SGA-2025

The SGA-2025 delivers multiwavelength imaging mosaics and ellipse photometry for nearly half a million galaxies with angular diameter \(>30~\mathrm{arcsec}\) across approximately 30,000 deg² of the extragalactic sky.

This page describes the organization of the public data release (v1.0), the naming conventions used throughout, and the data model of the primary science-ready merged catalogs.

Warning

SGA-2025 is not yet publicly released. This page, the linked catalogs, and the web galleries are in draft form, subject to change without notice, and should not be cited, redistributed, or relied upon as final data products.

Data Access

Merged catalogs (right-click → Save link as to download) (see Merged Catalogs for the data model and Imaging Regions for region descriptions):

Web visualizations (work in progress):

Per-group data files — for bulk download via wget or Globus (see Per-Group Data Files for the full file inventory):

Naming Conventions

We use two layers of identifiers throughout the SGA-2025: groups (mosaics containing one or more spatially/angularly associated galaxies) and objects (individual galaxies within each group). Most objects in the sample are in singleton groups.

Groups: GROUP_NAME and SGAGROUP

We process each SGA-2025 mosaic as a group — a set of one or more galaxies whose angular sizes and separations place them within the same mosaic footprint. The mosaic is centered on the diameter-weighted centroid of its members.

Two catalog columns identify a group:

  • GROUP_NAME (str10, e.g. 00000m0649) — the bare group name, used as the on-disk directory name.

  • SGAGROUP (str18, e.g. SGA2025_00000m0649) — the prefixed form, used as the file-name prefix for all per-group output files inside that directory.

The encoding is RRRRR[p|m]DDDD where RRRRR = RA × 100 (five digits, zero-padded), the separator is p (positive Dec) or m (negative Dec), and DDDD = abs(Dec) × 100 (four digits, zero-padded), both at a precision of 0.01° (36 arcsec). For example, a group centered at RA = 0.00°, Dec = −6.49° has GROUP_NAME = 00000m0649.

In groups with more than one member (GROUP_MULT > 1), GROUP_PRIMARY identifies the largest galaxy in the group based on its initial diameter. GROUP_RA and GROUP_DEC give the diameter-weighted centroid; GROUP_DIAMETER (in arcmin) gives the diameter of the smallest circle enclosing all member ellipses.

Objects: SGAID and SGANAME

Individual galaxies within a group are identified by:

  • SGAID (int64) — a unique integer identifier assigned to each source.

  • SGANAME (str25, e.g. SGA2025 J000.000−006.490) — an IAU-style coordinate name at 0.001° (3.6 arcsec) precision, formed from the fitted galaxy coordinates RA and DEC.

  • OBJNAME (str30) — the common name from the input parent catalog (HyperLeda, NED, LVD, etc.).

Galaxy Names: GALAXY and ALTNAMES

GALAXY (str35) is the primary human-readable name assigned to each source; ALTNAMES (str200) holds up to three additional, pipe-separated cross-identifications. Both are derived from NED’s cross-identification list for each object, guided by a few principles rather than a fixed procedure:

  • Prefer well-known, NED-searchable catalog names. Where an object has multiple valid identifiers, we favor “famous” designations – Messier, NGC, IC, UGC, ESO, and similar well-established catalogs – over obscure or machine-generated identifiers, and we favor a galaxy’s informal name (e.g. “Large Magellanic Cloud”, “Fornax Dwarf Spheroidal”) when one exists and is unambiguous.

  • Prefer names that actually identify the galaxy. NED sometimes attaches a supernova, nova, or other transient’s discovery name (e.g. ASASSN-16pd, SN 2008C HOST) to a host galaxy’s cross-match list; we treat these as poor primary names and prefer an actual galaxy identifier when one is available.

  • Normalize formatting for consistency. Catalog-name prefixes are cased consistently (e.g. MRK, ARP, PGC) and separator spacing is corrected where a name arrived without it (e.g. HyperLeda’s raw PGC513380 becomes PGC 513380).

  • Fall back gracefully. If NED returned no cross-match at all, we adopt the parent catalog’s own object name. A small, hand-curated list of overrides covers well-known objects – mostly Local Group dwarf galaxies – whose best name can’t be derived automatically.

ALTNAMES collects the remaining cross-identifications after GALAXY is chosen, deduplicated and prefix-normalized in the same way.

Imaging Regions

SGA-2025 covers two (mostly) non-overlapping imaging regions drawn from the DESI Legacy Imaging Surveys Data Release 11 (DR11) , which together span approximately 30,000 deg² of the sky (or 90% of the extragalactic sky at Galactic latitude |b|>10).

dr11-south uses g-, r-, i-, and z-band imaging from the Dark Energy Camera (DECam) at the 4-m Victor M. Blanco Telescope at Cerro Tololo Inter-American Observatory (CTIO).

dr11-north combines two surveys covering the sky north of roughly Dec ≈ +32°:

  • BASS (Beijing-Arizona Sky Survey) — provides g- and r-band imaging from the 90-Prime wide-field imager at the University of Arizona’s 2.3-m Bok telescope on Kitt Peak National Observatory.

  • MzLS (Mayall z-band Legacy Survey) — provides z-band imaging obtained with the Mosaic-3 camera at the NOAO 4-m Mayall telescope on Kitt Peak National Observatory.

