of extended Ly
structures with exquisite sensitivity, but current surveys only cover preselected targets over small solid angles based on deep photometric imaging, as is done, for example, in MUSE (L. Wisotzki et al.
2016
; F. Leclercq et al.
2017
; H. Kusakabe et al.
2022
) and KCWI (Y. Chen et al.
2021
). These targeted Ly
studies have revealed the detailed morphologies and kinematics of Ly
halos but lack the statistical power to probe rare luminous systems or mitigate cosmic variance.
The Hobby–Eberly Telescope Dark Energy Experiment (HETDEX; K. Gebhardt et al.
2021
) provides a unique statistical window into extended Ly
structures. Its wide-area, random-tiling, integral-field spectroscopic survey enables the first truly volumetric census of Ly
-emitting galaxies and their extended emission, covering 540 deg
and 1.9 <
< 3.5 with surface-brightness sensitivities of ∼2–5 × 10
−18
erg s
−1
cm
. HETDEX aims to constrain the Hubble parameter
) and angular diameter distance
) to better than 1% precision at
∼ 2.4, using LAEs as biased tracers of the underlying dark matter distribution (M. Shoji et al.
2009
). To achieve this goal, the project surveys 540 deg
—corresponding to a comoving volume of 10.9 Gpc
—in a noncontiguous tiling optimized for sampling large-scale structure (C.-T. Chiang et al.
2013
). Beyond its cosmological goals, HETDEX’s sensitivity and area enable a unique statistical investigation of extended LANs. Despite the low signal-to-noise ratio of most detected LAEs, a subset exhibits spatially resolved emission, providing an unprecedented opportunity to study the frequency, morphology, and environmental dependence of extended Ly
structures while minimizing cosmic variance. This combination of depth, volume, and unbiased selection makes HETDEX uniquely suited to build the first statistical sample of extended Ly
sources at Cosmic Noon.
In this work, we perform 2D surface-brightness modeling of every LAE in the internal HETDEX Data Release 5 (HDR5) catalog detected at a signal-to-noise ratio greater than six, bridging the populations of compact Ly
halos and large Ly
blobs. A description of a smaller subset of this catalog is described in E. Mentuch Cooper et al. (
2023
), and the full catalog will be provided in an upcoming HETDEX Data Release paper in E. Mentuch Cooper et al. (
2026
). Section
describes the HETDEX observations and sample selection, Section
outlines the generation of Ly
line-flux maps, and Section
details the surface-brightness model fitting. The accompanying catalog is described in Section
and includes positions, redshifts, and morphological parameters for both compact and extended systems, providing the largest homogeneous dataset of its kind. A summary of properties and statistical insights gained from the sample is provided in Section
All positions reported in this paper are in the International Celestial Reference System (ICRS). We adopt the flat Λ-cold-dark-matter cosmology with
= 67.7 km s
−1
Mpc
−1
and Ω
m,0
= 0.31 measured by Planck Collaboration et al. (
2020
, Planck18). All quoted sizes are expressed as physical transverse distances. All magnitudes are expressed in the AB system (J. B. Oke & J. E. Gunn
1983
). We assume a rest-frame vacuum wavelength of
= 1215.67 Å for Ly
. Observed wavelengths expressed in this paper and associated data products are as measured in air. All redshifts are appropriately calculated for any differences between air and vacuum wavelengths using the standard in D. C. Morton (
1991
).
2. Observations and Sample Selection
HETDEX uses the Visible Integral Field Unit (IFU) Replicable Unit Spectrograph (VIRUS; G. J. Hill et al.
2021
) on the Hobby–Eberly Telescope (HET; L. W. Ramsey et al.
1998
; G. J. Hill et al.
2021
) to search for Ly
-emitting galaxies (LAEs) at a redshift of 1.9 <
< 3.5. Equipped with an array of 78 IFUs, VIRUS simultaneously obtains ∼35 K fiber spectra in the wavelength range 3500 Å ≲
≲ 5500 Å with spectral resolving power 750 <
< 950. HETDEX survey tiling takes three-dithered ∼6 minute exposures to fill in the gaps between fibers on each individual IFU. In just 20 minutes, HETDEX/VIRUS can detect over 150 LAEs in a ∼55 arcmin
area of sky.
Each VIRUS IFU covers a 51″ × 51″ field-of-view and consists of 448
-diameter fibers coupled to a dual-channel spectrograph. Observations are typically executed in a three-point dither pattern with 360 s exposures at each position. This yields near-complete spatial coverage within the IFU footprints. For further details about the HETDEX survey design and its data products, see K. Gebhardt et al. (
2021
), and for instrumentation details, see G. J. Hill et al. (
2021
).
The sample used in this study is drawn from HETDEX data collected between 2017 January 1 and 2024 July 31. In total, the dataset includes 6771 individual observations, each composed of a varying number of functional IFUs. The full sky distribution of IFU observations and the outline of the main field regions are found in Figure
. The HETDEX survey consists of two main fields: the 390 deg
HETDEX Spring field (“dex-spring”) and the 150 deg
HETDEX Fall field (“dex-fall”). It also contains a nonuniform tiling in several legacy fields. The most extensive coverage comes from collaborative observations with the Texas Euclid Survey for Ly
(Ó. A. Chávez Ortiz et al.
2023
) of the North Ecliptic Pole (NEP). Nearly full field coverage of the central 1 deg
of the Cosmic Evolution Survey (COSMOS; N. Scoville et al.
2007
) is included. As well as sparse coverage of the SA22 (C. C. Steidel et al.
1998
) and GOODS-N (M. Dickinson et al.
2002
) fields. Field coordinates, coverage area, and the number of IFU observations contained in each field are summarized in Table
Figure 1.
Sky distribution of HETDEX IFU observations and Ly
Nebula (LAN) sources across six HETDEX fields. The average sky number density of resolved LAEs is 420 deg
−2
(for the redshift range 1.9 <
< 3.5 covered by HETDEX). Each panel shows the HETDEX IFU footprints in light gray (with each tile corresponding to a 51″ × 51″ IFU), overlaid with spectroscopically confirmed LANs color coded by redshift (from
hetdex
= 1.9 to
hetdex
= 3.5). The top two panels display the wide-area Spring (
dex-spring
) and Fall fields (
dex-fall
), with rectangular red outlines marking the approximate survey boundaries; these two regions encompass approximately 390 and 150 deg
of sky area, respectively. The bottom four panels show targeted deep fields: COSMOS, GOODS-N, SA22, and the North Ecliptic Pole (NEP). The scale bars in the lower left of each panel indicate angular size.
Download figure:
Standard image
High-resolution image
Table 1.
Summary of HETDEX Field Coverage Sorted by Area
Field
(IFU)
Area
LAE
LAN
LAN
(deg
(deg
−2
DEX-SPRING
222,822
44.72
50,195
21,563
482.19
DEX-FALL
126,454
25.38
20,221
9206
362.75
NEP
31,707
6.36
4708
1851
290.89
COSMOS
10,139
2.03
1702
648
318.44
SSA22
4142
0.83
531
230
276.69
GOODS-N
648
0.13
265
114
876.56
TOTAL
395,911
79.46
77,622
33,612
423.02
Note.
Columns include the number of IFUs, survey area, number of LAEs and LANs, and the surface density of LANs Σ
LAN
in units of deg
−2
Download table as:
ASCII
Typeset image
Early in the survey, as few as 16 IFUs were installed, with some only partially functional. However, by 2022 the full complement of 78 IFUs was deployed, with an average of 72 IFUs providing science-quality coverage each night. As described in E. Mentuch Cooper et al. (
2023
), some data are lost due to detector issues, saturation from bright stars or galaxies, and contamination from meteors and satellite streaks. After quality cuts, a total of 395,911 science-grade IFU observations remain. A breakdown by survey field is provided in Table
HETDEX attempts to achieve uniform sensitivity across the survey area by adjusting exposure times based on conditions such as image quality, atmospheric transparency, and instrument throughput. Despite these efforts, the detection limit for Ly
emission varies significantly across the dataset, with Ly
surface-brightness sensitivities ranging from ∼2 to 5 × 10
−18
erg s
−1
cm
for the typical spatial (FWHM = 1
2 to 3
0) and spectral resolution (Δ
= 5.6 Å) of HETDEX. Fiber-to-fiber and IFU-to-IFU sensitivity variations are also present due to instrumental differences and aging effects across the array.
Our analysis sample comes from the fifth internal HETDEX data release (HDR5; v5.0.2 internally), which includes more than 600 million fiber spectra and includes ∼78 deg
of sky. The full LAE catalog includes over one million candidates and is being prepared for a dedicated HETDEX release paper (E. Mentuch Cooper et al.
2026
). For selection methods and classification details, see D. Davis et al. (
2023
) and E. Mentuch Cooper et al. (
2023
). From this parent catalog, we select a sample of moderate to high signal-to-noise ratio (S/N > 6) LAEs; objects fainter than this do not have enough counts for surface photometry measurements. This S/N measure comes from the HETDEX detection pipeline and comes from fitting a Gaussian model to the 1D extracted spectrum. See K. Gebhardt et al. (
2021
) for more details on the detection search and emission-line fitting methods.
