BASS: BAT AGN Spectroscopic Survey

• BASS is an optical and near-IR spectroscopic survey targeting Swift/BAT ultra-hard X-ray AGN, providing a nearly unbiased census of local active galaxies.
• The survey combines dedicated observations with archival data to deliver comprehensive measurements of black hole masses, luminosities, and host-galaxy properties.
• BASS advances our understanding of AGN obscuration, dynamics, and multiwavelength calibrations, establishing a benchmark for AGN demographic and scaling studies.

The BAT AGN Spectroscopic Survey (BASS) is the optical and near-infrared spectroscopic program built around the all-sky Swift/BAT ultra-hard X-ray AGN catalog in the $14\text{--}195$ keV band. Its defining objective is a highly complete census of the key physical parameters of nearby active galactic nuclei, including black hole mass, bolometric luminosity, Eddington ratio, line-of-sight gas obscuration, and host-galaxy properties, using a parent selection that is minimally biased by dust and gas obscuration up to Compton-thick regimes. In practice, BASS links broadband X-ray spectroscopy to uniform optical and NIR measurements, making it a central reference set for local AGN demographics, obscuration studies, black-hole scaling work, and multiwavelength AGN calibrations (Koss et al., 2022, Koss et al., 2022, Koss et al., 2017).

1. Survey rationale and parent sample

BASS is anchored to the Swift/BAT all-sky survey, whose ultra-hard X-ray selection is far less affected by line-of-sight obscuration than optical or soft X-ray selection. This design reduces the classic biases against obscured sources and yields a census that includes both broad-line and narrow-line AGN, as well as many dusty systems underrepresented in optical spectroscopic surveys (Koss et al., 2022, Trakhtenbrot et al., 2017). BAT is largely insensitive to obscuration up to $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$, although the most heavily Compton-thick population can still be undercounted (Koss et al., 2022).

The early BASS program was built on the Swift/BAT 70-month catalog. Data Release 1 analyzed optical spectra for 641 of 836 BAT AGN, corresponding to 77% coverage, and already represented a significant hard-X-ray-selected census of the local universe, with about 90% of sources at $z<0.2$ (Koss et al., 2017). Data Release 2 expanded this framework to 858 AGN in the BAT 70-month sample and emphasized spectroscopic completeness, counterpart revision, and homogeneous measurement of line properties, velocity dispersions, black-hole masses, and accretion rates (Koss et al., 2022, Koss et al., 2022).

Counterpart identification is itself a nontrivial component of BASS because the BAT point-spread function is large. DR2 explicitly revised AGN counterparts across the full sample, identified dual or clustered BAT contributors, flagged flux-boosting cases, and separated beamed and lensed sources from the unbeamed AGN population used for most spectroscopic demographic work (Koss et al., 2022). This counterpart-cleaning function is integral rather than auxiliary: population inferences in BASS depend on a physically meaningful association between the BAT detection and the optical/NIR source being modeled.

2. Data releases and observational infrastructure

BASS combines dedicated observing campaigns with major public spectroscopic archives. The survey strategy favors broad wavelength coverage, high spectral resolution where needed for host-galaxy kinematics, and redundancy across facilities. DR2 is described as an unprecedented spectroscopic survey in spectral range, resolution, and sensitivity, with 1449 optical spectra over $\sim 3200\text{--}10000$ Å and 233 NIR spectra over $1\text{--}2.5\,\mu{\rm m}$ for the 858 BAT AGN (Koss et al., 2022).