The northern region therefore has three optical bands (g, r, z) with no i-band coverage. The BANDS column in the merged catalog records which optical bands are actually available for each galaxy.

In addition to the optical imaging, the SGA-2025 also includes:

  • Infrared: unWISE W1 (3.4 μm) and W2 (4.6 μm) and WISE W3 (12 μm) and W4 (22 μm) mosaics, derived from the full-depth coadded WISE satellite imaging.

  • Ultraviolet: GALEX FUV (~1530 Å) and NUV (~2310 Å) imaging (where GALEX sky coverage exists).

Merged Catalogs

The per-region merged catalogs combine all per-group ellipse photometry, NASA/IPAC Extragalactic Database (NED) cross-matched results, spectroscopic redshifts from the Dark Energy Spectroscopic Instrument Data Release 1 (DESI/DR1) and SDSS DR17, and direct distances from the Local Volume Database (LVD) and NED into a single FITS file per imaging region.

These catalogs are the starting point for most science use cases.

Catalog Files

We provide two versioned FITS files, one per imaging region (see Data Access for direct download links):

SGA2025-dr11-north-v1.0.fits    ( 90,504 galaxies, 340 MB)
SGA2025-dr11-south-v1.0.fits    (395,435 galaxies, 1.4 GB)

Each file has three binary table extensions:

  • SGA2025 — the primary science catalog: identification, geometry, cross-matched redshifts and distances, and a nominal flux per band for every galaxy.

  • ELLIPSEPHOT — the full aperture-photometry and curve-of-growth grid behind the nominal fluxes, row-matched to SGA2025 via SGAID (its leading column).

  • TRACTOR — Tractor source-fitting parameters, row-matched to SGA2025 via REF_ID (duplicated as a leading SGAID column too).

Reading the Catalogs

The catalogs are standard FITS binary tables. The recommended way to read them is via fitsio wrapped in an astropy Table:

import fitsio
from astropy.table import Table

sga     = Table(fitsio.read('SGA2025-dr11-south-v1.0.fits', ext='SGA2025'))
phot    = Table(fitsio.read('SGA2025-dr11-south-v1.0.fits', ext='ELLIPSEPHOT'))
tractor = Table(fitsio.read('SGA2025-dr11-south-v1.0.fits', ext='TRACTOR'))

fitsio also supports reading a subset of columns:

sga = Table(fitsio.read('SGA2025-dr11-south-v1.0.fits', ext='SGA2025',
                        columns=['GALAXY', 'RA', 'DEC', 'D26', 'Z']))

Common selections:

# Primary galaxies (unique list of groups)
I = sga['GROUP_PRIMARY']

# Diameter cuts (arcmin)
I = (sga['D26'] > 1.) & (sga['D26'] < 10.)

# Specific galaxies by name
I = np.isin(sga['GALAXY'], ['Messier 065', 'NGC 3621'])

# Local Volume Database objects (SAMPLE bit 1)
I = (sga['SAMPLE'] & 1) != 0

# Groups with 2–5 members
I = (sga['GROUP_MULT'] >= 2) & (sga['GROUP_MULT'] <= 5)

# Galaxies with a measured redshift
I = sga['Z_IVAR'] > 0

Data Model: SGA2025 Extension

The SGA2025 extension contains one row per galaxy. We organize columns into the groups below.

Identification and sky position

Column

Type

Description

SGAID

int64

Unique SGA identifier

SGANAME

str25

IAU-style name (SGA2025 JXXX.XXX±YY.YYY)

GALAXY

str35

Primary NED name (first entry from NED cross-ID list)

ALTNAMES

str200

Up to three additional NED cross-IDs, pipe-separated

OBJNAME

str30

Common galaxy name from the parent catalog

PGC

int32

PGC identifier (0 if unavailable)

RA, DEC

float64

Fitted galaxy coordinates (deg)

RA_TRACTOR, DEC_TRACTOR

float64

Tractor-fitted position (deg)

MORPH

str30

NED morphological type string

Status and quality bitmasks

Column

Type

Description

SGAGROUP

str18

Prefixed group name (e.g. SGA2025_00000m0649)

REGION

int16

Imaging region bitmask (see Bitmasks)

SAMPLE

int32

Sample classification bitmask (see Bitmasks)

ELLIPSEBIT

int32

Output quality and status flags (see Bitmasks)

ELLIPSEMODE

int32

Input fitting-mode flags requested for this object

BANDS

str4

Optical bands available (e.g. griz, grz)

Adopted redshift (see Redshifts for resolution logic)

Column

Type

Description

Z

float64

Adopted redshift (0 = no measurement; use Z_IVAR > 0 to test)

Z_IVAR

float64

Inverse variance of adopted redshift (0 = no measurement)

Z_REF

str4

Source of adopted redshift: LVD, DESI, SDSS, NED, or blank

Z_FLAG

int16

Redshift quality flag bitmask (see Bitmasks)

Adopted distance (see Distances for resolution logic)

Column

Type

Description

DIST

float32

Adopted distance (Mpc; 0 = no measurement)

DIST_IVAR

float32

Inverse variance of adopted distance (Mpc−2; 0 = no measurement)