To ensure robust surface-brightness profile fits, we exclude LAEs that fall within
of an IFU edge, i.e., those with
or
, where
ifu
ifu
is the detection positional distance from the IFU center in arcseconds. Additionally, to minimize contamination from nearby bright-continuum sources, we remove any LAE candidate within 5″ of a continuum source with
HETDEX
< 20, where
HETDEX
is the HETDEX
-band magnitude measured by convolving the point-spread function (PSF)-weighted spectrum with the Sloan Digital Sky Survey (SDSS)
-band filter response. These criteria reduce our final sample to 79,830 LAEs with clean, well-centered line emission.
3. Line Flux Maps
After defining our sample of bright LAEs, we generate their continuum-subtracted Ly
line-flux maps. These maps are created using a custom software package,
hetdex-api
21
developed to work with HETDEX data products. First, we construct a pseudonarrowband image at the detected central wavelength of an observed LAE by collapsing the fiber spectra in the spectral dimension around the emission line. This is done by collecting all fibers within a 20″ × 20″ region surrounding the emission-line detection position and taking the summed fiber spectra ±2
in the wavelength dimension centered on the emission line, where
is the best-fit Gaussian line width derived from the 1D spectral line fit from the HETDEX detection software as described in K. Gebhardt et al. (
2021
).
Our software flags any fiber spectral elements that are compromised by cosmic rays or other artifacts to NaN. The HET/VIRUS instrument does not contain an atmospheric diffraction corrector, so this correction must be applied in software. While each fiber is assigned a single sky coordinate in the HETDEX data model, the precise (R.A., decl.) for each element in the fiber spectra actually varies as a function of wavelength relative to its fiducial value of 4500 Å. Thus, a slight correction to the astrometry of up to ∼1″ across the 3470–5540 Å range is applied depending on the object’s wavelength. We then interpolate the fibers to an image of 0
25 per pixel using the
cubic
interpolation method from
scipy/interpolate/griddata
(P. Virtanen et al.
2020
) and convert the surface-brightness measurements to units of 10
−18
erg s
−1
cm
. We repeat this process for the fiber spectra uncertainty arrays to properly track measurement noise. As we collapse the spectra along the wavelength direction, we propagate the uncertainties by summing them in quadrature, ensuring that the final error maps reflect the combined contribution from all spectral channels.
Next, we subtract the continuum from each of these object spectra. The HETDEX data processing pipeline provides two sky-subtraction methods: one with a “local” sky-subtraction model, which uses sky fibers on a single IFU to model the sky, and the other using a “full-frame” sky-subtraction model, which considers all IFUs in the observation. (The 78 IFUs, which compose a single HETDEX/VIRUS observation, sample
of sky.) For our analysis, we use flux-calibrated fibers from the “local” sky-subtraction model, but because we perform continuum subtraction, the choice does not significantly affect our results. Using the local sky, we subtract the spectral continuum in the line-flux maps by generating two additional 50 Å-wide wavelength-collapsed images, ±10 Å from the emission-line map limits. We take an average of these two continuum images and subtract it from the Ly
line-flux map to create a continuum-subtracted emission-line-flux map. For broad-line AGN, we note that this continuum subtraction approach does not explicitly model spectral continuum slopes, but the continuum is estimated independently in each spatial element.
The continuum-subtracted surface-brightness sensitivity,
, is measured from the variance in the pixels in the continuum-subtracted line-flux map generated for each LAE detection. These maps have a pixel resolution of 0
25 and are collapsed spectrally based on the LAE detection’s fitted line width (
= 1.8–50 Å,
avg
= 4.4 Å). The value varies depending on observing conditions and wavelength, as well as subtle variations in the detector response. Figure
(left) shows
as a function of observed Ly
wavelength, color coded by the seeing full width at half-maximum (FWHM) for each exposure. The median trend (solid line) reveals a gradual increase in background noise toward shorter wavelengths. The median sensitivity across the survey is 3.8 × 10
−18
erg s
−1
cm
−2
arcsec
−2
, with 50% range of 3.1–4.8 × 10
−18
erg s
−1
cm
−2
arcsec
−2
. The sensitivity improves significantly at longer observed wavelengths where the instrument throughput is higher, and the sky background is lower. For wavelengths
obs
> 4000 Å, corresponding to
Ly
> 2.29, the median surface brightness limit is 3.4 × 10
−18
erg s
−1
cm
−2
arcsec
−2
, whereas for
obs
< 4000 Å (
Ly
< 2.29) it degrades to 5.3 × 10
−18
erg s
−1
cm
−2
arcsec
−2
due to reduced throughput and higher detector noise.
Figure 2.
surface-brightness sensitivity measured from the pixel variance in the continuum-subtracted line-flux maps and converted into intrinsic luminosity surface-density limits for individual HETDEX IFU observations. Left: the observed-frame 1
Ly
surface brightness,
(in units of 10
−18
erg s
−1
cm
−2
arcsec
−2
), as a function of observed wavelength. Each point represents an IFU exposure and is color coded by the image quality (FWHM) in arcseconds. The solid line traces the median trend in 50 Å bins, showing increased sensitivity toward the blue due to higher instrumental throughput and lower sky background. Middle: distribution of
values across all exposures, typically ranging from 2 to 10 × 10
−18
erg s
−1
cm
−2
arcsec
−2
. Right: histogram of the intrinsic luminosity surface-density limit,
, derived from each exposure using the luminosity distance and physical scale at its redshift. These limits correspond to
–40.1 erg s
−1
kpc
−2
, illustrating the range of Ly
sensitivity achieved across the survey.
Download figure:
Standard image
High-resolution image
The corresponding intrinsic luminosity surface-density limit,
, is derived for each exposure using the luminosity distance and angular-diameter scale at its redshift. The distribution, shown in the rightmost panel of Figure
, of
peaks near 39.5 erg s
−1
kpc
−2
with an interquartile range of 39.4–39.6 erg s
−1
kpc
−2
An example line-flux map from
HLAN 4025592924
is displayed in Figure
as a contour overlay. The source lies at
hetdex
=2.57 with Ly
-emission at
= 4344 Å. This source’s size is in the top 2% of our LANs, with an effective isophotal radius,
iso
, of 45.9 kpc. An inset in the main panel displays an expanded view of the line profile at seven apertures across the nebula, revealing the nebula’s velocity structure. This object is representative of the larger LANs in the sample, but is special in that it lies in the COSMOS-Web field (C. M. Casey et al.
2023
), has sensitive, high spatial resolution data in the near-infrared from JWST NIRCam (M. Franco et al.
2025
), and a high spectral resolution spectrum from DESI DR1 (DESI Collaboration et al.
2022
; DESI Collaboration
2025
). The DESI DR1 coadded spectrum (top row) clearly shows broad Ly
, but the absence of the C
iv
doublet suggests no strong AGN activity. The three-color JWST composite reveals multiple possible counterparts, but only one is bright in the Hyper Suprime-Cam
-band (HSC-
; S. Miyazaki et al.
2018
). The Ly
halo extends well beyond the stellar continuum.
Figure 3.
Multiwavelength view of HLAN 4025592924 at
hetdex
= 2.57. Situated in the COSMOS Deep Field, this LAN is among the largest in the HETDEX sample, with an isophotal radius of 45.9 kpc—placing it in the top 2% of the distribution. Despite its size, the structure’s Ly
luminosity is moderate at
log_L_lya
= 43.8 erg s
−1
, and it is not identified as an AGN. Top: DESI DR1 (
TARGETID 39627829524040740
; M. E. Levi et al.
2019
; DESI Collaboration
2025
) 1D spectrum on the central bright source (white “+” symbol). The lack of C
iv
suggests the source is not AGN dominated. Main: 30″ × 30″ JWST/NIRCam three-color composite (blue =
(F115W + F150W), green = F277W, red = F444W) from the COSMOS-Web DR1 mosaics (M. Franco et al.
2025
). The Ly
line-flux map (see Section
) from the HETDEX data cube is overplotted as contours (levels 3–15
). The top-left inset shows 1D Ly
profiles extracted along the positions marked by “+” symbols in the image. A 30 kpc scale bar is indicated at lower left. Right column: 30″ × 30″ postage stamp images (top to bottom) from Subaru/HSC-
band, and JWST filters F115W, F150W, F277W, and F444W reveal that multiple low-mass galaxies accompany the LAN.
Download figure:
Standard image
High-resolution image
Further examples of the line-flux maps are shown in the second column panel in Figures
. The leftmost panels in both figures, matched in coordinate space, show supplemental
-band images from the HSC. The dashed red contours denote the LANs’ 2
isophotes.
Figure 4.