Release or component	Scope	Principal content
DR1	641/836 AGN with optical spectra	First catalog of spectral measurements, derived quantities, and AGN demographics (Koss et al., 2017)
DR2 overview	1449 optical and 233 NIR spectra for 858 AGN	99.9% measured redshifts and 98% black hole masses for unbeamed AGN outside the Galactic plane (Koss et al., 2022)
NIR DR2	168 VLT/X-shooter NIR spectra; DR1+DR2 NIR sample of 266 AGN	High-ionization coronal lines and broad Paschen-line measurements (Brok et al., 2022)

The core optical facilities in DR2 include VLT/X-shooter, Palomar/DBSP, SDSS, SOAR/Goodman, Keck/LRIS, Magellan/MagE, VLT/FORS2, VLT/MUSE, Gemini/GMOS, and several additional instruments. Many DR2 spectra either have resolving power $R>2500$ or wide $3200\text{--}10000$ Å coverage, both of which are important for disentangling host-galaxy starlight, narrow-line kinematics, broad-line decomposition, and stellar velocity dispersions (Koss et al., 2022, Koss et al., 2022). In the obscured-AGN velocity-dispersion program, the best spectra are dominated by VLT/X-shooter, Palomar/DBSP, SDSS, and SOAR/Goodman (Koss et al., 2022).

The NIR branch of BASS is centered on VLT/X-shooter, whose simultaneous UVB, VIS, and NIR arms provide contiguous $0.3\text{--}2.5\,\mu{\rm m}$ coverage at medium resolution. The DR2 NIR study analyzed 168 nearby BAT AGN observed in the NIR arm, while the combined DR1+DR2 X-shooter-plus-earlier sample reached 266 unique BAT AGN, described as the largest rest-frame NIR spectroscopic set assembled to date for this population (Brok et al., 2022).

3. Spectroscopic methodology and derived quantities

The optical DR2 line-measurement framework is based on full-spectrum fitting rather than isolated local windows. Spectra are de-redshifted and corrected for Galactic extinction; the stellar continuum is modeled with pPXF using SSP templates and, for some VLT spectra, the X-shooter stellar library; emission lines are fitted simultaneously with gandalf; and a broad Balmer component is introduced when required, using Gaussians with ${\rm FWHM}>1000\ {\rm km\,s^{-1}}$. A detection requires amplitude-to-noise $A/N>3$, and otherwise a $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$0 upper limit is reported (Oh et al., 2022). This uniform procedure underlies the DR2 narrow-line catalog, from He II $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$1 through [S III] $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$2, and supports BPT-type classification, reddening work, and outflow analysis.

BASS adopts a prioritized strategy for black-hole mass estimation. The preferred order in DR2 is direct literature measurements, single-epoch virial masses from broad H$N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$3, then broad H$N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$4, then Mg II or C IV for high-$N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$5 beamed sources, then $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$6 from stellar velocity dispersion, with obscured broad-line masses treated cautiously when $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$7 (Koss et al., 2022). The general virial framework is

$N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$8

with $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$9 used consistently in DR2 single-epoch methods (Koss et al., 2022). For host-dominated AGN, BASS XXVI uses the Kormendy & Ho (2013) relation

$z<0.2$0

applied to central stellar velocity dispersions measured from the $z<0.2$1 Å region and the Ca triplet region at $z<0.2$2 Å (Koss et al., 2022).

The DR2 velocity-dispersion campaign reported new central stellar $z<0.2$3 measurements for 484 obscured AGN, with 956 independent determinations from 642 spectra, making it the largest $z<0.2$4 study of X-ray-selected obscured AGN to date. High-resolution observing was prioritized, with $z<0.2$5 and $z<0.2$6, enabling robust measurements down to $z<0.2$7. Typical uncertainties span $z<0.2$8, with a median of $z<0.2$9, and 281 AGN received a first published central velocity dispersion (Koss et al., 2022).

Bolometric luminosities and Eddington ratios in DR2 are tied to the BAT band. The standard BASS prescription uses the intrinsic $\sim 3200\text{--}10000$0 keV luminosity with a bolometric correction of 8, and

$\sim 3200\text{--}10000$1

The DR2 sample spans approximately $\sim 3200\text{--}10000$2 in $\sim 3200\text{--}10000$3, $\sim 3200\text{--}10000$4 in $\sim 3200\text{--}10000$5, and $\sim 3200\text{--}10000$6 in $\sim 3200\text{--}10000$7 (Koss et al., 2022), while the obscured-host $\sim 3200\text{--}10000$8 study maps $\sim 3200\text{--}10000$9 to $1\text{--}2.5\,\mu{\rm m}$0 and $1\text{--}2.5\,\mu{\rm m}$1 (Koss et al., 2022).

