# latexÓ¢ÎÄÅÅ°æ

Astronomy & Astrophysics manuscript no. DR9Q October 19, 2012

c ESO 2012

The Sloan Digital Sky Survey quasar catalog: ninth data release
? Aubourg3 , Stephen Bailey4 , Nicholas P. Ross4 , Adam D. Myers5,6 , Michael A. Isabelle P? aris1,2 , Patrick Petitjean1 , Eric 7 8 Strauss , Scott F. Anderson , Eduard Arnau9 , Julian Bautista3 , Dmitry Bizyaev10 , Adam S. Bolton11 , Jo Bovy? 12 , William N. Brandt13,14 , Howard Brewington10 , Joel R. Browstein11 , Nicolas Busca3 , Daniel Capellupo15,16 , William Carithers4 , Rupert A.C. Croft17 , Kyle Dawson11 , Timoth? ee Delubac18 , Garrett Ebelke10 , Daniel J. Eisenstein19 , Philip 20 21 13 , 14 , 22 Engelke , Xiaohui Fan , Nur Filiz Ak , Hayley Finley1 , Andreu Font-Ribera4,23 , Jian Ge15 , Robert R. Gibson8 , 24 15 ¡¦ Patrick B. Hall , Fred Hamann , Joseph F. Hennawi6 , Shirley Ho17 , David W. Hogg25 , Zeljko Ivezi? c8 , Linhua 21 8 , 26 27 4 6 , 28 Jiang , Amy E. Kimball , David Kirkby , Jessica A. Kirkpatrick , Khee-Gan Lee , Jean-Marc Le Go?18 , Britt Lundgren20 , Chelsea L. MacLeod9 , Elena Malanushenko10 , Viktor Malanushenko10 , Claudia Maraston29 , Ian D. McGreer21 , Richard G. McMahon30 , Jordi Miralda-Escud? e9,31 , Demitri Muna32 , Pasquier Noterdaeme1 , Daniel 10 18 10 Oravetz , Nathalie Palanque-Delabrouille , Kaike Pan , Isma¡§ el Perez-Fournon33,34 , Matthew M. Pieri29 , Gordon T. 35 1 36 Richards , Emmanuel Rollinde , Erin S. Sheldon , David J. Schlegel4 , Donald P. Schneider13,14 , Anze Slosar36 , Alaina Shelden10 , Yue Shen19 , Audrey Simmons10 , Stephanie Snedden10 , Nao Suzuki4,37 , Jeremy Tinker32 , Matteo Viel38,39 , Benjamin A. Weaver32 , David H. Weinberg40 , Martin White4 , W. Michael Wood-Vasey41 , and Christophe Y eche18
(A?liations can be found after the references) Received xxx; accepted xxx
ABSTRACT

arXiv:1210.5166v1 [astro-ph.CO] 18 Oct 2012

We present the Data Release 9 Quasar (DR9Q) catalog from the Baryon Oscillation Spectroscopic Survey (BOSS) of the Sloan Digital Sky Survey III. The catalog includes all BOSS objects that were targeted as quasar candidates during the survey, are spectrocopically con?rmed as quasars via visual inspection, have luminosities Mi [z=2] < ?20.5 (in a ¦«CDM cosmology with H0 = 70 km s?1 Mpc?1 , ?M = 0.3, and ?¦« = 0.7) and either display at least one emission line with full width at half maximum (FWHM) larger than 500 km s?1 or, if not, have interesting/complex absorption features. It includes as well, known quasars (mostly from SDSS-I and II) that were reobserved by BOSS. This catalog contains 87,822 quasars (78,086 are new discoveries) detected over 3,275 deg2 with robust identi?cation and redshift measured by a combination of principal component eigenspectra newly derived from a training set of 8,632 spectra from SDSS-DR7. The number of quasars with z > 2.15 (61,931) is ?2.8 times larger than the number of z > 2.15 quasars previously known. Redshifts and FWHMs are provided for the strongest emission lines (C iv, C iii], Mg ii). The catalog identi?es 7,533 broad absorption line quasars and gives their characteristics. For each object the catalog presents ?ve-band (u, g, r, i, z) CCD-based photometry with typical accuracy of 0.03 mag, and information on the morphology and selection method. The catalog also contains X-ray, ultraviolet, near-infrared, and radio emission properties of the quasars, when available, from other large-area surveys. The calibrated digital spectra cover the wavelength region 3,600-10,500 ? at a spectral resolution in the range 1,300 < R < 2,500; the spectra can be retrieved from the SDSS Catalog Archive Server. We also provide a supplemental list of an additional 949 quasars that have been identi?ed, among galaxy targets of the BOSS or among quasar targets after DR9 was frozen.
Key words. Keywords: catalogs, surveys, quasars: general

