COS Legacy Archive Spectroscopic SurveY: 

A Treasury of Star-Forming Galaxies


PIs: Danielle Berg, John Chisholm, Tim Heckman, 

Bethan James, Crystal Martin, Dan Stark 


A Cycle 27 large HST program (135 new + 177 archival orbits) to study 45 nearby star-forming galaxies with full rest-frame far-UV spectral coverage of stellar, nebular, and ISM features


CLASSY is a treasury survey that builds on the Hubble Space Telescope (HST) archive to construct the first high-quality, high-resolution far-ultra violet (far-UV) spectral database of nearby star-forming galaxies. The survey will combine 177 orbits of archival observations with 135 new orbits (312 total orbits) of HST observations, or more than 600 total spectral images  taken with the Cosmic Origins Spectrograph (COS). 

Far-UV spectra are fundamental to our understanding of star-forming galaxies, providing a unique window on massive stellar populations, chemical evolution, feedback processes, and reionization. The launch of the James Webb Space Telescope (JWST) and construction of the extremely large telescopes (ELTs) will soon usher in a new era, pushing the UV spectroscopic frontier to z~15-20. The success of these future endeavors hinges on a comprehensive understanding of the massive star populations and interstellar medium (ISM) gas conditions that power the observed UV spectral features. This requires the level of detail that is only possible in local galaxies.  


The CLASSY Treasury sample contains 45 star-forming galaxies selected to span similar properties as seen at high redshift, with a large range of masses, metallicities, star formation rates, ionization parameters, and densities. The spectra will detect a suite of emission and absorption lines from massive stars and the interstellar medium (ISM). The CLASSY spectral atlas will be used to investigate the massive stellar populations that populate metal poor galaxies, the physical properties of large-scale outflows that regulate star formation, and the chemical abundance patterns of the gas and stars. CLASSY will improve the diagnostic power of the UV lines for future JWST/ELT surveys, providing a long-lasting legacy to the astronomical community for decades to come.

In order to understand the complex interplay between stars and gas in star forming galaxies, CLASSY will focus on 6 areas of important science: (1) the science of massive stars, (2) photoionization and chemical evolution modeling, (3) reionization physics, (4) Lyman-α physics, (5) feedback physics, and (6) Far-UV nebular diagnostics. As a result of these areas of study, CLASSY will provide enduring value and utility via state-of-the-art data products (listed below) that will be provided to the astronomical community via MAST and the HLSP (high-level science products) websites:

  1. CLASSY spectral atlas of individual and stacked galaxies

  2. Compiled ancillary data across the EM spectrum for all CLASSY galaxies

  3. CLASSY stellar continuum fits and stellar population properties

  4. Database of spectral emission and absorption features in CLASSY spectra

  5. Lyα emission profile fits 

  6. Database of individual, tailored photoionization models of CLASSY galaxies

  7. Derived nebular properties of CLASSY galaxies

  8. New UV diagnostic tools calibrated to existing optical spectra

  9. Database of chemical abundances and chemical evolution models

  10. Database of feedback properties

  11. Predictive tools of inferred escape fractions of ionizing photons

  12. Improved stellar models and prescriptions

  13. User-friendly CLASSY tutorials




Danielle Berg | PI

The Ohio State


John Chisholm | co-PI

University of California Santa Cruz

Tim Heckman | co-PI

Johns Hopkins


Bethan James | co-PI

Space Telescope

Science Institute

Crystal Martin | co-PI

University of California Santa Barbara

Dan Stark | co-PI

University of


Peter Senchyna

Jarle Brinchmann

Themiya Nanayakkara

Jane Rigby

Matthew Hayes

Masami Ouchi

Yuma Sugahara

Swara Ravindranath

Allison Strom

Nell Byler

Stephane Charlot

Claus Leitherer

Alaina Henry

Svea Hernandez

Dawn Erb

Jacopo Chevellard

Matthew Bayliss

Richard Pogge

Chuck Steidel

Alessandra Aloisi

Aida Wofford

Lisa Kewley

Rongmon Bordoloi

Anna Feltre

Ricardo Amorin

Evan Skillman

Claudia Scarlata

Anne Jaskot

Stephen Wilkins

Mario Llerena

University of Arizona

Universidade do Porto

Universiteit Leiden

NASA Goddard

Stockholm University

University of Tokoyo

University of Tokoyo

Space Telescope Science Institute

Carnegie Observatories

Australian National Univeristy

Institut d'Astrophysique de Paris

Space Telescope Science Institute

Space Telescope Science Institute

Space Telescope Science Institute

University of Wisconsin Milwaukee

Institut d'Astrophysique de Paris

University of Cincinnati

The Ohio State University

California Institute of Technology

Space Telescope Science Institute

Universidad Nacional Autonoma de Mexico

Australian National Univeristy

North Carolina State University

SISSA International School for Advanced Studies

Universidad de la Serena

University of Minnesota

University of Minnesota

Williams College

University of Sussex

Universidad de la Serena



In order to achieve the science objectives of CLASSY, we require S/N~7 for moderate resolutions (R > 4000; FWHM 100 km/s) to resolve blended components and kinematic details and detect faint stellar features. We inspected the Hubble Spectra Legacy Archive, which provides S/N measurements per native resolution element (binned by 6 pixels; FWHM ∼15 km s−1), for spectra meeting this criteria, finding 101 nearby (z < 0.3) galaxies with S/N>7 per 100 km/s resolution element in at least one of the required FUV gratings (G130M, G160M, or G185M). The following selection criteria were then implemented to ensure the efficient completion of high-quality comprehensive rest-FUV spectra for a large, diverse sample of star-forming galaxies:

  • Star-forming galaxies: targets with secondary classifications (or visually confirmed spectral features) of QSO or Seyfert in the HSLA were removed.