In the NIR, line fitting is performed with PySpecKit over segmented spectral regions centered on Paschen lines and key coronal transitions such as [Si VI], [Si X], [S VIII], and [S IX]. Broad and narrow components are separated at $1\text{--}2.5\,\mu{\rm m}$2, with centroids and widths tied where warranted by line physics and spectral quality (Brok et al., 2022). This NIR framework is especially important in obscured AGN because broad Paschen lines remain detectable when Balmer lines are strongly attenuated.

4. Obscuration, spectral classes, and accretion diagnostics

BASS DR2 established a dense optical classification framework for the BAT AGN population. Depending on the diagnostic, 48%–75% of BAT AGN are classified as Seyfert; the most efficient classification is the classic [O III]/H$1\text{--}2.5\,\mu{\rm m}$3 versus [N II]/H$1\text{--}2.5\,\mu{\rm m}$4 diagram, for which the Seyfert fraction is 75.4% (560/743) (Oh et al., 2022). At the same time, BAT selection reveals a sizable dusty narrow-line population missed by optical samples: $1\text{--}2.5\,\mu{\rm m}$5 occurs in $1\text{--}2.5\,\mu{\rm m}$6 of BAT narrow-line AGN (Oh et al., 2022).

The optical/X-ray correspondence is one of the core BASS results. DR1 reported broad agreement at the $1\text{--}2.5\,\mu{\rm m}$7 level between optical broad-line classification and X-ray obscuration, with Seyfert 1–1.8 generally below $1\text{--}2.5\,\mu{\rm m}$8 and Seyfert 2 generally above that threshold (Koss et al., 2017). DR2 sharpened the result into a clear dichotomy at $1\text{--}2.5\,\mu{\rm m}$9: 85% of Sy1–1.5 have $R>2500$0, 78% of Sy1.8–1.9 have $R>2500$1, and 91% of Sy1.9–2 have $R>2500$2 (Oh et al., 2022).

BASS also documents structured departures from the simplest one-absorber picture. In the Type 1-focused obscuration analysis, the BLR-facing extinction inferred from broad H$R>2500$3 is often orders of magnitude smaller than the X-ray $R>2500$4, implying that much of the X-ray-absorbing gas is located on scales smaller than, or internal to, the BLR. After removing $R>2500$5 of Sy1.9 classifications potentially contaminated by outflows, 86% of Type 1 AGN are X-ray unabsorbed and 14% are X-ray absorbed; about 70% of the absorbed Type 1 subset are Sy1.9 (Shimizu et al., 2017). The clustering analysis extends the challenge to orientation-only unification by showing that obscured AGN inhabit denser environments, or earlier-forming halos at fixed halo mass, than unobscured AGN even after matching in luminosity, redshift, stellar mass, and Eddington ratio (Powell et al., 2018).

Accretion-state diagnostics are another recurring BASS theme. The BASS III study showed that the narrow-line ratio [N II] $R>2500$6 anti-correlates significantly with $R>2500$7, with Pearson $R>2500$8, $R>2500$9, and $3200\text{--}10000$0 dex, and proposed it as an empirical Eddington-ratio indicator with $3200\text{--}10000$1 dex rms scatter in the inverse relation (Oh et al., 2016). By contrast, BASS VI found that the hard X-ray photon index $3200\text{--}10000$2 is only weakly correlated with $3200\text{--}10000$3 in the full hard-X-ray-selected sample, with a best-fit relation $3200\text{--}10000$4, and no robust evidence for such a correlation in the direct-mass subset (Trakhtenbrot et al., 2017). This clarified a methodological controversy: steeper $3200\text{--}10000$5 relations are recovered when using simplified X-ray modeling and optical-continuum-based accretion estimates, whereas broad-band X-ray modeling and X-ray-anchored $3200\text{--}10000$6 produce much flatter trends (Trakhtenbrot et al., 2017).