DIST_REF

str10

Source of adopted distance: NED_DIRECT, LVD, NED, or ZCOSMO

DIST_METHOD

str16

Specific distance technique (e.g. TRGB, SBF, LVD’s own reported method, NED_MEAN, or HUBBLE_FLOW)

Photometric redshift (see Redshifts for details)

Column

Type

Description

Z_PHOT

float64

Random-forest photometric redshift (0 = no measurement; use Z_PHOT_IVAR > 0 to test)

Z_PHOT_IVAR

float64

Inverse variance of Z_PHOT (0 = no measurement)

Peculiar-velocity-corrected redshift (see Redshifts for details)

Column

Type

Description

Z_COSMO

float64

CMB-frame redshift with the mean pvhub-predicted peculiar velocity removed (0 = no measurement; use Z_COSMO_IVAR > 0 to test)

Z_COSMO_IVAR

float64

Inverse variance of Z_COSMO (0 = no measurement)

Group membership

Column

Type

Description

GROUP_NAME

str10

Bare group name (= on-disk directory name)

GROUP_MULT

int16

Group multiplicity (i.e., number of group members)

GROUP_PRIMARY

bool

True for the primary (largest) group member based on the initial diameter

GROUP_RA, GROUP_DEC

float64

Diameter-weighted group centroid coordinates (deg)

GROUP_DIAMETER

float32

Angular diameter of the full group (arcmin)

Fitted geometry

Column

Type

Description

D26

float32

Isophotal diameter at μ = 26 mag arcsec⁻² (arcmin)

D26_ERR

float32

1σ uncertainty on D26 (arcmin)

D26_REF

str3

Band from which D26 was derived (one of g, r, i, z; mom if the moment-based SMA was the only available estimate; fix when the FIXGEO ELLIPSEMODE bit is set and the geometry was held fixed)

BA

float32

Final axis ratio b/a

PA

float32

Final position angle (deg, N through E)

SMA_MOMENT

float32

Semi-major axis from flux-weighted moments (arcsec)

BA_MOMENT

float32

Axis ratio b/a from moments

PA_MOMENT

float32

Position angle from moments (deg, N through E)

SMA50_{G,R,I,Z} / SMA50_ERR_{G,R,I,Z}

float32

Optical half-light semi-major axis ± uncertainty, from the curve-of-growth fit (arcsec); IR/UV bands are in ELLIPSEPHOT

Nominal photometry (default aperture/method; see the ELLIPSEPHOT HDU for the full aperture grid and curve-of-growth fits)

Column

Type

Description

SMA_FLUX

float32

Aperture semi-major axis the nominal flux was measured within (arcsec); −1 for a source (e.g. a curve-of-growth total) with no finite aperture

FLUX_REF

str8

Source of the nominal flux, per galaxy (currently AP04 for every galaxy, one of the aperture radii in ELLIPSEPHOT; future releases may select a different aperture or method per galaxy)

FLUX_{band}

float32

Nominal flux (nanomaggies); all 10 bands

FLUX_IVAR_{band}

float32

Inverse variance of FLUX_{band} (nanomaggies−2); all 10 bands

Galaxy extinction and image depth

Column

Type

Description

EBV

float32

Milky Way E(B−V) from Schlegel et al. (1998) (mag)

MW_TRANSMISSION_{band}

float32

Milky Way transmission fraction (0–1); all 10 bands

PSFSIZE_{G,R,I,Z}

float32

Effective PSF FWHM in each optical band (arcsec)

PSFDEPTH_{G,R,I,Z}

float32

5σ point-source depth in each optical band (mag)

DESI DR1 spectroscopy

The columns below record the single best DESI spectrum selected for each galaxy (see Redshifts). Sentinel values: TARGETID_DESI and ZWARN_DESI are −1 when no spectrum was selected; Z_DESI and Z_IVAR_DESI are 0; DELTACHI2_DESI is −1; string columns are blank.

Column

Type

Description

TARGETID_DESI

int64

DESI target identifier (−1 if no spectrum)

SURVEY_DESI

str7

DESI survey (e.g. main, sv1, sv3)

PROGRAM_DESI

str6

DESI observing program (e.g. bright, dark)

Z_DESI

float64

DESI spectroscopic redshift (0 if no valid spectrum)

Z_IVAR_DESI

float64

Inverse variance of Z_DESI (0 if no valid spectrum)

ZWARN_DESI

int64

DESI redshift warning bitmask (−1 if no spectrum)

DELTACHI2_DESI

float64

DESI Δχ² between best and second-best redshift fit (−1 if no spectrum)

SPECTYPE_DESI

str6

DESI spectral classification (e.g. GALAXY, QSO, STAR)

NSPEC_DESI

int32

Number of DESI spectra (any ZWARN) within the SGA ellipse (0 if none)

SDSS DR17 spectroscopy

The closest quality-filtered SDSS DR17 spectrum matched within 3.5 arcsec of the galaxy position (see Redshifts). Sentinels: Z_SDSS and Z_IVAR_SDSS are 0; string columns are blank.