1. Introduction
Since their discovery (Schmidt 1963), interest in quasars has grown steadily, both because of their peculiar properties and because of their importance for cosmology and galaxy evolution. Many catalogs have gathered together increasing numbers of quasars either from heterogeneous samples (see Hewitt & Burbidge 1993; V? eron-Cetty & V? eron 2006, and references therein) or from large surveys, most importantly: the Large Bright Quasar Survey (LBQS, Morris et al. 1991; Hewett et al. 1995); the 2dF Quasar Redshift Survey (2QZ; Boyle et al. 2000; Croom et al. 2001) and the successive releases of the Sloan Digital Sky Survey (SDSS, York et al. 2000) Quasar Catalogs (e.g., Schneider et al. 2010, for DR7). This paper describes the ?rst quasar catalog of the Baryon Oscillation Spectroscopic Survey (BOSS, Schlegel et al. 2007;
?

Dawson et al. 2012). BOSS is the main dark time legacy survey of the third stage of the Sloan Digital Sky Survey (SDSS-III, Eisenstein et al. 2011). It is based on the ninth data release of the SDSS (Ahn et al. 2012). BOSS is a ?ve-year program to obtain spectra of 1.5 million of galaxies and over 150,000 z > 2.15 quasars. The main goal of the survey is to detect the characteristic scale imprinted by baryon acoustic oscillations (BAO) in the early universe from the spatial distribution of both luminous red galaxies at z ? 0.7 and H i absorption lines in the intergalactic medium (IGM) at z ? 2.5. BOSS uses the same imaging data as in SDSS-I and II, with an extension in the South Galactic Cap (SGC). The BAO clustering measurements in the IGM require a quasar catalog of maximal purity and accurate redshifts. Indeed the spectra of any non-quasar object, especially at high signal-tonoise ratio, will dilute the signal and/or increase the noise in the clustering measurement. The automated processing of the spec1

Hubble fellow

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

tra (Bolton et al. 2012) is sophisticated, but is not perfect. The identi?cation of the objects and their redshifts have therefore to be certi?ed before any analysis is performed. The present catalog, henceforth denoted DR9Q catalog, contains 87,822 quasars identi?ed among the objects targeted as quasar candidates over an area of 3,275 deg2 surveyed during the ?rst two years of BOSS operations. We also give a supplemental list of quasars identi?ed among galaxy targets. This catalog keeps the tradition of producing quasar catalogs (Schneider et al. 2002, 2003, 2005, 2007, 2010) from SDSS-I and II (York et al. 2000). The ?nal version of the SDSS-II quasar catalog (Schneider et al. 2010) based on the seventh SDSS data release (Abazajian et al. 2009) contains 105,783 objects mostly at z < 2 (see Shen et al. 2011, for their properties). Note that the DR9Q catalog does not contain all DR7 quasars but only those DR7 quasars that were reobserved during the two ?rst years of BOSS1 . High redshift (z > 2) quasar continua together with pixel masks, improved noise estimates, and other products designed to aid in the BAO-Lyman-¦Á clustering analysis will be released in Lee et al. (2012, in prep.). The selection of candidates and observations are summarized in Section 2. We describe the visual inspection of all targets in Section 3, present accurate redshifts for the quasars in Section 4 and describe the detection and measurement of broad absorption lines (BALs) in Section 5. The catalog is described in Section 6. We give a catalog summary in Section 7 and comment on the supplemental lists of quasars in Section 8. We conclude in Section 9. In the following we will use a ¦«CDM cosmology with H0 = 70 km s?1 Mpc?1 , ?M = 0.3, and ?¦« = 0.7 (Spergel et al. 2003). Most of the objects in the catalog show at least an emission line with FWHM > 500 km s?1 in their spectra. However, there are a few exceptions: a few objects have emission lines with smaller FWHM due to noise or dust obscuration (Type II quasars) others have very weak emission lines but are identi?ed as quasars because of the presence of the Lyman-¦Á forest (Diamond-Stanic et al. 2009). We will call a quasar an object with a luminosity Mi [z=2] < ?20.5 and either displaying at least one emission line with FWHM greater than 500 km s?1 or, if not, having interesting/complex absorption features. This de?nition is slightly di?erent from the one used in SDSS-DR7. The change in absolute magnitude is to include a few low-z objects in the catalog. Because BOSS is targeting z > 2.15 quasars, the median absolute luminosity is higher in BOSS than in SDSSDR7. All BOSS objects with z > 2 qualify for the SDSS-DR7 de?nition: FWHM > 1000 km s?1 and Mi [z=0] < ?22 (adopting the same cosmology and ¦Á¦Í = ?0.5). In the following, all magnitudes will be PSF magnitudes.