Examples of some of the largest extended Ly
-emitters. Photometric imaging from HSC-
is shown on the left, with the 2
boundary of the Ly
emission shown as a dashed red contour. The second column displays the Ly
line-flux map centered on the wavelength listed in white text. The third column presents the radial surface-brightness profile in blue and our best-fit two-component model (PSF core + 2D exponential) in red. The dashed green line is the measured PSF from stars in the same observation as the LAN. The HETDEX/VIRUS spectrum for the central HETDEX detection is given in the rightmost panel. This spectrum is the PSF-weighted spectrum from the HETDEX pipeline. The spectral width of the line-flux map is highlighted in yellow on the spectrum. The line shapes are asymmetrical, and some appear to have multiple associated continuum counterparts in HSC-
images.
Download figure:
Standard image
High-resolution image
Figure 5.
Top four panels: examples of extended Ly
-emitters with a strong central AGN counterpart. The panels have the same format as described in Figure
. Unlike the sources in Figure
, the peak Ly
-emission coincides with peak continuum emission in accompanying photometric data. Bottom two panels: examples of bright AGN that prefer a single-component, point-source model. These are not considered extended by our method.
Download figure:
Standard image
High-resolution image
Figure 6.
Representative examples of smaller LAN in the HETDEX LAE sample. These are plotted from top to bottom in decreasing
iso
. The panels have the same format as described in Figure
Download figure:
Standard image
High-resolution image
4. Model Fitting
For every Ly
line-flux map in the sample, we determine whether a source’s Ly
-emission is extended by comparing two surface-brightness models produced by the software package
pyimfit/imfit
(P. Erwin
2015
). The first model represents point-source emission, as determined by nearby field stars; the second includes this point source, but also adds in a symmetric exponential component to model the extended emission.
4.1. Point-source Model
The impact of the atmosphere and the HET/VIRUS optical path is well described by a 1D Moffat function (A. F. J. Moffat
1969
) with
= 3.5 (G. J. Hill et al.
2021
where
is the angular separation on the sky,
is the central intensity,
, fixed to 3.5, determines the steepness of the wings, and
defines the width of the profile, via
For each HETDEX observation, the image quality or FWHM is determined as part of the reduction pipeline described by K. Gebhardt et al. (
2021
). An average value across all stars in a single HETDEX observation (typically 20–30 stars) is used as a representative value for each observation and is stored in each internal HETDEX data release.
To test for possible differences between the HETDEX pipeline image quality and the image quality applicable to our line-flux maps, we independently assess the latter using spectrally collapsed, interpolated fiber data of stellar spectra. This approach follows the same procedure used to create the line-flux maps described in Section
, except that the spectrum is collapsed over the wavelength range from 4000 to 5000 Å. We compute the mean, median, and standard deviation of the collapsed image and subtract the mean sky to normalize the background. Pixels located beyond a radius of 8″ from the star’s center and exceeding 5 times the sky background level are masked, as they are likely contaminated by background sources.
Using the software
pyimfit/imfit
22
(P. Erwin
2015
), we fit a symmetric 2D Moffat profile with fixed
= 3.5 to each star with a
HETDEX
magnitude between ∼16 and 21. The intensity and FWHM of the Moffat function are treated as free parameters. In total, 42,923 stars are analyzed.
The PSF measurements derived from this approach exhibit a systematic offset of +0
14 compared to the FWHM measurements obtained from the HETDEX reduction pipeline (K. Gebhardt et al.
2021
). This offset is smaller than the standard deviation between the two measurements (0
23), and can likely be explained by minor differences in masking, image interpolation, normalization, and fitting procedures.
To incorporate this offset into our PSF modeling, we adjust the FWHM used in subsequent fits. When modeling both a simple point source and the core component of the two-component model, we fix the FWHM to the value measured in the reduction pipeline, with an additional 0
14 added to account for the systematic offset described above.
Additionally, the modeling of the core is assumed to be circular, and the intensity is treated as a free parameter. The spatial center of the model is allowed to vary within ±5″of the central coordinate reported by the HETDEX detection. This offset is intentionally large to account for any possible astrometric errors due to detection grouping.
4.2. Core Plus Exponential Model
Building on the simple point-source model, we implement a two-component approach that includes both a compact core and an extended exponential halo, a common approach adopted for LAHs (for example, in L. Wisotzki et al.
2016
). The core is modeled as a point source using the same method described above. This is a reasonable assumption, as the typical seeing of HETDEX observations (
to
, or ∼10 to 25 kpc at
∼ 2.5) is comfortably larger than the UV continuum core-scale lengths for LAEs in the Hubble Ultra-Deep Field, constrained by Hubble Space Telescope imaging (F. Leclercq et al.
2017
).
While LANs display a variety of morphologies, the signal-to-noise of individual surface-brightness maps is insufficient for morphological analysis for most objects in our sample. Instead, we only focus on the circularly averaged extent through our two-component surface-brightness model defined as
where
core
) is defined by Equations (
) and (
) (using the FWHM from the HETDEX reduction pipeline increased by +0
14 to correct for the systematic offset described above) and
is the scale length of the exponential. After fitting, we report the scale lengths in physical kpc.
In our fitting, the intensity of the exponential component,
, is allowed to vary, but we require that this extended component contribute at least half of the total intensity at
= 0. This constraint reduces the likelihood of spurious fits in low S/N regions and ensures that a meaningful fraction of the total light comes from the exponential halo. There is evidence, particularly among sources with strong continuum counterparts, that supports a population of objects with a strong central core component plus a moderately bright exponential envelope that contributes less than half of the total intensity. An example of such a source is found in the bottom panel of Figure
. Detailed analysis of the surface-brightness profiles of AGN will follow in future HETDEX papers.
4.3. Fitting Method and Fit Quality
We fit every Ly
line-flux map in the sample with both a core and a core-plus-exponential model using
pyimfit
(P. Erwin
2015
). To estimate the uncertainties in the fitted parameters, we perform 20 bootstrap iterations for each fit.
Two criteria are applied to evaluate the quality of the fits, including the reduced chi-squared statistic and sufficient signal-to-noise:
1.
Lyα Flux Significance
. The flux measured within the isophotal aperture must be detected at greater than 3
, i.e.,
flux_lya
>3 ×
flux_lya_err
2.
Goodness of Fit
. The reduced chi-squared value of the exponential model must satisfy
chi2_exp_reduced
< 3, or 3 ≤
chi2_exp_reduced
< 5 for sources with exceptionally high signal-to-noise, defined as
flux_lya
flux_lya_err
> 300.
Applying these quality cuts to the full sample of 79,830 LAEs yields a final sample of 70,691 sources.
4.4. Determining Size Significance
With the PSF-only model nested within the PSF+exponential model, we opt to use a classical nested-model
-test (W. H. Press et al.
2007
) to assess whether adding the exponential halo yields a statistically significant improvement as opposed to a least-squares likelihood approach. We use this in combination with a difference in Bayesian information criterion (ΔBIC; A. R. Liddle
2007
) as well as other significance measures to determine whether a source is extended.
The F-statistic measures how much
improves per additional parameter, relative to the residual variance of the more complex model.
where
psf
and
are the numbers of free parameters in the PSF and PSF+Exp models, respectively, and
eff
are their degrees of freedom, with
eff
denoting the number of unmasked pixels used in the fit. The full chi-square values are denoted
(PSF) and
(PSF+Exp).
The corresponding
-value,
, quantifies the probability of obtaining an
-statistic as large as or larger than the observed value under the null hypothesis that the PSF-only model is sufficient. It is computed from the cumulative distribution function (CDF) of the
distribution as
where
and
are the numerator and denominator degrees of freedom, respectively. In practice, a small value of
(e.g.,
< 0.05) indicates that the extended model provides a statistically significant improvement in the fit, whereas a large
suggests that the simpler PSF-only model is sufficient.
In addition to the
-statistic, we also compute the difference in Bayesian Information Criterion (ΔBIC; A. R. Liddle
2007
) between the PSF-only and PSF+Exp models, defined as
where Bayesian Information Criterion (BIC; G. Schwarz
1978
) is defined as
where
is the full chi-square value of the fit,
is the number of free parameters in the model, and
eff
denotes the number of unmasked pixels used in the fit. This expression follows from the general definition
under the assumption of Gaussian-distributed residuals with approximately constant variance, for which
reduces to
up to an additive constant. The Gaussian likelihood approximation is well justified for our fits, as the pixel-level uncertainties in the model images are dominated by nearly Gaussian detector and sky noise.
Differences in BIC are widely used as approximate measures of relative model evidence, with values of ∣ΔBIC∣ ≳ 6–10 often interpreted as strong to very strong support for the model with lower BIC (e.g., R. E. Kass & A. E. Raftery
1995
; A. R. Liddle
2007
). In this work, we adopt a slightly more permissive threshold of ΔBIC ≤ −5 to identify extended sources. This choice is empirically motivated by visual inspection of all objects with −10 < ΔBIC < 0, which shows that sources with ΔBIC ≤ −5 consistently exhibit clear spatial extension relative to the PSF, while objects with less negative values are increasingly ambiguous.