Column

Type

Description

SPECOBJID_SDSS

str22

SDSS spectrum identifier (encodes plate/mjd/fiber/run2d; blank if no match)

Z_SDSS

float64

SDSS spectroscopic redshift (0 if no match)

Z_IVAR_SDSS

float64

Inverse variance of Z_SDSS (0 if no match)

CLASS_SDSS

str6

SDSS spectral classification (GALAXY or QSO)

NED redshift and distance

Column

Type

Description

Z_NED

float64

NED redshift (0 = no NED redshift)

Z_IVAR_NED

float64

Inverse variance of Z_NED; derived from NED_Z_UNC. When NED_Z_UNC = 0 (error not reported in NED), a nominal uncertainty of 100 km/s is assumed. Set to 0 only when no NED redshift exists.

Z_REF_NED

str

NED reference for the redshift

DIST_NED

float32

NED distance (Mpc; 0 = no NED distance)

DIST_IVAR_NED

float32

Inverse variance of DIST_NED (Mpc−2; 0 if no NED distance or no formal uncertainty was reported)

DIST_NED_DIRECT

float32

NED “gold-tier” direct distance, from a single high-precision measurement from the literature (Mpc; 0 = none available)

DIST_IVAR_NED_DIRECT

float32

Inverse variance of DIST_NED_DIRECT (Mpc−2; 0 if no uncertainty was reported)

DIST_NED_DIRECT_METHOD

str16

Distance technique used for DIST_NED_DIRECT (e.g. TRGB, SBF, Cepheids)

NED_FLAG

int32

NED cross-match quality flag (see Bitmasks)

NED_Z_FLAG

int32

NED redshift-resolution flag (see Bitmasks)

LVD redshift and distance

These columns are populated only for objects in the Local Volume Database (SAMPLE & LVD != 0). Velocities are converted from km/s using Z = v_los / c.

Column

Type

Description

Z_LVD

float64

LVD systemic redshift from vlos_systemic (0 if not in LVD)

Z_IVAR_LVD

float64

Inverse variance of Z_LVD from vlos_sigma (0 if no LVD velocity error)

DIST_LVD

float32

LVD distance (Mpc, converted from kpc; 0 if not in LVD)

DIST_IVAR_LVD

float32

Inverse variance of DIST_LVD (Mpc−2; 0 if no LVD distance uncertainty)

Initial geometry

These columns record the geometry adopted from the input parent catalog, before any ellipse fitting, for comparison against the final fitted values. INIT_REF encodes the source of the initial diameter: one of SGA2020, HYPERLEDA, LIT, LVD, SMUDGes, RC3, or VI (visually inspected and added during processing).

Column

Type

Description

RA_INIT, DEC_INIT

float64

Parent-catalog coordinates used to initialize processing (deg; −1 if not available)

DIAM_INIT

float32

Initial diameter from the parent catalog (arcmin; −1 if not available)

BA_INIT

float32

Initial axis ratio b/a from the parent catalog (−1 if not available)

PA_INIT

float32

Initial position angle from the parent catalog (deg, N through E; −1 if not available)

MAG_INIT

float32

Initial magnitude from the parent catalog (mag; −1 if not available)

OPTFLUX

float32

Optical reference flux used to set source-masking priority within a group (nanomaggies; 0 if not set)

SMA_MASK

float32

Semi-major axis used for neighbor masking during ellipse fitting (arcsec; 0 if not set); overrides the moment-based SMA when positive

INIT_REF

str14

Source catalog for DIAM_INIT (blank if not available)

Data Model: ELLIPSEPHOT Extension

The ELLIPSEPHOT extension is row-matched to SGA2025 via SGAID (its leading column) and holds the full aperture-photometry and curve-of-growth grid behind SGA2025’s nominal FLUX_{band} columns, along with related morphology and radius measurements most analyses won’t need directly.

Morphology and total magnitudes — all 10 bands: G, R, I, Z, W1, W2, W3, W4, FUV, NUV

Column pattern

Description

GINI_{band}

Gini coefficient of the light distribution

COG_MTOT_{band} / COG_MTOT_ERR_{band}

Curve-of-growth total magnitude ± uncertainty (mag)

COG_DMAG_{band} / COG_DMAG_ERR_{band}

COG aperture correction ± uncertainty (mag)

COG_LNALPHA1_{band} / COG_LNALPHA1_ERR_{band}

COG Sérsic inner shape parameter ± uncertainty

COG_LNALPHA2_{band} / COG_LNALPHA2_ERR_{band}

COG Sérsic outer shape parameter ± uncertainty

COG_CHI2_{band}

COG fit χ² (float32)

COG_NDOF_{band}

COG fit degrees of freedom (int32)

Half-light radius — IR/UV bands only (optical SMA50_{G,R,I,Z} is in SGA2025)

Column pattern

Description

SMA50_{band} / SMA50_ERR_{band}

Half-light semi-major axis ± uncertainty, from the curve-of-growth fit (arcsec); W1, W2, W3, W4, FUV, NUV

Isophotal radii — optical bands G, R, I, Z only

Column pattern

Description

R{22,23,24,25,26}_{band}

Isophotal semi-major axis at μ = 22–26 mag arcsec⁻² (arcsec)

R{22,23,24,25,26}_ERR_{band}

1σ uncertainty on the isophotal radius (arcsec)

Aperture photometry — all 10 bands

Column pattern

Description

SMA_AP{00–04}

Aperture semi-major axes (arcsec); AP00–AP04 correspond to [0.5, 1.0, 1.25, 1.5, 2.0] × SMA_MOMENT

FLUX_AP{00–04}_{band}

Aperture flux (nanomaggies)

FLUX_ERR_AP{00–04}_{band}

Aperture flux uncertainty (nanomaggies)

FMASKED_AP{00–04}_{band}

Masked pixel fraction within aperture (0–1)

SGA2025’s FLUX_{band}/FLUX_IVAR_{band} are drawn from one of these apertures per galaxy, recorded in FLUX_REF and SMA_FLUX.