of speci?c ancillary science programs or as a consequence of imperfect high-redshift quasar selection. To detect the BAO signal, a surface density of 15 quasars with z ¡Ý 2.15 per square degree is required (McDonald & Eisenstein 2007). For comparison, SDSS-I/II targeted about ? 14, 000 z ¡Ý 2.15 quasars over the full survey, e.g. ?8,400 deg2 (Schneider et al. 2010), leading to a surface density of ?2 quasars per square degree in the redshift range of interest for BOSS. To reach the BAO quasar density requirement implies targeting to fainter magnitudes than SDSS-I/II. The BOSS limiting magnitude for target selection is r ¡Ü 21.85 or g ¡Ü 22 (Ross et al. 2012), while z ¡Ý 3 quasars were selected to be brighter than i ? 20.2 in SDSS-I/II (Richards et al. 2002).
2.1. Imaging data

BOSS uses the same imaging data as that of the original SDSSI/II survey, with an extension in the SGC. These data were gathered using a dedicated 2.5 m wide-?eld telescope (Gunn et al. 2006) to collect light for a camera with 30 2k¡Á2k CCDs (Gunn et al. 1998) over ?ve broad bands - ugriz (Fukugita et al. 1996); this camera has imaged 14,555 unique square degrees of the sky, including ?7,500 deg2 in the NGC and ?3,100 deg2 in the SGC (Aihara et al. 2011). The imaging data were taken on dark photometric nights of good seeing (Hogg et al. 2001). Objects were detected and their properties were measured (Lupton et al. 2001; Stoughton et al. 2002) and calibrated photometrically (Smith et al. 2002; Ivezi? c et al. 2004; Tucker et al. 2006; Padmanabhan et al. 2008), and astrometrically (Pier et al. 2003).
2.2. Target selection

2. Survey outline
In order to measure the BAO scale in the Lyman-¦Á forest at z ? 2.5, BOSS aims to obtain spectra of over 150,000 quasars in the redshift range 2.15 ¡Ü z ¡Ü 3.5, where at least part of the Lyman-¦Á forest lies in the BOSS spectral range. The measurement of clustering in the IGM is independent of the properties of background quasars. Therefore the quasar sample does not need to be uniform and a variety of selection methods are used to increase the surface density of high redshift quasars (Ross et al. 2012). Some quasars with z < 2 will be targeted in the course
All known z > 2.15 quasars in BOSS footprint are being reobserved to obtain spectra of uniformly high SNR in the Lyman-¦Á forest and to enable variability studies 2
1