We find one additional criterion to be useful for determining whether a source is resolved. We compare the measured isophotal radius,
iso
, to that predicted by the best-fit exponential model based on the surface-brightness limit of each image. The predicted isophote is computed as
where
is the model’s central surface brightness, and
pix
is the background rms per pixel, derived from the measured sky noise (
SB_1sigma_obs
) and pixel scale (0
25).
We then quantify the fractional mismatch between the measured and predicted isophotal radii as
This quantity, stored as the catalog column
iso_rel_err
, measures the relative deviation between the observed morphology and that expected from the best-fit exponential profile. Sources with iso_rel_err > 1 are flagged as inconsistent (
flag_iso_mismatch
) and excluded from the extended-source sample, as such discrepancies typically indicate unstable or nonphysical fits, generally due to low signal-to-noise or from contamination of foreground objects. This test provides an additional safeguard against false extensions, ensuring that only objects whose observed isophotal radii are consistent with their modeled surface-brightness profiles are classified as genuinely extended. Values of
iso_rel_err
close to zero indicate good agreement between the observed and model-predicted isophotal radii, while large values signify inconsistency between the fitted profile and the measured morphology.
A source is deemed to be the best fit by a two-component model when the following quantitative criteria are satisfied:
1.
F-test criterion:
log10_pF
< −12.
2.
ΔBIC criterion:
dBIC
< −5
3.
Isophotal consistency: The observed and model-predicted isophotal radii must agree within a factor of 2 (
iso_rel_err
< 1).
4.
significance: To ensure the halo is genuinely resolved, we additionally require that the exponential scale length is significantly detected,
r_s
> 3 ×
r_s_err
Out of the sample of 70,691 sources that satisfy the quality criteria described in Section
4.3
, 33,612 are best fit by a two-component, extended model, indicating that nearly half of the sample exhibits statistically significant extended Ly
emission in which the exponential halo contributes at least half of the total model intensity.
4.5. Isophotal Effective Radius
We measure the radial surface-brightness profile of each source using a sequence of circular annuli centered on the best-fit emission peak of the two-component model. These profiles provide a direct, model-independent view of the Ly
surface-brightness decline. Examples are shown in the third column of Figures
, where blue points indicate the radial profile and the horizontal dashed line marks the standard deviation of the sky background,
sky
The isophotal radius,
iso
(catalog column
r_iso
), is defined as the radius at which the azimuthally averaged Ly
surface brightness falls below the local background standard deviation,
sky
(catalog column
SB_1sigma_obs)
. The corresponding isophotal flux (
flux_lya
) and circularized area (
area_r_iso_circ
, in arcsec
) are measured within the aperture defined by
iso
and listed in Table
Table 2.
Column Information for the HETDEX Ly
Nebulae Catalog
Column Name
Unit
Description
name
HLAN+
detectid
Unique observation-specific HETDEX Lyα Nebula (HLAN) identifier.
ra
deg
Best-fit model coordinate R.A.
dec
deg
Best-fit model decl.
source_type
Source class (
lae
or
agn
if in HETDEX AGN catalogs; C. Liu et al.
2022
; A. R. Liddle
2025
).
z_hetdex
Spectroscopic redshift (see E. Mentuch Cooper et al.
2023
).
z_hetdex_src
Redshift source as described in E. Mentuch Cooper et al. (
2023
).
detectid
HETDEX unique internal detection ID.
shotid
HETDEX observation ID.
field
Survey field label.
SB_1sigma_obs
10
−18
erg s
−1
cm
−2
arcsec
−2
Observed-frame 1
Ly
surface-brightness sensitivity per IFU exposure.
r_iso
kpc
Isophotal radius where Ly
intensity in a circular aperture falls below the sky background level (
sky).
r_s
kpc
Best-fit scale length of the exponential component.
r_s_err
kpc
error on
r_s
area_iso_2sigma
arcsec
Isophotal sky area based on 2
sky + continuum-subtracted background.
area_r_iso_circ
arcsec
Isophotal sky area based on circular aperture defined by
r_iso
logL_lya
(erg s
−1
Ly
luminosity converted from
flux_lya
logL_lya_err
(erg s
−1
error on
logL_lya
flux_lya
erg s
−1
cm
−2
Ly
flux in the circular aperture defined by
r_iso
flux_lya_err
erg s
−1
cm
−2
error on
flux_lya
gmag
mag
-band AB magnitude as measured in the HETDEX PSF spectrum.
HSC-r_mag
mag
HSC-
AB magnitude in a 2
aperture.
HSC-r_mag_err
mag
error on HSC magnitude.
combined_eqw_rest_lya
Rest-frame Ly
equivalent width from D. Davis et al. (
2023
).
flag_resolved
1 = resolved, 0 = point-source
Final decision whether the source is resolved based on the criteria described in Section
4.4
chi2_ext_reduced
Reduced
of the extended, two-component (PSF + exponential) fit.
chi2_psf_reduced
Reduced
of the PSF-only fit.
log10_pF
statistic from the F-test. A source is accepted as resolved if
dBIC
Difference in the Bayesian Information Criterion. Resolved if dBIC < −5.
iso_rel_err
Fractional deviation between the measured and model-predicted isophotal radii is described in Section
4.4
dups_detectid
List of duplicate HLAN names for repeat observations.
Note.
The HLAN Catalog can be accessed online at
and in the electronic version of this paper. The column names in the online journal version differ slightly to conform to journal standards.
Only a portion of this table is shown here to demonstrate its form and content. A
machine-readable
version of the full table is available.
Download table as:
Machine-readable (MRT)
Typeset image
This definition differs from the isophotal contour-based approaches used in previous work (e.g., Y. Matsuda et al.
2011
; S. Cantalupo et al.
2014
; J. F. Hennawi et al.
2015
; Z. Cai et al.
2017
), which are circularized from the contour area and converted to an effective radius. Our method instead derives
iso
directly from the azimuthally averaged profile, ensuring a uniform and reproducible measurement across the full sample. The difference arises from methodology rather than PSF or sensitivity variations: our profiles extend to the 1
surface-brightness limit and neglect ellipticity, typically yielding slightly larger radii than contour-based measurements. This approach is particularly robust for faint or low S/N halos where 2D contouring becomes unreliable and averaging the signal in annuli provides higher S/N. Although our circular-profile method yields systematically larger effective radii, it provides a consistent comparison to surface-brightness analyses of lower-luminosity halos from MUSE surveys (e.g., L. Wisotzki et al.
2016
; F. Leclercq et al.
2017
) and stacked HETDEX samples (M. Lujan Niemeyer et al.
2022a
2022b
; B. McKay et al.
2026
).
4.6. Isophotal Area
To compare our approach with standard isocontour methods, we also measure the isophotal area,
iso
, directly from the Ly
line-flux maps. The area enclosed by the 2
surface-brightness contour—corresponding to twice the observed 1
limit (
SB_1sigma_obs
)—defines
area_iso_2sigma
in arcsec
. This quantity allows us to compare contour-based areas with those inferred from the circularized isophotal radius. Both metrics are included in the catalog as described in Table
. The comparison is discussed in detail in Section
6.1
5. Catalog Description
The HETDEX Ly
Nebulae (HLAN) Catalog, containing the best-fit surface-brightness model parameters and LAN source information, is available online at
and in the electronic version of this paper. A description of the table columns is given in Table
. The name given to each object follows the nomenclature “HLAN” supplemented by a representative
detectid
value, where
detectid
is the integer ID used by the internal HETDEX data model to represent a HETDEX detection.
In the HETDEX catalog, every point-source emission-line detection is assigned a unique
detectid
. For spatially resolved objects, such as LANs, multiple
detectids
can represent a single source—either as a result of multiple point-source detections at the same wavelength but different spatial positions, or multiple spectral emission-line detections at different wavelengths but coincident spatial locations. During catalog generation, described in detail in E. Mentuch Cooper et al. (
2023
), a series of friends-of-friends detection groupings in both 2D and 3D space is applied together with additional logic to reduce multiple detections to a single source. This consolidation is performed on a per-observation basis. Ultimately, a single representative
detectid
is retained for each LAN. This “selected”
detectid
often corresponds to the brightest continuum detection.
If multiple HETDEX observations (labeled
shotid
in the catalog) cover the same source, this may result in multiple HETDEX detections for the same source and ultimately multiple LAN entries in the HLAN Catalog. We identify HLAN objects with duplicate observations and give their
detectid
matches in column
dups_detectid
As stated above, each surface-brightness model is allowed to raster its location of peak intensity (
= 0) ±5″ around the source’s reported coordinates in HETDEX. The computed centroids of are provided in the catalog. In many cases, these positions agree well with the HETDEX
detectid
coordinates, but subtle differences in the source-grouping code or redshift assignment do sometimes produce offsets. We store the best-fit coordinates in the catalog in the columns
ra
and
dec
. The median separation between this best-fit peak in emission and the representative detection of the source is 0
51 with an interquartile range of 0
35–0
74.
