Galaxy Classification In 1924 Edwin Hubble divided galaxies

Galaxy Classification In 1924, Edwin Hubble divided galaxies into different “classes” based on their appearance. Why begin here? • Hubble classification serves as the basic language of the field. • The morphological sequence reflects a fundamental physical and evolutionary sequence, which offers important clues to galactic structure, formation and evolution.

Hubble Tuning Fork diagram (Hubble 1936) Ellipticals Lenticular (S 0) Spiral and Barred Spiral Irregular

Spiral Galaxies • Disk + spiral arms + bulge (usually) • Subtype a b c defined by 3 criteria: • Bulge/disk luminosity ratio • Sa: B/D>1 Sc: B/D<0. 2 • Spiral pitch angle • Sa: tightly wound arms Sc: loosely wound arms • Degree of resolution into knots, HII regions, etc.

Barred Spiral Galaxies • Contain a linear feature of nearly uniform brightness centered on nucleus • Subclasses follow those of spirals with subtypes a b and c

Elliptical Galaxies • Smooth structure and symmetric, elliptical contours • Subtype E 0 - E 7 defined by flattening • En where n = 10(a-b)/a where a and b are the projected major and minor axes (doesn’t tell what the 3 -D shape is)

Lenticulars or S 0 Galaxies • Smooth, central brightness concentration (bulge similar to E) surrounded by a large region of less steeply declining brightness (similar to a disk) • No spiral arm structure • Originally thought to be transition objects between Sa and E but typical S 0 is 1 -2 mags fainter than typical Sa, E (van den Bergh 1998)

Irregular Galaxies NGC 4485 -Irr II M 82 -Irr II Irr I • No morphological symmetry • Lots of young, blue stars and interstellar material • Smaller than most spirals and elliptical galaxies • Two major subtypes: • Irr I: spiral-like but without defined arms, show bright knots with O, B stars • Irr II: asymmetrical with dust lanes and gas filaments, often interacting

General trends within Hubble sequence E Sc: • • Decreasing Bulge/Disk Decreasing stellar age Increasing fractional gas content Increasing ongoing star formation Limitations of the Hubble Classification Scheme 1. Only includes massive galaxies (doesn’t include dwarf spheroidals, dwarf irregulars, blue compact dwarfs) 2. Three different parameters for classifying spirals is unsatisfactory because the parameters are not perfectly correlated. 3. Bars are not all-or-nothing. There is a continuum of bar strengths.

de Vaucouleurs’ Revised Hubble Classification System (de Vaucouleurs 1958, Handbuch der Phys. 53, 275) (de Vaucouleurs 2 1964, Reference Catalog of Bright Galaxies) Basic idea: retain Hubble system, but add lots of optional bells and whistles • Mixed types: E/S 0, Sab, Sbc • Mixed barred/normal: SA (unbarred), SB (barred), SAB (in between) • Inner rings: S(s) (arms out of ring), S(r) (arms in ring), S(rs) • Outer rings: (R) S • Extended spiral, irr types: Sm (between spiral and Irr), Im (magellanic), Sd (extreme Sc), Sdm (between Sd and Im) • “t-types” scale Added in later editions of the Reference Catalog (de Vaucouleurs 2, Corwin 1976) E 0 Sa Sb Sc Im -5 -1 1 3 5 10 (t-type)

Schematic Diagram of Revised Hubble Classification E E+ S 0 - S 0+ Sa Sb Sc Sd Sm Im Cross section of diagram No Bar Limitations: Ring shaped Spiral shaped • E Im is not a linear sequence of one parameter • Rings and bars are not independent • Does not take into consideration mass or other important parameters. All based on optical surface brightness morphology. Bar

Luminosity Classification or “DDO System” van den Bergh (1960) - working at David Dunlop Observatory in Ontario, Canada - hence the “DDO” In spirals and irregular galaxies, some properties correlate with galaxy mass rather than type. For spirals, the key parameter is arm development (i. e. arm length, continuity and width relative to size) Sc I - long, well-developed arms Sc III - short, stubby arms Sc IV - dwarf, spiral galaxy -faint hint of spiral structure Revised DDO - van den Bergh (1976): Placed disk galaxies into 3 parallel classes based on luminosity: Gas-rich, anemics and lenticulars Anemics have weak and diffuse spiral arms and low level of ongoing SF Parameters which change systematically from Lenticular to Gas-rich • Mean stellar age • Gas fraction • Recent SF