The target selection of quasar candidates is crucial for the goals of the quasar BOSS survey. On average 40 ?bers per square degree are allocated by the survey to the quasar project. The surface density of z ¡Ý 2.15 quasars to the BOSS magnitude limit is approximately 28 per deg2 (see Palanque-Delabrouille et al. 2012). Thus, recovering these quasars from 40 targets per square degree in single-epoch SDSS imaging is challenging because photometric errors are signi?cant at this depth and because the quasar locus (in ugriz) crosses the stellar locus at z ? 2.7 (Fan 1999; Richards et al. 2002; Ross et al. 2012). All objects classi?ed as point-sources in the imaging data and brighter than either r = 21.85 or g = 22 (or both, magnitudes dereddened for Galactic extinction) are passed through the various quasar target selection algorithms. The quasar target selection for the ?rst two years of BOSS operation is fully described in Ross et al. (2012). We brie?y summarize here the key steps. The target selection algorithm is designed to maximize the number of quasars useful for the Lyman-¦Á forest analyses and reach the requirement of 15 deg?2 quasars with z ¡Ý 2.15. Several target selection methods are therefore combined and data in other wavelength bands are used when available. At the same time, in order to use the quasars themselves for statistical studies, such as the quasar luminosity function or clustering analyses (e.g. White et al. 2012), part of the sample must be uniformly selected. Thus, the BOSS quasar target selection is split in two parts: ¨C About half of the targets are selected as part of the so-called ¡°CORE¡± sample using a single uniform target selection algorithm. The likelihood method (Kirkpatrick et al. 2011) was adopted for the CORE selection during the ?rst year of ob-

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

servations. Starting with the second year of operation, it was replaced by the extreme deconvolution method (XDQSO; Bovy et al. 2011) which better takes photometric errors into account. ¨C Most of the remaining quasar candidates are selected as part of the so-called ¡°BONUS¡± sample through a combination of four methods: the Non-Parametric Bayesian Classi?cation and Kernel Density Estimator (KDE; Richards et al. 2004, 2009), the likelihood method (Kirkpatrick et al. 2011), a neural network (Y eche et al. 2010) and the XDQSO method (Bovy et al. 2011, 2012, objects for lower likelihood than in the CORE sample, over a slightly expanded redshift range, and incorporating data from UKIDSS; Lawrence et al. (2007); from GALEX; Martin et al. (2005); and, where available, from coadded imaging in overlapping SDSS runs). The outputs of all of these BONUS methods are combined using a neural network Point-sources that match FIRST (Becker et al. 1995) and that are not blue in u ? g (which would be characteristic of z < 2 quasars) are also always included. In addition, previously known z > 2.15 quasars (mostly from SDSS I/II) were also re-targeted for several reasons: (i) the BOSS wavelength range is more extended than in previous surveys; (ii) BOSS spectra have usually higher signal-to-noise ratio (SNR) than SDSS spectra (Ahn et al. 2012); (iii) the two epoch data will allow spectral variability studies. This sample is selected using the SDSS-DR7 quasar catalog (Schneider et al. 2010), the 2dF QSO Redshift Survey (2QZ; Croom et al. 2004), the 2dFSDSS LRG and QSO Survey (2SLAQ; Croom et al. 2009), the AAT-UKIDSS-SDSS (AUS) survey, and the MMT-BOSS pilot survey (Appendix C in Ross et al. 2012). Quasars observed at high spectral resolution by UVES-VLT and HIRESKeck were also included in the sample. Finally, BOSS includes targeting of a number of ancillary programs, some designed speci?cally to target quasars (e.g., the variability programs; Palanque-Delabrouille et al. 2011; MacLeod et al. 2012). The corresponding programs include: ? Reddened Quasars: Quasar candidates with high intrinsic reddening. ? No Quasar Left Behind: Bright variable quasars on Stripe 82. ? Variability-Selected Quasars: Variable quasars on Stripe82, focused on z > 2.15. ? K-band Limited Sample of Quasars: Quasars selected from SDSS and UKIDSS K photometry. ? High-Energy Blazars and Optical Counterpars of Gamma-Ray Sources: Fermi sources, plus blazar candidates from radio and X-ray. ? Remarkable X-ray Source Populations: XMM-Newton and Chandra sources with optical counterparts. ? BAL Quasar Variability Survey: Known BALs from SDSSI/II. ? Variable Quasar Absorption: Known Narrow-line absorption quasars from SDSS-I/II. ? Double-Lobed Radio Quasars: Point sources lying between pairs of FIRST radio sources. ? High-Redshift Quasars: Candidates at z > 3.5 in overlap between scanlines. ? High-Redshift Quasars from SDSS and UKIDSS: Candidates at z > 5.5 from SDSS and UKIDSS photometry. ? Previously Known Quasars with 1.8 < z < 2.15: Reobserved to constrain metal absorption in the Ly¦Á forest. ? Variable Quasars: selected from repeat observations in overlaps of SDSS imaging runs.