From the initial sample of 79,830 LAE candidates, 33,612 sources are best fit by a two-component surface-brightness model consisting of a point-source core plus an exponential halo whose flux contributes more than 50% of the total Ly
intensity. To identify these objects in the catalog, select column
flag_resolved
, where 1 = PSF+EXP model preferred, and 0 = PSF model preferred.
Ultimately, HETDEX is limited in both image quality and sensitivity, and thus, if a source is not identified as extended in this work, it does not mean that it lacks a significant Ly
halo component. Indeed, studies (L. Wisotzki et al.
2016
; F. Leclercq et al.
2017
) at higher spatial resolution and more sensitive surface-brightness limits find that all LAEs have an extended component.
5.1. Duplicate LANs and Internal Consistency
We searched for repeat detections of the same source by crossmatching sky positions. All pairs of detections within 5″ were identified and grouped using a connectivity graph, resulting in 1837 distinct duplicate sources in the catalog. The unique identifiers (
detectid
) of neighboring detections are stored in the catalog column
dups_detectid
, which lists the set of duplicate sources (comma separated) for each entry.
The presence of repeat observations enables an internal check on the robustness of our extended-source classification. Among all duplicate groupings, we find that approximately 60% of sources receive consistent resolved/unresolved classifications. In regions with exceptionally deep data (reaching ∼2 ×  10
−18
erg s
−1
cm
), this agreement increases to about 80%. This trend reflects the strong dependence of extended-source identification on surface-brightness sensitivity. When stacking HETDEX sources, extended Ly
emission is detected for stacked LAE surface-brightness profiles (M. Lujan Niemeyer et al.
2022a
2022b
; B. McKay et al.
2026
), and it is likely that, given better sensitivity, every HETDEX source is likely extended in Ly
emission as is found in L. Wisotzki et al. (
2016
).
To evaluate the internal repeatability of key observables, we compared measurements of the isophotal radius (
iso
), exponential scale radius (
), isophotal area (
iso,2
), Ly
flux (
Ly
), and Ly
luminosity (
Ly
) among all members of each duplicate group. The median fractional scatter among repeats was ∼0.08 for
iso
, 0.12 for
, 0.13 for
iso,2
, and 0.12 for
Ly
, corresponding to a typical variation of 10%–16%. The derived luminosities show similarly small differences, confirming that photometric and flux calibration consistency across independent HETDEX observations is excellent.
5.2. Active Galactic Nucleus Hosts
The internal HETDEX AGN Catalog for HDR5, (public versions are available in C. Liu et al.
2022
; A. R. Liddle
2025
), contains 11,995 AGN in the redshift interval where the Ly
line falls within the VIRUS spectral window (1.9 <
< 3.5) but only 7890 persist after selection cuts and the analysis is cleaned for poor model fits (see Sections
and
4.3
). Thus the parent sample has an incidence of 11.2% AGN. AGN identifications are given in column
source_type
of the HLAN catalog, and rely on at least one of the following criteria:
1.
Broad rest-frame ultraviolet emission with FWHM > 1000 km s
−1
2.
A prior AGN identification in SDSS spectroscopy,
3.
The simultaneous detection of Ly
and a high-ionization companion line (typically C
iv
) at a consistent redshift.
In total, 4106 out of 7890 AGN (52.5% of the AGN sample) are better fit with a two-component (core + exponential) surface-brightness model than by a single point-source profile. The incidence of AGN in the parent sample is similar to the fraction of AGN in the LAN sample. We find that 12.2% of LANs are identified as AGN.
The 50% halo-flux threshold adopted in our classification is intentionally conservative. Visual inspection of several AGN whose best-fit solution is the single-component model still reveals low-surface-brightness Ly
emission beyond the seeing disk. For these brightest AGN, the strong central core dominates the emission, and the sources are categorized best by the one-component model. Two examples can be seen at the bottom of Figure
. This leads to a decrease in the extended fraction among the most luminous AGN, as is discussed later in Section
6.3
. A more detailed decomposition of nuclear and extended contributions for the entire AGN sample is deferred to future work.
6. Results
The resolved LAN sample contains 33,612 sources. The redshift distributions of the HLAN LANs and the S/N > 6 HETDEX LAEs are shown in the left panel of Figure
. The distributions appear quite similar except for a noticeable difference at higher redshifts (
> 2.8). This decrease is likely due to the effects of cosmological surface-brightness dimming rather than any evolutionary effect.
Figure 7.
Left: redshift distribution; the green histogram shows the full S/N > 6 LAE sample, while the blue histogram shows the resolved LAN subset. Middle/right: distributions for the resolved subset of (from left to right)
(erg s
−1
), isophotal radius
iso
(kpc), and isophotal area
iso
measured at the 2
surface-brightness contour (arcsec
). All panels share a common
-axis (counts); tick labels are shown on the far left and right for readability. In the
iso
panel, vertical reference lines mark
πθ
for seeing FWHM
= 1
8 (dashed) and 2
5 (dotted).
Download figure:
Standard image
High-resolution image
The LAN sample spans a range of sizes and luminosities, with their distributions plotted in Figure
, with a median isophotal radius of
iso
= 21.7 kpc (16th–84th percentile range: 17.3–29.3 kpc), and 152 systems exceeding 50 kpc. For comparison, the exponential scale lengths have a median value of
= 11.8 kpc (8.55–16.9 kpc). A comparison of these two radii is shown in the left panel of Figure
. This demonstrates that the measured isophotal radius is governed primarily by the nebula’s total Ly
flux rather than an intrinsic size–luminosity relation. Because
iso
is defined at a fixed 1
sky
surface-brightness level, galaxies with low integrated flux fall below the detection threshold at smaller radii, whereas more luminous systems remain detectable to larger distances. The fitted exponential scale length
is less sensitive to this surface-brightness limit and therefore provides a more robust tracer of the intrinsic physical extent. Any interpretation of
iso
must therefore account for these surface-brightness selection effects, which are discussed further in Section
6.4
Figure 8.
Structural comparisons among HETDEX Lyα Nebulae (LANs). (a) Exponential scale length,
, versus isophotal radius,
iso
, measured at the 1
surface-brightness limit. Blue and green contours denote LAEs and AGN, respectively; the dashed line marks
iso
. Most sources lie below this relation, confirming that detectable Ly
emission generally extends beyond the fitted exponential scale length. (b) Comparison between the measured isophotal area,
iso
(2
Ly
contour), and the circularized area derived from the 1
isophotal radius,
. The solid line shows the one-to-one relation, and the dashed line marks
circ
= 1.5
iso
, close to the sample median. Both panels demonstrate that the circularized radius provides a reliable approximation to the isophotal extent of the Ly
emission, while the exponential component typically underestimates the full halo size. The median ratio
circ
iso
= 1.42 corresponds to radii that are ≃20% larger, confirming that the circularized
iso
closely reproduces the measured isophotal extent.
Download figure:
Standard image
High-resolution image
Their Ly
luminosities span
–44.82 erg s
−1
, with a median of 43.15 and a 16th–84th percentile range of 42.92–43.45. The median Ly
flux is 19.9 × 10
−17
erg s
−1
cm
−2
(12.6–38.8). The isophotal area measured at the 2
surface-brightness contour has a median of 14.8 arcsec
(9.94–24.4 arcsec
), while the circularized area from
iso
is systematically larger, with a median of 21.8 arcsec
(16.6–32.8 arcsec
).
As a robustness check, we repeated the analysis for a representative test sample using fixed-width pseudo-NB images. We find that a wide fixed window (
= 8) yields results consistent with our fiducial variable-width approach, while an overly narrow choice (
= 2) systematically suppresses extended Ly
emission and increases classification changes for fainter halos that lie on the decision boundary between core and halo dominated.
6.1. Isophotal Area Comparison
To assess the correspondence between the effective isophotal radius (
iso
, measured at 1
SB_1sigma_obs
in the catalog) and the total area enclosed by the 2
surface-brightness contour (column
area_iso_2sigma
), we compare the circularized area inferred from
iso
, available in column
area_r_iso_circ
) with the directly measured isophotal area,
iso,2
, in the right panel of Figure
. The majority of sources cluster around the one-to-one relation, but with a systematic offset, indicating that the circularized radius provides a robust proxy for the total isophotal extent, albeit with moderate scatter toward smaller
iso
values at higher surface-brightness thresholds. The distribution is centered above unity with median(
) = 1.42 and an interquartile range of [1.33,  1.75], implying that the circularized area inferred from
iso
typically exceeds the 2
isophotal area by approximately 50%.
A small minority of systems show
< 1 (
), consistent with cases where the 2
isophote extends asymmetrically beyond the circularized aperture—such as filamentary morphologies, centroid offsets, or blended structures. Conversely, large
outliers likely occur when circularization overestimates the extent of elongated or patchy halos, when
iso
measured at the 1
level captures low S/N wings absent at 2
, or from minor measurement systematics (e.g., segmentation or masking effects, deblending errors, or redshift-dependent kpc/″ conversions for a small subset of sources).
Overall, this comparison confirms that the circularized radius measurement provides a reliable and easily reproducible proxy for total Ly
extent, with predictable biases that can be statistically characterized for large samples.