Data Model: TRACTOR Extension

The TRACTOR extension is row-matched to SGA2025 via REF_ID (duplicated as a leading SGAID column) and contains Tractor source-fitting parameters for each SGA galaxy and all sources detected within its ellipse. See the Legacy Surveys documentation for the full Tractor data model.

Redshifts

The merged catalogs cross-match four independent redshift sources — DESI DR1, SDSS DR17, NED, and the Local Volume Database (LVD).

Adopted redshift (Z, Z_IVAR, Z_REF)

Priority order LVD > DESI > SDSS > NED:

  1. Z_REF = LVD when Z_IVAR_LVD > 0.

  2. Z_REF = DESI when Z_IVAR_DESI > 0.

  3. Z_REF = SDSS when Z_IVAR_SDSS > 0.

  4. Z_REF = NED when Z_IVAR_NED > 0.

  5. Z_REF = '' (blank) when no source has a valid redshift.

Important

Use Z_IVAR > 0 to test for a valid redshift. Z = 0 is the sentinel for no measurement, but is also a valid (if rare) redshift for very nearby galaxies.

DESI DR1 spectrum selection

We apply this to all spectra within 3.5 arcsec of the SGA position:

  1. Keep only ZCAT_PRIMARY = True.

  2. Drop SPECTYPE = STAR, and any GALAXY/QSO spectrum with Z > 1.0.

  3. Prefer SPECTYPE = GALAXY with ZWARN = 0; otherwise any ZWARN = 0.

  4. Take the highest DELTACHI2 among survivors.

Z_IVAR_DESI = 0 if no spectrum survives. NSPEC_DESI counts all DESI spectra within the ellipse regardless of selection outcome.

SDSS DR17 spectrum selection

We match against SDSS DR17 specObj using a single fixed-radius (3.5 arcsec) positional search, pre-filtered to SPECPRIMARY = 1, CLASS in (GALAXY, QSO), and clean redshift warnings (ZWARNING_NOQSO = 0 for QSOs, ZWARNING = 0 for galaxies). We drop Z > 1.0 as physically implausible for this sample (mirroring the DESI cut above), and keep the closest remaining match when more than one falls within the radius. This match is independent of NED’s own SDSS-derived redshifts (see NED_Z_FLAG below) and takes priority over them, since NED does not propagate SDSS’s own quality flags.

NED redshift resolution

We replace NED’s aggregated (“fiducial”) Z_NED with a value drawn from NED’s full redshift-measurement list for the object when it:

  • exceeds a threshold (0.4 by default),

  • is missing entirely, or

  • comes from a censored source: the Bilicki et al. (2014) 2MASS photometric-redshift catalog, or an SDSS-compilation reference (superseded by the direct SDSS DR17 match above).

We group candidate measurements into clusters of mutually agreeing values; if one cluster dominates, we adopt its highest-quality measurement. Otherwise, or if no candidates exist, we flag the result rather than silently trust it (see NED_Z_FLAG / Z_FLAG in Bitmasks).

Consensus override

DESI always wins normal priority when measured, so it is the only source that can end up adopted despite being the outlier: when DESI, SDSS, and NED are all measured and SDSS/NED agree with each other but not with DESI, we replace the adopted redshift with SDSS’s and flag it CONSENSUS_OVERRIDE (see Bitmasks).

Automatic NED override

This sample is overwhelmingly z < 0.3, so we treat an adopted DESI/SDSS redshift above 0.3 with an independent NED measurement on file as a fiber landing on a background source: we replace the adopted redshift with NED’s and flag it AUTO_NED_OVERRIDE (see Bitmasks).

Manual VI overrides

We resolve a small residual of flagged redshifts (see Z_FLAG in Bitmasks) by hand and feed them back in as a worksheet CSV keyed by SGAID, overriding or discarding the adopted redshift and setting VI_REVIEWED.

Photometric redshift (Z_PHOT, Z_PHOT_IVAR)

We train a scikit-learn RandomForestRegressor on the subsample of galaxies with an adopted spectroscopic redshift, using Milky Way extinction-corrected optical/infrared colors, an apparent magnitude, mean surface brightness, angular diameter, aspect ratio (b/a), and the Tractor Sérsic index as features. Apparent magnitude and surface brightness, \(m\) and \(\langle\mu\rangle\), respectively, use the r band, falling back to i and then g when r is unavailable, where

\[\langle\mu\rangle = m + 2.5\,\log_{10}\!\left(\frac{\pi}{4}\,\frac{b}{a}\,(60\,D_{26})^2\right)\]

and D26 is the major-axis diameter at the \(26~\mathrm{mag~arcsec}^{-2}\) isophote in arcmin.