These programs are described in detail in the Appendix and Tables 6 and 7 of Dawson et al. (2012).
2.3. Spectroscopy

Because BOSS was designed to observe targets two magnitudes fainter than the original SDSS spectroscopic targets, substantial upgrades to the SDSS spectrographs were required and prepared during the ?rst year of SDSS-III (Smee et al. 2012). New CCDs were installed in both red and blue arms, with much higher quantum e?ciencies both at the reddest and bluest wavelengths. These are larger format CCDs with smaller pixels, that match the upgrade of the ?ber system from 640 ?bers with 3 arcsec optical diameter to 1,000 ?bers (500 per spectrograph) with 2 arcsec diameter. The larger number of ?bers alone improves survey e?ciency by 50%, and because BOSS observes point sources (quasar targets) and distant galaxies in the skydominated regime the smaller ?bers yield somewhat higher SNR spectra in typical APO seeing, though they place sti?er demands on guiding accuracy and di?erential refraction. The original di?raction gratings were replaced with higher throughput, volume-phase holographic (VPH) transmission gratings, and other optical elements were also replaced or recoated to improve throughput. The spectral resolution varies from ?1,300 at 3,600 ? to 2,500 at 10,000 ? The instrument is described in detail in Smee et al. (2012) and the BOSS survey is explained in Dawson et al. (2012). BOSS spectroscopic observations are taken in a series of at least three 15-minute exposures. Additional exposures are taken until the squared signal-to-noise ratio per pixel, (SNR)2 , reaches the survey-quality threshold for each CCD. These thresholds are (SNR)2 ¡Ý 22 at i-band magnitude 21 for the red camera and (SNR)2 ¡Ý 10 at g-band magnitude 22 for the blue camera (extinction corrected magnitudes). Recall that the pixels are co-added, linear in log ¦Ë with sampling from 0.82 to 2.39 ? over the wavelength range from 3,610 to 10,140 ?. The current spectroscopic reduction pipeline for BOSS spectra is described in Bolton et al. (2012). SDSS-III uses plates with 1000 spectra each, more than one plate can cover a tile (Dawson et al. 2012). 819 plates were observed between December 2009 and July 2011. Some have been observed multiple times. In total, 87,822 unique quasars have been spectroscopically con?rmed based on our visual inspection. Fig. 1 shows the observed area in the sky. The total area covered by the SDSS-DR9 is 3,275 deg2 . Fig. 2 displays the cumulative number of quasars as a function of the observation date.

+60

+30

300

240

180

120 ?30

60

0

?60

Fig. 1. The space distribution in equatorial coordinates of the SDSS-III DR9 data release quasars.

3

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

104 Quasars (cumulative number)

8

Lyman?¦Á

40
6

CIV CIII] MgII (z>2)