6.2. Optical Counterparts
Nearly the entire HETDEX spring field has been imaged in the
-band to
∼ 26.1 with Hyper Suprime-Cam (HSC; D. Davis et al.
2023
); similarly, the other HETDEX fields considered in this paper have
grizy
coverage down to
∼ 26 from Subaru Strategic Program (HSC-SSP; H. Aihara et al.
2018
). These data enable a direct comparison between the continuum counterparts of our Ly
detections and their emission-line properties. Aperture magnitudes on their
-band continuum images were measured in a series of circular apertures centered on the LAN centroid position and grown until the enclosed optical flux converged. These are found in the catalog in column
HSC-r_mag
if available. Where no counterpart is present at the 3
level (
≳ 29), we fix an upper limit at
= 30 in the catalog, but do not plot these values.
Figure
summarizes the relation between continuum brightness, Ly
flux, and the extent of the Ly
emission for HETDEX LANs. Panels (a)–(c) show that LAEs (blue) and AGN (green) span similar Ly
fluxes and halo sizes despite a systematic offset in continuum brightness, with LAEs typically ∼1–1.5 mag fainter in HSC-r. In (a), we see that Ly
flux generally scales with continuum magnitude. In (b) and (c), the exponential scale length
and isophotal radius
iso
display little dependence on continuum magnitude, suggesting that the spatial extent of the Ly
emission is only weakly tied to the stellar or AGN continuum luminosity. However, the largest haloes (
≳ 20 kpc;
iso
≳ 40 kpc) are preferentially found among optically bright AGN, consistent with enhanced Ly
scattering or illumination in AGN-driven ionization cones.
Figure 9.
Optical magnitudes of HETDEX Lyα Nebulae (LANs) with identified HSC counterparts. From left to right: (a) Ly
flux versus HSC-r aperture magnitude; (b) exponential scale length,
, as a function of HSC-r; (c) isophotal radius,
iso
, as a function of HSC-r; and (d) isophotal radius versus rest-frame Ly
equivalent width (EW). Blue and green contours correspond to LAEs and AGN, respectively. LAEs are typically ∼1–1.5 mag fainter in the optical continuum than AGN yet span comparable Ly
fluxes. Both
and
iso
show weak dependence on continuum brightness, though the most extended haloes (
≳ 20 kpc;
iso
≳ 40 kpc) are preferentially associated with the optically brightest AGN.
Download figure:
Standard image
High-resolution image
Panel (d) relates the physical halo size to the rest-frame equivalent width (EW). Generally, the larger
iso
LANs have lower EWs as they tend to be associated with brighter continuum counterparts (as seen in Panel (c)). This trend supports a picture in which large-scale, low-surface-brightness Ly
emission contributes substantially to the total line flux, boosting EW measurements for galaxies embedded in more diffuse, resonantly scattered haloes which are not associated with bright-continuum counterparts.
Together, these trends underscore two key points and echo the recent results of M. Li et al. (
2024
): (i) the majority of LAEs in HETDEX are faint or even undetected in deep
-band imaging, reinforcing the power of blind IFU surveys to uncover low-mass, continuum-faint galaxies; and (ii) AGNs contribute disproportionately to the bright-continuum, low-EW tail of the LAE distribution but represent only a minor fraction (∼12%) of the overall population of Ly
emitters.
6.3. How Many LAEs are Extended?
Figure
10
quantifies, for the first time in an wide-area, untargeted IFU survey, how frequently luminous LAEs at
≃ 2–3 at 1
Ly
sensitivities of ∼2–5 × 10
−18
erg s
−1
cm
display spatially extended Ly
emission. We first consider the fraction of sources that prefer the two-component (PSF+exponential) model fit. For a series of Ly
flux bins, we calculate the fraction of sources that satisfy the extended-source criteria described in Section
4.4
, divided by the total number of sources in each flux bin. The ensemble sample is shown in solid gray, while those classified as LAEs and AGNs are shown by the blue and green curves, respectively. At both low (
Ly
< 10 ×  10
−17
erg s
−1
cm
−2
) and high (
Ly
> 100 ×  10
−17
erg s
−1
cm
−2
) line-flux values, both AGN and LAEs exhibit higher incidences of point-source-preferred models.
Figure 10.
Incidence of spatially extended Ly
nebulae in the S/N > 6 parent LAE sample. Left: fraction of sources that prefer the two-component (PSF+exponential envelope) model over a single PSF-only fit. At low (
Ly
< 10 ×  10
−17
erg s
−1
cm
−2
) and high (
Ly
> 100 ×  10
−17
erg s
−1
cm
−2
) line-flux values, both AGN and LAE have higher incidences of point-source preferred models. Middle: fraction of sources whose isophotal radius exceeds either 20 kpc (solid) or 30 kpc (dashed) as a function of the Ly
flux measured at
iso
. Right: same as the middle panel, but binned by the Ly
luminosity enclosed within
iso
. Blue and green curves denote galaxies classified as LAEs and optically identified AGN (A. R. Liddle
2025
), respectively. Gray lines show the PSF-only null hypothesis computed for each object from its own seeing, surface-brightness limit, and redshift. The observed incidence of extended emission rises steeply with increasing Ly
surface brightness or luminosity, well above the expectation for unresolved sources.
Download figure:
Standard image
High-resolution image
In the lower flux regime, the data lack sufficient S/N to provide resolved fits to the data, so we cannot conclude that these LAEs truly lack extended Ly
halos. In B. McKay et al. (
2026
), we show that stacks of HETDEX LAEs with good image quality reveal that extended emission is still present on average.
At higher fluxes, visual inspection shows that sources best described by the PSF-only model—without a significant exponential component—tend to have bright optical counterparts. In the AGN sample, there are spectral signatures of AGN hosts (broad emission, AGN emission-line pairs, bright-continuum optical counterparts). Two examples of bright AGN with strong Ly
emission but little extended structure are shown in the bottom two panels of Figure
. In the LAE sample, the sources that prefer a one-component model appear similar to the AGN cases: they have luminous optical counterparts but lack clear AGN features in the HETDEX spectra. It is plausible that many of these are, in fact, narrow-line AGN with other emission lines, such as C
iv
, lying outside the HETDEX spectral window. We note that, upon visual inspection, even among the brightest systems, some extended emission remains evident, but it contributes less than 50% of the total Ly
flux. Consequently, the two-component model offers no statistical preference over the simpler PSF-only fit. This behavior reflects the fact that our classification is based on statistical model comparison rather than on residual flux after PSF subtraction, and is therefore sensitive to the dominance of the central source at high luminosities, particularly because the extended component must contribute a substantial fraction of the total Ly
flux to be statistically favored.
In the middle panel, we consider the fraction of objects whose isophotal radius exceeds either 20 or 30 kpc—thresholds that bracket the classical definitions of “medium” and “large” LANs—as a function of both the Ly
flux, in the middle panel, and intrinsic luminosity in the right panel. These are aperture values determined by the circular aperture defined by
iso
. In these panels, we separate those defined as AGN (green) from those defined as LAE (blue). The fractions are determined for the full parent sample of 70,691 LAEs.
The extent of
iso
is dependent on both line flux and the surface brightness of the data, and will increase even in the case of a PSF-only surface-brightness model. Even if the sample consisted entirely of unresolved sources, a trend would be seen with flux. We consider this PSF-only null hypothesis in the gray curves in the middle and right panels. We estimate the apparent isophotal radius expected for an unresolved source by convolving a point source with the seeing profile described by a circular Moffat function and evaluating where the surface brightness falls below the 1
detection limit. The calculation is performed for each object using its own FWHM, surface-brightness limit, and redshift to account for variations in angular scale and depth. We then determine the fraction of such point-source models that would exceed 20 kpc (solid gray curve) or 30 kpc (dashed gray curve) at the survey isophote.
At low fluxes, the observed fractions closely follow the null prediction, indicating that nearly all sources are unresolved. As flux increases, however, both LAE and AGN populations rise well above the gray curves, implying that the measured sizes cannot be explained by PSF broadening alone. The excess fraction above the null model, therefore, represents genuine spatially extended Ly
emission. LAEs, particularly those with
iso
> 30 kpc (dashed blue curve), show a systematically higher incidence of extended radii than AGN (dashed green curve) at fixed flux, suggesting that their halos are more prevalent or more diffuse for a given brightness.
The right panel presents the same metric as a function of intrinsic Ly
luminosity. The trends mirror those in the middle panel but remove the influence of distance on the observed flux. The offset between the colored and gray curves again indicates that extended emission is common beyond what would be expected for unresolved sources and that LAEs are more likely to be extended than an AGN-dominated sample. The 50% incidence threshold for
iso
> 30 kpc occurs near
Ly
∼ 3 × 10
43
erg s
−1
for LAEs and slightly higher for AGN at
Ly
∼ 3 × 10
43
erg s
−1
, consistent with a luminosity-dependent transition in which nearly all systems above 10
44
erg s
−1
host large-scale Ly
halos.