We validate the model with 5-fold cross-validation, then retrain it on the full spectroscopic subsample and predict Z_PHOT for every galaxy in the catalog. Z_PHOT_IVAR is the inverse variance of the spread of predictions across the individual trees of the forest.

Feature

Importance

g − r

0.5622

Angular diameter (D26)

0.0805

Apparent magnitude \(m\)

0.0786

z − W1

0.0559

Surface brightness \(\langle\mu\rangle\)

0.0538

W1 − W2

0.0510

Sérsic index (SERSIC)

0.0312

r − i

0.0286

NUV − g

0.0196

Aspect ratio b/a (BA)

0.0157

i − z

0.0153

Peculiar-velocity-corrected redshift (Z_COSMO, Z_COSMO_IVAR)

For every galaxy with Z_IVAR > 0, we convert the heliocentric Z to the CMB frame using the standard dipole (Planck 2018: 369.82 km/s toward RA = 167.942°, Dec = −6.944°, J2000).

Below 20,000 km/s, pvhub (Said et al.) predicts the line-of-sight peculiar velocity from four independent linear reconstructions (2M++-SDSS, 2M++-SDSS+6dF, 2MRS, 2M++); Z_COSMO is the CMB-frame redshift with their mean peculiar velocity removed, and Z_COSMO_IVAR combines the model-to-model scatter with the measurement uncertainty in quadrature. Beyond 20,000 km/s, Z_COSMO is simply the CMB-frame redshift and Z_COSMO_IVAR carries over Z_IVAR unchanged.

When this leaves Z_COSMO undefined, we substitute, in priority order:

  1. The equivalent Hubble-flow redshift from an independently measured distance (NED_DIRECT, LVD, or NED; see Distances), inverted through our fiducial cosmology.

  2. The CMB-frame-converted photometric redshift, Z_PHOT, with Z_COSMO_IVAR set from Z_PHOT_IVAR.

Distances

We assign distances from four sources, in priority order: high-precision “gold-tier” measurements compiled by NED, the Local Volume Database (LVD), NED’s own blended “mean” distance, and — for every remaining galaxy with a measured redshift — a luminosity distance derived from Z_COSMO under our fiducial cosmology, flat ΛCDM with H0 = 70 km/s/Mpc and Ωm = 0.3.

Adopted distance (DIST, DIST_IVAR, DIST_REF, DIST_METHOD)

Priority order NED direct > LVD > NED mean > Hubble flow:

  1. DIST_REF = NED_DIRECT when a high-precision “gold-tier” distance is available from NED, in priority order Cepheids, TRGB, PNLF, SBF, SNIa, SNII, GCLF, SNIa-SDSS. The highest-priority method present must be confirmed by at least 3 independent references. If it is not, and a lower-priority method is confirmed by at least 3 references, agrees with itself to 15%, and disagrees with the top method by more than 15%, we adopt that lower-priority method’s mean and standard deviation instead.

  2. DIST_REF = LVD when DIST_LVD > 0 (converted from kpc to Mpc).

  3. DIST_REF = NED when DIST_NED > 0 (NED’s blended “mean” distance).

  4. DIST_REF = ZCOSMO for every remaining galaxy with a measured redshift: the luminosity distance at Z_COSMO under our fiducial cosmology.

DIST_METHOD records the specific technique behind the adopted value. For any adopted distance with no reported uncertainty, we assign a 15% fractional floor rather than DIST_IVAR = 0.

Note

Gold-tier method abbreviations, in priority order, with their typical distance-modulus uncertainty (mag) inferred from the values compiled in NED:

Method

Full name

Typical uncertainty

Cepheids

Cepheid variable stars

0.08 mag

TRGB

Tip of the Red Giant Branch

0.10 mag

PNLF

Planetary Nebula Luminosity Function

0.13 mag

SBF

Surface Brightness Fluctuations

0.15 mag

SNIa

Type Ia supernovae

0.16 mag

SNII

Type II supernovae (optical)

0.16 mag

GCLF

Globular Cluster Luminosity Function

0.17 mag

SNIa SDSS

Type Ia supernovae, SDSS-II Supernova Survey

0.26 mag

Sentinel conventions

Column type

Sentinel / missing value

Redshifts and their IVARs (Z, Z_IVAR, etc.)

0 — use the paired *_IVAR > 0 column to test for a valid measurement

Distances and their IVARs (DIST, DIST_IVAR, etc.)

0 — same convention as redshifts

Other float columns

−1 (strictly negative, unphysical for distances and separations)

Integer IDs (TARGETID_DESI, etc.)

−1

String columns

blank (empty string)

Bitmasks

SAMPLE

We set SAMPLE in the parent catalog and propagate it to the merged catalogs. Multiple bits may be set on the same object simultaneously.

Bit name

Value

Description

LVD

1

Source from the Local Volume Database of nearby galaxies

MCLOUDS

2

Located within the Magellanic Cloud footprint

GCLPNE

4

Located within a globular cluster or planetary nebula mask

NEARSTAR

8

Near a bright star (star–galaxy separation < 1.2 × Gaia/Tycho mask)

INSTAR

16

Inside a bright star (star–galaxy separation < 0.5 × Gaia/Tycho mask)

OVERLAP

32

Initial ellipse overlaps another (initial) SGA ellipse

ELLIPSEMODE — input fitting flags

We request this before processing; it describes the intended fitting strategy.