  • Grating efficiency: 34 galaxies have observations of sufficient quality in two of the proposed gratings, allowing their rest-FUV coverage to be completed with the addition of a single grating. G130M and G185M/G225M spectra will be obtained for subsamples of 17 and 17 galaxies, respectively. From the galaxies with only observations in G130M or G160M, we preferentially chose five additional galaxies with existing high-quality G160M observations because G130M observations are less time intensive (due to its high sensitivity).

  • Low redshift: z < 0.2 allows us to prioritize the FUV sensitivity of COS, which is significantly higher than the rest-NUV detectors (COS, STIS).

  • Compact: visually selected targets with compact morphologies from the existing SDSS and/or COS acquisition images (Figure 1) so that the COS 2.5" aperture (optimized for point sources) will capture most of the light from the galaxy. Additionally, when comparing existing GALEX FUV fluxes (PSF∼5") to archival COS continuum fluxes for the CLASSY sample, we find that 63% of the total FUV flux is recovered by the COS aperture on average. 

  • Bright: for an efficient observing strategy, we require GALEX FUV magnitudes < 21 mag/arcsec^2, or <~3 orbits per grating setting.

  • Parameter Space: The sample fulfills criteria 1-5 broadly and efficiently samples the SFR, sSFR, M⋆, metallicity, and ionization parameter properties of star-forming galaxies near and far (z < 7; see Figure 2). To better sample the higher nebular densities observed for z > 1 galaxies (e.g., Shirazi+14; Sanders+16; Kassinen+17), we exhausted the SDSS DR8 and existing literature in order to unearth additional targets with [S II] densities of ne > 400 cm^-3. Although these targets are extremely rare at z ∼ 0, we were able to compile a sample of 5 galaxies with no previous COS observations (+5 galaxies in the G185M sample) that meet our criteria. With 10 high-density targets, CLASSY will be the first sample to directly study the effects of electron density on rest-FUV galaxy properties.

Figure 1: COS acquisition images from previous observations for a sub-sample of CLASSY. The CLASSY sample is composed of star-forming galaxies that are compact in the UV such that most of their light enters the 2.5" COS aperture (red circle). In fact, for the median galaxy in CLASSY, 63% of the GALEX FUV flux is recovered in the COS aperture, as measured by their existing spectra.

Figure 2: CLASSY sample properties (detailed in Table 1) of the G185M sample (17 galaxies; teal points), G130M sample (17 galaxies; gold points), G130M+G160M sample (5 galaxies grey diamonds), and high electron density sample (ne > 400 cm^−3 (10 galaxies; purple crosses). The CLASSY sample spans a broad dynamic range of M⋆, SFR, metallicity, ionization parameter (e.g., O32), and electron density. For comparison, we plot the M⋆ vs. SFR relationships for 0.5 < z < 2.5 galaxies from Whitaker+14. Note, in particular, that we sample the highest densities (e.g., Sanders+16; Kassinen+17) and lowest M⋆and metallicities that are thought to be important in the epoch of reionization (e.g., Schaerer+10), and high ionization parameter, which predicts LyC leakage (e.g., Izotov+18).

Details of the CLASSY sample are given in the table below. We derive initial sample properties from SDSS spectra and photometry (when available) in order to provide consistent values. A key goal of the CLASSY project will be to update these values with more robust measures. All objects are UV bright, nearby galaxies covering a range of metallicity, mass, SFR, and gas density.  Columns 2-5 give the target name used in this work, R.A., Dec., and redshift. Columns 6-9 list the O3N2 strong-line oxygen abundances, median total stellar masses, average star formation rates, and average specific star formation rates from the MPA-JHU DR8 database. Note that the masses of targets that are more extended than the SDSS fiber will be underestimated, but are still in the low- to moderate-mass range. Columns 10-12 lists the [SII] electron density (where values of ne < 100 cm^-3 are consistent with the low-density limit), [OIII] electron temperature, and the E(B-V) reddening value. Note that J0036-3333, J1444+4237, J0405-3648, and J0127-0619 do no have SDSS observations, and consequently their properties are taken from literature sources.



HST/COS began observing for CLASSY began in December of 2019 and is expected to continue through spring 2021. As of May 2020, about half of the new CLASSY observations have been successfully acquired. Below we show some preliminary data examples.   

Figure 3: Example COS NUV acquisition images from recent CLASSY observations. By the sample design, these galaxies show compact ionizing stellar clusters such that most of their light enters the 2.5" COS aperture (gold circles). The spatial axes are also shown (green lines), where the trace has been plotted in the right panels. As a result, more extended objects will result in lower spectra resolutions. 

Figure 4: Interactive example CLASSY spectrum, combining archival and new FUV G130M+G160M+ G185M observations. Note that geocoronal lines have been removed and chip gaps are indicated by blank spaces, however, Milky Way absorption features are still present. By zooming in on different regions of the spectrum, you will see different stellar, ISM, and nebular features. Some of the common features are marked by colored bands and are labeled when you hover. You can click on the different types of spectral features in the legend to turn them on and off.

We extend our gratitude to the Lorentz Center for useful discussions during the “Characterizing Galaxies with Spectroscopy with a view for JWST” 2017 workshop that led to the formation of the CLASSY collaboration and survey.

© 2020 by Danielle A Berg. Proudly created with