Together, the middle and right panels demonstrate that the frequency and extent of extended Ly
emission increase steeply with both flux and luminosity, and that LAEs remain more spatially extended than AGN over the full dynamic range. These results show that extended LANs are common but not ubiquitous. Their prevalence depends on both intrinsic Ly
output and the nature of the power source, with the systematic offset between AGN and star-forming LAEs suggesting fundamental differences in the mechanisms that excite Ly
emission.
6.4. The Ly
Halo Size–Luminosity Relation
To explore the relationship between Ly
halo size and luminosity, we compare HETDEX LAEs to a compilation of literature sources spanning diverse environments and measurement techniques. Figure
11
shows the distribution of isophotal radii (
iso
, left panel) and exponential scale radii (
, right panel) as functions of
Ly
. In the left panel, we include isophotal measurements from LABs compiled from E. Borisova et al. (
2016
), Y. Matsuda et al. (
2011
), Z. Cai et al. (
2017
), S. Cantalupo et al. (
2014
), J. F. Hennawi et al. (
2015
), and M. Li et al. (
2024
). We apply a correction factor of 0.5 to approximate isophotal radii from full angular extent measurements and note that these will naturally be dependent on the observation’s surface-brightness sensitivity. The right panel shows exponential scale lengths from the LAB sample from E. Borisova et al. (
2016
), and from the LAH samples from L. Wisotzki et al. (
2016
), F. Leclercq et al. (
2017
), and R. Xue et al. (
2017
), alongside HETDEX LAE stacks (B. McKay et al.
2026
).
Figure 11.
Radial extent versus Ly
luminosity measurements based on different size estimators. Left: isophotal radius (
iso
) versus Ly
luminosity (
Ly
) for HETDEX LAEs and AGN (gray contours), compared with previous studies (Y. Matsuda et al.
2011
; S. Cantalupo et al.
2014
; J. F. Hennawi et al.
2015
; E. Borisova et al.
2016
; Z. Cai et al.
2017
; M. Li et al.
2024
) of extended Ly
emission (colored symbols). Median HETDEX sizes in bins of luminosity are shown separately for LAEs (blue), AGN (green), and the combined sample (gray). Right: exponential scale length (
) versus
Ly
for the same sample from E. Borisova et al. (
2016
), and additional measurements of Ly
halos from F. Leclercq et al. (
2017
), L. Wisotzki et al. (
2016
), and HETDEX LAE stacks from B. McKay et al. (
2026
). Gray shaded regions in both panels show kernel-density estimates (KDEs) enclosing 25%–95% of the HETDEX LAE+AGN population.
Download figure:
Standard image
High-resolution image
To visualize the distribution of the 33,612 HETDEX sources in the
Ly
–size plane, we computed 2D kernel-density estimates (KDEs) using a Gaussian kernel, evaluated on a uniform logarithmic grid in both luminosity and size. The KDEs are normalized such that their total integral equals unity, and the contours in Figure
11
correspond to the highest-density probability (HDP) regions enclosing 25%, 50%, 75%, and 95% of the total LAN population. In both panels, HETDEX sources occupy a relatively narrow region in luminosity–size space, with a trend of increasing
iso
with luminosity in the left panel, in line with other sources in the literature. No significant trend in the distribution is seen in the right panel with exponential scale length,
The ensemble averages shown for LAEs (blue), AGN (green), and the combined sample (gray) confirm that
iso
increases systematically with
Ly
, consistent with literature data points. In contrast, the exponential scale lengths show little luminosity dependence, implying that the intrinsic halo profiles remain broadly similar across luminosities. This behavior reflects the known sensitivity of isophotal measurements to surface-brightness limits: as emphasized by C. C. Steidel et al. (
2011
), deeper data reveal more extended emission even at fixed luminosity. The scatter in
iso
in HETDEX sources likely arises from a combination of intrinsic variation in the individual source luminosity, redshift, and Ly
surface-brightness distribution, as well as measurement differences in sensitivity and image quality. For example, much of the observed scatter can likely be mitigated by adopting a universal surface-brightness threshold that corrects for cosmological dimming, following the approach of F. Arrigoni Battaia et al. (
2023
).
Across the combined HETDEX LAE+AGN sample, the median exponential scale length is
= 11.6 ± 1.9 kpc. Separating by source type, LAEs exhibit larger median scale lengths (12.9 ± 1.5 kpc) than AGN (10.3 ± 1.4 kpc), revealing differences in emission related to AGN and star-forming regions. Future work will relate HETDEX LAN profiles to galaxy properties in greater depth.
6.5. Resolved versus PSF Flux Comparisons
IFU observations of LAEs with VLT/MUSE show that much of the Ly
energy budget lives in this halo component: individual LAEs contain ∼40%–90% of their line flux at radii that exceed the seeing disk (L. Wisotzki et al.
2016
), with a median halo contribution of ≃65% in the MUSE Hubble Ultra-Deep-Field sample (F. Leclercq et al.
2017
). PSF-weighted or circular apertures recover only a fraction of the total Ly
emission. In the narrowband LAE sample of Y. Huang et al. (
2021
), for example, point-source or fixed angular apertures miss ∼30% of the Ly
flux in a
∼ 3.1 protocluster.
Because the HETDEX pipeline extracts a PSF-weighted spectrum by construction, its reported fluxes can be affected by this bias. Figure
12
quantifies the effect on a source-by-source basis. Nearly every LAE lies above the 1:1 line, confirming that PSF extractions underestimate the integrated Ly
flux. The offset correlates with
iso
(shown in color), with the largest objects resulting in larger flux measurement offsets. Across the full dynamic range, the median
Ly
,iso
Ly
,psf
is ∼1.3, implying a systematic flux deficit of roughly 30%.
Figure 12.
Comparison between the Ly
fluxes measured in isophotal apertures,
iso
, and fluxes measured from PSF measurements in the HETDEX reduction pipeline (
psf
). Top: individual LAEs are color coded by their physical isophotal radius
iso
; the dashed light-gray line marks the 1:1 relation. Most objects lie above this line, showing that a PSF-based extraction systematically misses flux from extended nebular emission. Bottom: median flux ratio
iso
psf
in 12 adaptive bins. The median ratio is ≃1.3 with little dependence on flux.
Download figure:
Standard image
High-resolution image
This systematic offset will propagate directly into luminosity functions, equivalent-width distributions, star formation rate estimates, and any quantity that depends on absolute Ly
luminosity.
6.6. HETDEX LANs Observed in Other Surveys
A comparison of our catalog to the recently released LAN sample from the MAMMOTH-Subaru survey (M. Li et al.
2024
) reveals four LANs common to both catalogs. This survey reaches similar Ly
surface-brightness limits to HETDEX, with a typical 2
Ly
surface brightness of 5–10 × 10
−18
erg s
−1
cm
−2
arcsec
−2
. As with HETDEX, the surface brightness varies from field to field. Table
summarizes the comparison, including positions, isophotal areas, and Ly
luminosities from both studies. The matched LANs between our sample and MAMMOTH agree well in position, with coordinate offsets always less than 1
3 (median offset: 0
6). The Ly
luminosities (measured within the encompassing isophotal areas) are also consistent, differing by less than 0.16 dex (median
dex), well within typical systematic uncertainties from differing measurement apertures and surface-brightness limits. In both cases, the isophotal area comes from the area enclosed in the 2
isophotal boundary (
area_iso_2sigma
in the HETDEX catalog). Overall, the HETDEX and MAMMOTH measurements agree quite well—typical area differences are around 15%, consistent with expected variations from slightly different surface-brightness thresholds and segmentation methods.
Table 3.
Comparison of LABs Matched between Our Sample and MAMMOTH
MAMMOTH
This Work (HETDEX)
Name
R.A.
Decl.
Area
log
HETDEX Name
R.A.
Decl.
Area
log
(deg)
(deg)
(arcsec
(erg s
−1
(deg)
(deg)
(arcsec
(erg s
−1
MLAN43
32.1770
0.6582
41.6
43.49
HLAN4016274387
32.1769
0.6582
39.38
43.65
MLAN78
32.7873
0.7366
26.1
43.34
HLAN4015512497
32.7870
0.7367
21.50
43.19
MLAN60
32.9132
1.0364
32.2
43.36
HLAN3010736450
32.9131
1.0361
36.06
43.43
MLAN85
35.5727
−2.1925
24.3
43.05
HLAN3009861289
35.5726
−2.1925
13.94
43.08
Note.
Coordinates, solid angle, and Ly
luminosity from both studies are listed. Area is measured from isophotal contours drawn at the varying 2
surface brightness (column
area_iso_2sigma
) limit of the HETDEX observation.
Download table as:
ASCII
Typeset image
6.7. Radio Counterparts
Radio observations of the HETDEX Spring field from the Low Frequency Array (LOFAR; M. P. van Haarlem et al.
2013
) are fully included in the second data release from the LOFAR Two-Metre Sky Survey (LOTSS; T. W. Shimwell et al.