Bit name

Value

Description

FIXGEO

1

Fix the ellipse geometry to initial values (no free fitting)

RESOLVED

2

Milky Way satellite (effectively resolved into stars); Tractor fitting not performed

FORCEPSF

4

Force all Tractor sources to be PSF

FORCEGAIA

8

Force Tractor to only fit Gaia sources (not used)

LESSMASKING

16

Apply less aggressive masking during ellipse-fitting

MOREMASKING

32

Apply more aggressive masking during ellipse-fitting

MOMENTPOS

64

Use moment-based position instead of Tractor position

TRACTORGEO

128

Adopt Tractor geometry instead of ellipse-fit geometry

NORADWEIGHT

256

Disable radial weighting in when determining light-weighted geometry

ELLIPSEBIT — output fitting and quality flags

We set this after processing; it records the actual fitting outcome and data quality.

Bit name

Value

Description

NOTRACTOR

1

No Tractor model (dropped or never fitted)

TRACTORPSF

2

Tractor fit this source as a point source (PSF model)

FIXGEO

4

Ellipse geometry was fixed, not freely fitted

BLENDED

8

Source center is within the ellipse of another SGA source

LARGESHIFT

16

Large positional shift between initial and fitted ellipse center

LARGESHIFT_TRACTOR

32

Large positional shift between initial and Tractor center

MAJORGAL

64

Major galaxy requiring special processing treatment

OVERLAP

128

Fitted ellipse overlaps another SGA ellipse

SATELLITE

256

Satellite galaxy located within the ellipse of a larger system

MOMENTPOS

512

Moment-based position was adopted

TRACTORGEO

1024

Tractor geometry was adopted

NORADWEIGHT

2048

Radial weighting was not applied to isophote fitting

LESSMASKING

4096

Less aggressive neighbor masking was applied

MOREMASKING

8192

More aggressive neighbor masking was applied

FAILGEO

16384

Isophote fitting failed; D26 may be unreliable

SKIPTRACTOR

32768

Tractor fitting was skipped entirely for this mosaic

NED_FLAG — NED cross-match quality

Records how we resolved the NED cross-match for each galaxy. The byname query searches NED by galaxy name; the bypos query searches by sky position. We may set multiple bits simultaneously. NED_FLAG covers cross-match resolution only; see NED_Z_FLAG below for how NED’s own redshift was resolved, and Redshifts for how that feeds into Z_FLAG.

Bit name

Value

Description

LARGE_SEP

1

Byname positional offset exceeds threshold (3.5 arcsec); bypos result used instead

BYNAME_FAIL

2

Byname query returned an error or no match; bypos result used instead

BYPOS_FAIL

4

Bypos query also failed; NED cross-match result may be incomplete

Note

LVD galaxies are exempt from the LARGE_SEP flag: their NED coordinates can legitimately differ from the SGA centroid (extended dwarfs), so we always accept a successful byname match regardless of positional offset.

NED_Z_FLAG — NED redshift resolution

Records how we arrived at NED’s own default redshift, Z_NED, for each galaxy (see Redshifts). We may set multiple bits simultaneously.

Bit name

Value

Description

Z_SUSPECT

1

Adopted Z_NED exceeded the redshift-override threshold (0.4 by default); alternative NED redshift measurements were considered

Z_UNRESOLVED

2

Z_SUSPECT or Z_MISSING was set but no confident replacement redshift was found (disagreeing alternatives, or none below threshold); the adopted Z_NED should be treated with caution

Z_MISSING

4

NED’s own default redshift was missing entirely; Z_NED was adopted from NED’s full list of redshift measurements instead

Z_CENSORED_PHOTOZ

8

Adopted default was a photometric-redshift measurement (Bilicki et al. 2014 2MASS Photo-z catalog); discarded, alternates considered

Z_CENSORED_SDSS

16

Adopted default came from an SDSS-compilation reference; discarded in favor of the direct SDSS DR17 match (see Redshifts)

Z_FLAG — redshift quality

Records how we resolved the adopted, catalog-wide redshift (Z) across all sources — distinct from NED_Z_FLAG, which only describes NED’s own internal resolution.

Bit name

Value

Description

DISCREPANCY

1

Any two measured redshifts (LVD, DESI, SDSS, NED) differ by more than 1000 km/s in the rest frame. Deliberately sensitive – a coarse NED redshift can exceed this against a precise DESI/SDSS spectrum that is otherwise fine, so this bit alone does not imply the adopted redshift is wrong

NED_SUSPECT

2

NED is the sole redshift measurement (no LVD, DESI, or SDSS value) and its NED_Z_FLAG has Z_SUSPECT set

NED_UNRESOLVED

4

As NED_SUSPECT, and its NED_Z_FLAG also has Z_UNRESOLVED set: no confident replacement redshift was found, so the adopted redshift may still be wrong

NED_MISSING

8

NED is the sole redshift measurement (no LVD, DESI, or SDSS value) and its NED_Z_FLAG has Z_MISSING set: NED had no default redshift, so the adopted value came from NED’s full redshift list

CONSENSUS_OVERRIDE

16

DESI disagreed with SDSS and NED, which agreed with each other; we replaced the adopted redshift with SDSS’s