30

MgII (z<2) Continuum

4

SNR
20 10
Jul 1st 2010 Jan 1st 2011 Jul 1st 2011

2

0
Jan 1st 2010

0

18

20

22

Date

Fig. 2. Cumulative number of quasars as a function of observation date during the ?rst two years of the survey. Horizontal times are due to the yearly summer shutdown during monsoon rains (summer 2010 at MJD = 55400) and the monthly bright time when BOSS does not observe. As z > 2 quasars are usually identi?ed by the presence of strong Lyman-¦Á and C iv emission lines, we determine the SNR e?ectively achieved at the position of these lines. The median SNR per pixel at the position of various emission lines (Lyman¦Á, C iv, C iii] complex and Mg ii) and in the continuum are shown in Fig. 3. While the SNR per pixel in regions free of emission lines (black histogram) drops to be equal to ? 1 at r ? 22, the SNR at the top of the Lyman-¦Á (green histogram) and C iv (red histogram) emission lines stays above about 4, allowing the identi?cation of a fair fraction of these objects at this magnitude. In order to classify the object, each spectrum is ?t by the BOSS pipeline2 with a library of star templates, a PCA decomposition of galaxy spectra and a PCA decomposition of quasar spectra. Each class of templates is ?t over a range of redshifts: galaxies from z = ?0.01 to 1.00 quasars from z = 0.0033 to 7.00; and stars from z = ?0.004 to 0.004 (¡À1200 km/s). The combination of redshift and template with the overall best ?t (in terms of the lowest reduced chi-squared) is adopted as the pipeline classi?cation (CLASS) and redshift measurement (Z ¡À Z ERR). A warning bitmask (ZWARNING) is set to indicate poor wavelength coverage, negative star template ?ts, broken/dropped ?bers, ?bers assigned to mesure sky background, and ?ts which are within ?¦Ö2 /dof = 0.01 of the next best ?t (comparing only ?ts with a velocity di?erence of 1,000 km s?1 ). A ZWARNING equals to zero indicates a robust classi?cation with no pipeline-identi?ed problem (Aihara et al. 2011; Bolton et al. 2012). The classi?cations by the BOSS pipeline are not perfect however and visual inspection is required. Most misclassi?ed spectra have low SNR. At SNR per pixel ? 2, some objects are ?t equally well by a star and a quasar template. Even if the object is correctly identi?ed as a quasar, the redshift can be erroneous, because one line is misidenti?ed; the most common case is Mg ii¦Ë2800 is misidenti?ed as Lyman-¦Á. But this can be also because of a strong absorption feature (e.g. a damped Lyman-¦Á
The software used is called idlspec2d and is publicly available. The current version is v5 4 45. Details can be found at http://www.sdss3.org/dr9/software, Bolton et al. (2012) 4
2

r?PSF magnitude

Fig. 3. Median observed SNR per pixel at the top of the Lyman¦Á (green), C iv (red), C iii] complex (blue), Mg ii at z > 2 (orange) and Mg ii at z < 2 (yellow) emission lines and in emission-line free regions (black) versus r-PSF magnitude (corrected for Galactic extinction). Two redshift ranges are considered for Mg ii because the emission line is redshifted in regions of the spectra with very di?erent characteristics. At r ? 22, the median SNR per pixel at the top of the Lyman-¦Á and C iv emission lines is about 4; su?cient to identify most of the quasars. Outside of the emission-line regions, at the same magnitude, the SNR per pixel is about unity. system, DLA, or a BAL) spoils the pro?le of an emission line and the pipeline is unable to recover it.
2.4. Calibration warnings
2.4.1. Excess ?ux in the blue

The BOSS spectra often show excess light at the blue end (a similar problem was found in SDSS-DR7 spectra; P? aris et al., 2011). To quantify this problem we selected spectra where a damped Lyman-¦Á system (DLA) is observed with aborption redshift greater than 3.385 and with a column density N (H i) ¡Ý 1020.5 cm?2 . There are 402 such quasars in the sample. In these spectra, and because of the presence of the DLA, the ?ux is expected to be zero at ¦Ëobs ¡Ü 4100 ? (e.g. below the Lyman limit of all DLAs). When stacking the selected lines of sight (Fig. 4), we note instead that the ?ux increases for wavelengths below 4000 ?. The excess light at ¦Ëobs ? 3600 ? is 10% of the ?ux at ¦Ërest = 1280 ? where the spectra are normalized. This problem can a?ect the analysis of the Lyman-¦Á forest (see e.g. Font-Ribera et al. 2012) and is probably a consequence of imperfect sky subtraction (Dawson et al. 2012). It will be corrected in a future version of the pipeline.
2.4.2. Spectrophotometric calibration

To maximize the ?ux in the blue part of the quasar spectra, where the Lyman-¦Á forest lies, it was decided to o?set the position of the quasar target ?bers to compensate for atmospheric refraction and di?erent focus in the blue (Dawson et al. 2012).