2022
). HETDEX counterparts of LOTSS sources were previously identified in M. H. Debski et al. (
2025
) from an earlier HETDEX data release, but here we are able to consider radio counterpart fractions obtained from the LOFAR DR2 catalog (M. J. Hardcastle et al.
2023
) for the full parent sample of HETDEX LAEs and compare this to the LAN radio counterpart fraction.
In the full parent sample of LAEs with S/N > 6, 51,570 LAEs are in the HETDEX Spring field, and 1523 (2.9%) have radio counterparts. Of these LAEs, roughly half the sample (979/1523; 64%) prefer an extended emission model. Inspection of those LAEs with radio emission that are best represented by a point-source model are generally AGN (296/596) or LAEs with lower S/N. Half of the LAEs in this sample are at S/N < 7 and lack sufficient signal to detect extended emission.
In total, there are 21,566 LANs in the HLAN Catalog in the HETDEX Spring field. Of these objects, roughly 4.5% (979/21,566) have radio counterparts within 5″of the peak LAN emission. Of these LAN-LOFAR sources, 371 are labeled AGN. The sample is biased toward higher isophotal radii and Ly
luminosities. For example, the LAN-LOFAR sources have a median
iso
= 28.9 kpc compared to 21.7 kpc and a median
of 43.4 erg s
−1
compared to 43.1 erg s
−1
for the full LAN sample. The fraction of LANs with radio counterparts increases for larger LANs. For instance, LANs with
iso
> 30 kpc, 14% of the LANs have radio counterparts, and at
iso
> 50 kpc, 35% of the LANs have radio counterparts. This can be seen in the rightmost panel of Figure
13
where the fraction of LANs with radio counterparts increases with
iso
Figure 13.
Left: Ly
line flux versus integrated LOFAR 144 MHz flux density. Blue and green symbols indicate LAE and AGN, respectively; gray open circles highlight sources with spatially resolved LOFAR emission (
Resolved
= True in T. W. Shimwell et al.
2022
). Middle: comparison between the Ly
isophotal radius measured by HETDEX (
iso
) and the LOFAR largest angular scale converted to proper kpc (
LAS
). Right: fraction of HETDEX sources with LOFAR counterparts as a function of
iso
, shown separately for LAE (blue), AGN (green), and the combined sample (gray). Larger Ly
halos show a higher probability of association with LOFAR emission, particularly among AGN.
Download figure:
Standard image
High-resolution image
Some extended radio counterparts have HETDEX Ly
counterparts. In the full S/N > 6 sample, 66 extended radio sources are found. Nearly two-thirds of these extended radio sources (41/66) are also in the HETDEX LAN sample. For HETDEX LANs with resolved LOFAR detections, we observe a diverse range of radio morphologies. Broadly, they fall into two categories, which we highlight in an example in Figure
14
: some sources, such as the one on the left, exhibit one or two LOFAR lobes flanking the central Ly
-emitting region, while others possess extended LOFAR emission that is spatially coincident with the Ly
emission but generally on much larger scales, as seen in the right panel. The morphologies of these Ly
-radio sources suggest that many LANs harbor AGN, even those LAEs not previously classified as hosting AGN.
Figure 14.
Multiwavelength overlays for two HETDEX-detected sources exhibiting extended Ly
emission. Left: 60″ wode region centered on
HLAN5002600476
= 2.23). Right: 50″ wide region centered on
HLAN4028050917
= 2.26), a HETDEX AGN source identified by Ly
+ CIV emission. The gray-scale background displays
-band imaging from HSC-SSP. Overlaid in blue/green are LOFAR 144 MHz radio continuum contours, and in orange/pink are HETDEX Ly
narrowband emission contours. The LOFAR contours trace extended radio emission, often associated with AGN activity or star-forming regions, while the Ly
contours reveal the spatial extent of ionized gas at high redshift. A 100 kpc scale bar is shown in the lower left of each panel.
Download figure:
Standard image
High-resolution image
Figure
13
compares the Ly
and radio properties of every HETDEX–LOFAR match in the Spring field. Resolved sources generally lie toward the upper envelope of both panels, underscoring that the most extended LANs also tend to host the brightest and most spatially extended radio emission. This figure highlights the size and flux selection effects discussed above and provides motivation for the morphological categories shown in Figure
14
7. Summary
This study measures the extended surface-brightness profiles of an emission-line-selected sample of Ly
-emitting galaxies drawn from the Hobby–Eberly Telescope Dark Energy Experiment (HETDEX). We model 70,691 LAEs as a point-source component with an optional exponential envelope. From this parent sample, we identify 33,612 LANs that are better described by the core+exponential envelope model, representing nearly half (47.5%) of all LAEs in the parent sample. We also provide area measurements based on 2
isophotal contours, as is common in LAB studies, and an effective isophotal radius measured in circular annuli. While this value trends with Ly
flux and luminosity, we show that the exponential scale length of the sample is not Ly
luminosity dependent
Roughly 12% of LANs coincide with AGN identified in the HETDEX AGN catalogs of C. Liu et al. (
2022
) and A. R. Liddle (
2025
). Not all AGN exhibit extended emission, with those LAEs with the highest Ly
line fluxes preferring the single PSF component model. HSC-
imaging reveals diverse morphologies, ranging from compact, continuum-bright sources to diffuse, continuum-faint systems with rest-frame equivalent widths exceeding 100 Å. Crossmatching with LOFAR radio data shows that radio counterparts are more common among the largest LANs.
Flux recovery comparisons confirm that PSF-based extractions systematically underestimate total Ly
flux. For HETDEX’s median seeing (FWHM ≃ 1
8), extended-aperture fluxes are, on average, 30% times larger than pipeline point-source values, implying that a significant fraction of Ly
light lies outside the core component.
The resulting
HETDEX LAN Catalog
, which accompanies this paper, provides positions, redshifts, and morphological and photometric parameters for the full 70,691 LAEs. It can be found at
and in the electronic version of this paper. This catalog establishes the largest statistical census of extended Ly
emission to date, bridging the regimes of compact Ly
halos and luminous Ly
blobs across Cosmic Noon.
Acknowledgments
E.M.C. gratefully acknowledges the late Peter Erwin, developer of
pyimfit
(P. Erwin
2015
). His exceptionally efficient surface-brightness-modeling code made this work possible, enabling us to fit hundreds of thousands of objects over many iterative analyses.
HETDEX is led by the University of Texas at Austin McDonald Observatory and Department of Astronomy, with participation from the Ludwig-Maximilians-Universität München, Max-Planck-Institut für Extraterrestrische Physik (MPE), Leibniz-Institut für Astrophysik Potsdam (AIP), Texas A&M University, The Pennsylvania State University, Institut für Astrophysik Göttingen, The University of Oxford, Max-Planck-Institut für Astrophysik (MPA), The University of Tokyo, and Missouri University of Science and Technology. In addition to Institutional support, HETDEX is funded by the National Science Foundation (grant AST-0926815), the State of Texas, the US Air Force (AFRL FA9451-04-2-0355), and generous support from private individuals and foundations.
Observations for HETDEX were obtained with the Hobby–Eberly Telescope (HET), which is a joint project of the University of Texas at Austin, the Pennsylvania State University, Ludwig-Maximilians-Universität München, and Georg-August-Universität Göttingen. The HET is named in honor of its principal benefactors, William P. Hobby and Robert E. Eberly.
The Visible Integral-field Replicable Unit Spectrograph (VIRUS) was used for HETDEX observations. VIRUS is a joint project of the University of Texas at Austin, Leibniz-Institut für Astrophysik Potsdam (AIP), Texas A&M University (TAMU), Max-Planck-Institut für Extraterrestrische Physik (MPE), Ludwig-Maximilians-Universität München, Pennsylvania State University, Institut für Astrophysik Göttingen, University of Oxford, and the Max-Planck-Institut für Astrophysik (MPA). In addition to Institutional support, VIRUS was partially funded by the National Science Foundation, the State of Texas, and generous support from private individuals and foundations.
The authors acknowledge the Texas Advanced Computing Center (TACC) at The University of Texas at Austin for providing high-performance computing, visualization, and storage resources that have contributed to the research results reported within this paper. URL:
The Institute for Gravitation and the Cosmos is supported by the Eberly College of Science and the Office of the Senior Vice President for Research at the Pennsylvania State University. The Kavli IPMU is supported by the World Premier International Research Center Initiative (WPI), MEXT, Japan.
K.G. acknowledges support from NSF-2008793.
Facility:
HET - McDonald Observatory's Hobby-Eberly Telescope.
Software:
This research was made possible by the open-source projects
hetdex-api
pyimfit/imfit
(P. Erwin
2015
),
astropy
(Astropy Collaboration et al.
2018
),
scipy
(P. Virtanen et al.
2020
),
Python
(G. Van Rossum & F. L. Drake
2009
), and
numpy
(C. R. Harris et al.
2020
).
Footnotes
21
22
Please wait… references are loading.
10.3847/1538-4357/ae44f3