At the population-function level, BASS XXX derived the first directly observationally constrained black-hole mass function and Eddington-ratio distribution function for Type 2 AGN. After correcting for selection biases, the intrinsic ERDF of Type 2 AGN is significantly skewed toward lower Eddington ratios than that of Type 1 AGN, a result interpreted as support for radiation-regulated unification, in which radiation pressure shapes the geometry of the dusty obscuring structure (Ananna et al., 2022).

5. Host galaxies, environments, and dynamical context

BASS situates BAT AGN in a specific host-galaxy regime rather than treating them as purely nuclear sources. In the DR2 velocity-dispersion study, obscured BAT AGN have higher central stellar dispersions than optically selected SDSS narrow-line AGN, with typical $3200\text{--}10000$7 versus $3200\text{--}10000$8, but they are not biased toward the extreme dispersions of massive ellipticals with $3200\text{--}10000$9. The same analysis argues that BAT AGN preferentially occupy intermediate-to-massive spirals and lenticulars, and estimates that direct stellar or gas dynamical black-hole mass measurements are feasible for more than $0.3\text{--}2.5\,\mu{\rm m}$0 BASS AGN with existing facilities (Koss et al., 2022).

The environmental analysis in BASS IX places local hard-X-ray AGN in group-scale halos. For the full luminosity-limited sample, the inferred host-halo masses are $0.3\text{--}2.5\,\mu{\rm m}$1 in the median and $0.3\text{--}2.5\,\mu{\rm m}$2 in the mean. On average, BAT AGN occupy halos similarly to inactive galaxies of comparable stellar mass, but obscured AGN show enhanced small-scale clustering relative to unobscured AGN and therefore reside in denser environments or earlier-forming halos (Powell et al., 2018). This environmental asymmetry is one of the strongest BASS-based arguments that obscuration is not reducible to a line-of-sight orientation parameter alone.

Morphology work has extended the host picture to the 105-month BAT catalog. Visual classifications for 1189 hard-X-ray-selected AGN hosts show a deficiency of smooth ellipticals by $0.3\text{--}2.5\,\mu{\rm m}$3 and disks with prominent arms by $0.3\text{--}2.5\,\mu{\rm m}$4, together with an excess of mergers or disturbed systems by $0.3\text{--}2.5\,\mu{\rm m}$5 and disk galaxies without spiral structure by $0.3\text{--}2.5\,\mu{\rm m}$6, relative to a control sample matched in redshift and $0.3\text{--}2.5\,\mu{\rm m}$7-band magnitude. The same study finds a higher bar fraction among AGN hosts than in the control sample, at $0.3\text{--}2.5\,\mu{\rm m}$8 versus $0.3\text{--}2.5\,\mu{\rm m}$9, and reports that high-luminosity and high-accretion AGN preferentially reside in smooth or point-like hosts, while lower-luminosity AGN are more common in disk galaxies (Tello et al., 26 Jun 2025). A plausible implication is that BASS resolves multiple fueling channels within one selection function: disturbed or transitional disks, barred systems, and more spheroidal high-accretion hosts.

The luminous obscured tail of the BASS population has also been studied separately. Among 28 of the most luminous low-redshift narrow-line BAT AGN, hosts are predominantly massive galaxies with ${\rm FWHM}>1000\ {\rm km\,s^{-1}}$0, visually mostly ellipticals where imaging is available, and the radio properties span almost four orders of magnitude. In the NVSS-covered subset, 11 of 19 sources are radio-loud, and seven of the 28 show double radio lobes (Bär et al., 2019). This luminous obscured subset therefore connects BASS host work to jet production and feedback in the local universe.