AUTO_NED_OVERRIDE

32

Adopted DESI/SDSS redshift exceeded 0.3, so we automatically replaced it with an independent NED measurement

NED_CENSORED_PHOTOZ

64

As NED_SUSPECT, and its NED_Z_FLAG has Z_CENSORED_PHOTOZ set: NED’s default was a discarded photometric-redshift measurement

NED_CENSORED_SDSS

128

As NED_SUSPECT, and its NED_Z_FLAG has Z_CENSORED_SDSS set: NED’s default was a discarded SDSS-compilation reference

VI_REVIEWED

256

A manual visual-inspection pass set or confirmed the adopted redshift; the final, human-authoritative bit

OPTMASKBITS, GALEXMASKBITS, UNWISEMASKBITS

We store per-band imaging mask bits in the per-group -maskbits.fits[.fz] files and the per-galaxy -ellipse-{griz,galex,unwise}.fits files. All three share the same low-order bits:

Bit name

Value

Description

brightstar

1

Bright star, medium star, or star cluster (legacypipe MASKBITS)

gaiastar

2

Gaia point source (type = PSF)

galaxy

4

Extended (non-reference) source

reference

8

SGA reference source

Per-band bits:

  • OPTMASKBITS: g = 16, r = 32, i = 64, z = 128

  • GALEXMASKBITS: FUV = 16, NUV = 32

  • UNWISEMASKBITS: W1 = 16, W2 = 32, W3 = 64, W4 = 128

Per-Group Data Files

Directory Layout

We organize the public data release as:

data/
  {region}/                  # dr11-south or dr11-north
    {raslice}/               # 3-digit integer part of GROUP_RA (000–359)
      {group_name}/          # e.g. 00000m0649
        SGA2025_{group_name}-*

The raslice is the zero-padded integer part of GROUP_RA (e.g. RA = 33.98° → 033). The group_name is the bare GROUP_NAME column value.

To construct the path to a group directory in Python:

import os

datadir = '/path/to/data'   # local copy of the data release
region  = 'dr11-south'      # or 'dr11-north'

for row in cat:
    raslice   = f'{int(row["GROUP_RA"]):03d}'
    group_name = row['GROUP_NAME'].strip()
    galaxydir  = os.path.join(datadir, region, raslice, group_name)
    prefix     = f'SGA2025_{group_name}'

File Inventory

We prefix all files in a group directory with g = SGA2025_{group_name}.

Two processing modes exist. Full mode applies to groups where we performed Tractor source fitting (the file {g}-tractor.fits is present). Coadds-only mode applies to groups where we produced only image mosaics (no Tractor fitting).

Files present in both modes:

File

Description

{g}-sample.fits

Parent-catalog rows for this group

{g}-ccds.fits

Metadata for all CCDs overlapping this mosaic

{g}-coadds.isdone

Sentinel file written when coadd step completes

{g}-ellipse.isdone

Sentinel file written when ellipse step completes

{g}-coadds.log

Coadd processing log

{g}-ellipse.log

Ellipse processing log

{g}-image.jpg

Optical grz color composite

{g}-image-W1W2.jpg

unWISE W1/W2 color composite

{g}-image-FUVNUV.jpg

GALEX FUV/NUV color composite

Full mode only:

File

Description

{g}-tractor.fits

Tractor source catalog for this mosaic

{g}-maskbits.fits.fz

Imaging mask-bit array (compressed)

{g}-image-{band}.fits.fz

Image mosaic per band (compressed); bands = g, r, i, z, FUV, NUV, W1–W4

{g}-invvar-{band}.fits.fz

Inverse-variance map per band (compressed)

{g}-model-{band}.fits.fz

Tractor model image per band (compressed)

{g}-psf-{band}.fits.fz

PSF image per band (compressed)

{g}-model.jpg

Optical grz Tractor model composite

{g}-model-W1W2.jpg

unWISE W1/W2 Tractor model composite

{g}-model-FUVNUV.jpg

GALEX FUV/NUV Tractor model composite

{g}-resid.jpg

Optical grz residual (image − model)

{g}-resid-W1W2.jpg

unWISE W1/W2 residual

{g}-resid-FUVNUV.jpg

GALEX FUV/NUV residual

Coadds-only mode only:

File

Description

{g}-maskbits.fits

Imaging mask-bit array (uncompressed)

{g}-image-{band}.fits

Image mosaic per band (uncompressed); at least one of g, r, i, z plus FUV, NUV, W1–W4

Per-Galaxy Ellipse Files

In addition to the per-group files, each galaxy within a group has up to three per-galaxy ellipse files named after the galaxy’s IAU coordinate name (without spaces):

SGA2025_J{RA}[+|-]{DEC}-ellipse-griz.fits
SGA2025_J{RA}[+|-]{DEC}-ellipse-galex.fits
SGA2025_J{RA}[+|-]{DEC}-ellipse-unwise.fits

Each file contains four FITS extensions:

Extension

Description

MODELS

Image cube of the ellipse model and data for each band

MASKBITS

Per-band imaging mask array (same footprint as MODELS)

ELLIPSE

Ellipse photometry table (one row per ellipse semi-major axis step)

SBPROFILES

Azimuthally averaged surface-brightness and color profiles

Release History

v1.0 (August 2026)

First public release.