Yerkes System (Morgan 1958) Strong correlation noted between the nuclear light concentration (how big the bulge is) and its integrated spectrum. Type is based on this one parameter - integrated spectral type. • E, S 0 • S • Irr K-type spectrum F-K stars dominate A stars dominate Nomenclature: g. S 2 Spectral type (dominant stars) Hubble type flattening (i. e. bulge/disk) E - elliptical 10(a-b) a, af, f, fg, g, gk, k D - S 0 a S - spiral B - barred I - Irregular R - rotationally symmetric but no S or E structure

Kennicutt (1992) Galaxies shown in order of increasing Hubble type from top to bottom.

A couple of galaxy classes not addressed in these systems…. Dwarf Ellipticals – d. E • much less luminous • than the normal elliptical galaxy. Typically a few kpc across and contain 1 million stars. NGC 205 Dwarf Spheroidals – d. Sph • overall low star density • appear as a cluster of faint stars. The Sculptor system (Shapley 1938) was the first to be discovered. d. Sph are the low-luminosity counterparts of d. Es. Leo I d. Sph

Morphological Distributions The morphological type of galaxy present depends to some extent on where you look (more detailed discussion of this later…). Some key results: • The Local Group includes a significant number of very faint galaxies. Of the ~35 galaxies, only the 3 brightest (M 31, MW and M 33) are spirals, the remainder are equally divided between irregular and dwarf elliptical /spheroidal galaxies. • Galaxies outside of clusters (in the “field”) are biased towards late-type (Sc) spirals. A typical field sample might be 80% S galaxies, 10% S 0 galaxies, and 10% E galaxies. Within rich clusters, the distribution is dominated by early-type systems (Dressler 1980). An intermediate density cluster will have 40% S galaxies, 40% S 0 galaxies, and 20% E galaxies. A high density cluster will have 10% S, 50% S 0, and 40% E.

Automated Classification Visual classification is inherently time consuming and different observers are unlikely to agree in ambiguous cases. This motivates the development of algorithms to automatically and impartially classify galaxy images - very important for large surveys like 2 MASS and SDSS. Abraham et al. (1994, 1996): Concentration parameter C - fraction of light within ellipsoidal radius 0. 3 x outer isophotal radius (1. 5 above sky level). Asymmetry parameter A - fraction of light in features not symmetric wrt a 180 degree rotation Naim, Ratnatunga & Griffiths (1997) use 4 parameters: blobbiness, asymmetry, filling factor and elongation. Naim et al. (1995) used artificial neural nets to classify galaxies into the numerical T types. Achieved uncertainty of +/- 1. 8 in T which is comparable to the dispersion between observers. For distant galaxies (greater than z=0. 5), classification is difficult because of small angular size and apparent faintness of galaxies. HST galaxies (z~1) classified by 2 experts (Ellis and van den Bergh) and also using A and C parameters of Abraham. For faint galaxies, C parameter alone is fairly good. For brighter galaxies, C is degenerate between E and S 0.

Abraham et al. (1996)

The Gini Coefficient and M 20 parameter Used in economics to measure distribution of wealth in population G = relative distribution of flux in galaxy’s pixels (Abraham et al. 2003) G=0 for completely egalitarian society (uniform surf brightness) G=1 for absolute monarchy (all flux in single pixel) Constant 2 x area= G galaxy M 20 = 2 nd order moment of the brightest 20% of the galaxy measures concentration

Mergers more light in fewer pix E/S 0/Sa G Sb/Sbc more uniform surface brightness Lotz et al. 2005 Sc/Sd/Irr less concentrated M 20 centrally concentrated

gim 2 d bulge fraction B/T versus gim 2 d smoothness s 2. Judy Y. Cheng et al. MNRAS 2011; 412: 727 -747 Using surface brightness fitting to classify ~1000 galaxies from SDSS

. . . but we haven’t seen the end of visual classification! No matter how good the automated classifications become, the human eye is still better at determining patterns than neural networks (e. g. detecting spiral structure, smoothness) Galaxy Zoo is a “citizen science” project started in 2007, employing volunteers to classify galaxies imaged in the Sloan Digital Sky Survey, HST, and other galaxy survey projects. Well over 250, 000 people have participated in this project to visually classify about a million galaxies. Each galaxy receives over 20 classifications and the results are used together to determine the true classification. Some results? go to http: //www. galaxyzoo. org/