6. Multiwavelength extensions and scientific legacy

A major strength of BASS is that it repeatedly tests optical diagnostics against less extinction-sensitive tracers. In the NIR DR2 study, 49/109 Seyfert 2 and 35/58 Seyfert 1 galaxies show at least one high-ionization coronal line. The [Si VI] ${\rm FWHM}>1000\ {\rm km\,s^{-1}}$1 line correlates with BAT X-ray emission with a scatter of 0.37 dex, substantially tighter than the optical [O III] ${\rm FWHM}>1000\ {\rm km\,s^{-1}}$2 relation at 0.71 dex, and no significant correlation is found between coronal-line luminosity and the X-ray photon index ${\rm FWHM}>1000\ {\rm km\,s^{-1}}$3. The same study shows that broad Pa${\rm FWHM}>1000\ {\rm km\,s^{-1}}$4 and Pa${\rm FWHM}>1000\ {\rm km\,s^{-1}}$5-based masses agree with the ${\rm FWHM}>1000\ {\rm km\,s^{-1}}$6 relation, whereas broad H${\rm FWHM}>1000\ {\rm km\,s^{-1}}$7-based masses in Sy1.9 can be underestimated by up to ${\rm FWHM}>1000\ {\rm km\,s^{-1}}$8 dex because of dust obscuration and BLR stratification (Brok et al., 2022).

The MIR extension pushes the same logic farther into obscuration-resistant emission. In 140 BASS AGN with Spitzer/IRS high-resolution spectra, high-ionization MIR lines are detected at very high rates: 88% for [Ne V] ${\rm FWHM}>1000\ {\rm km\,s^{-1}}$9, 85% for [Ne V] $A/N>3$0, and 96% for [O IV] $A/N>3$1. Their luminosities correlate with $A/N>3$2 keV X-ray luminosity with typical scatter $A/N>3$3 dex, and even more tightly with SED-based $A/N>3$4, with $A/N>3$5 dex. The line-to-continuum relations show no significant dependence on black-hole mass, Eddington ratio, or X-ray column density across the range probed (Bierschenk et al., 2024).

BASS has also become a platform for contemporaneous SED work. The optical–UV–X-ray study of 236 unobscured BAT AGN used simultaneous Swift/UVOT and Swift/XRT data to derive host-corrected SEDs, total bolometric luminosities, $A/N>3$6, and band-dependent bolometric corrections. It reports a significant decrease in $A/N>3$7 with $A/N>3$8 and $A/N>3$9, and provides a second-order regression between the $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$00 keV bolometric correction and $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$01, while explicitly recommending the use of scaling relations with their measured scatter rather than fixed median corrections (Gupta et al., 2024). This has made BASS relevant not only to AGN demographics but also to practical luminosity estimation in heterogeneous survey data.

Time-domain behavior is another established BASS extension. A multi-epoch analysis of 412 BAT AGN with repeated optical spectroscopy identified eight new changing-look events and brought the number of known BASS changing-look AGN to 21. Using contemporaneous $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$02 keV BAT light curves where available, the study found that many type transitions coincide with substantial ultra-hard X-ray flux changes and derived a changing-look rate of 0.7%–6.2% on $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$03-year timescales (Temple et al., 2022). Because the BAT band is comparatively insensitive to Compton-thin obscuration, these results argue that most BASS changing-look events are driven by intrinsic accretion variability rather than solely by line-of-sight absorption changes.

Recent BASS work has additionally positioned the local BAT sample as a benchmark for high-redshift obscured AGN. In a study of 21 highly luminous, obscured Sy1.9–2 AGN at $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$04, [Ne V] $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$05 is detected in 17 of 20 sources with optical spectra, and the sample is explicitly used as a comparison set for JWST-selected narrow-line AGN at $N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$06 (Peca et al., 14 Jul 2025). This suggests a continuing role for BASS as the low-$N_{\rm H} \approx 10^{24}\,{\rm cm^{-2}}$07 calibration standard against which future high-redshift obscured populations will be interpreted.

Taken together, BASS is not simply a catalog series but a coherent observational framework. Its distinctive contribution lies in combining ultra-hard X-ray selection, near-complete spectroscopy, and cross-band modeling to map AGN structure from the BLR to the host halo. That combination has enabled direct constraints on local X-ray luminosity functions, black-hole mass functions, Eddington-ratio distributions, host-galaxy morphology, environmental dependence, obscuration geometry, coronal-line physics, and time-domain state changes, all within a single, explicitly selection-aware AGN census (Ananna et al., 2022, Koss et al., 2022).