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release
0.20

Normalized near 1280? (rest) flux

Normalized near 1450? (rest) flux

0.15

4
SDSS?DR7 spectra

HI?Ly?¦Â

0.10

HI?Ly?¦Á

SDSS?DR9 spectra

SiIV

0.05

2

CIV

CIII]

0.00

3600

3800

4000

4200

Wavelength (?)

0 1000 1500 2000 2500 3000

Fig. 4. Stack of DR9 BOSS spectra where a damped Lyman¦Á system is seen at an absorpstion redshift higher than 3.385 with a column density N (H i) ¡Ý 1020.5 cm?2 . The spectra are normalized to unity near 1280 ? in the quasar rest frame. Owing to the presence of the DLA, the ?ux is expected to be zero at observed wavelengths below ?4100 ? (e.g. below the Lyman limit of all DLAs). This is not the case in the very blue part of the spectrum (¦Ëobs ¡Ü 4000?) where the mean observed ?ux appears to increase (spuriously).

Restframe wavelength (?)

Fig. 5. Composite spectra of 6,459 quasars observed both by SDSS-DR7 and BOSS for (i) SDSS-DR7 spectra (red) and, (ii) SDSS-DR9 spectra (black). The slope of the two composite spectra should be similar (as any variability should be averaged out). This is not the case because of the di?erence in focus of the BOSS quasars and standard stars. Note that this ?ux miscalibration is di?erent from object to object. check performed afterwards by the scanners or by a user of the data. When a new version of the pipeline is made available, all the data are re-reduced. We then reinspect objects with uncertain identi?cations (QSO ?, QSO Z?, Star ?, see Section 3.2) or spectra that are not quali?ed (Bad) but we do not reinspect the objects with ?rm identi?cations. It can happen that the spectra of a few objects are of lesser quality with the new version of the pipeline. These objects are still in the catalog. Even the most thorough work of the kind described here cannot be absolutely ?awless. We encourage the reader to signal any mistake to the ?rst author of this paper in order to ensure highest quality of the information provided in the catalog.

These o?sets were not applied to the standard stars. The current pipeline ?ux calibration does not take these ?ber o?sets into account, therefore the spectrophotometry of the main QSO targets (e.g. not the ancillary targets) is biased toward bluer colors over the full wavelength range. Spectrophotometry of these objects will preferentially exhibit excess ?ux relative to the SDSS imaging data at ¦Ë < 4000 ? and a ?ux decrement at longer wavelengths. Because the ?ber o?sets are intended to account for atmospheric di?erential refraction, data will show larger o?sets in spectrophotometric ?uxes relative to imaging photometry for observations performed at higher airmass. Dawson et al. (2012) discuss in details the quality of the BOSS spectrophotometry and reports that stellar contaminants in the quasar sample (i.e. quasar candidates that are actually stars) have g ? r colors 0.038 magnitudes bluer than the photometry with an RMS dispersion of 0.158 magnitudes. This problem is illustrated in Fig. 5 where the median composite spectra of quasars observed by both SDSS-I/II and BOSS are plotted together. The resulting SDSS-DR7 spectrum is in red and BOSS spectrum in black. The BOSS composite spectrum is bluer than the same composite from SDSS-DR7 spectra. Note that this ?ux mis-calibration is di?erent from object to object so that Fig. 5 shows only the mean di?erence between DR7 and DR9 spectra.
2.4.3. Identi?ed quasars with bad spectra

3. Construction of the DR9Q catalog
In order to optimally measure the BAO clustering signal in the IGM, we must have as pure a catalog of quasars as possible. In this catalog, peculiar features such as broad absorption lines (BAL) or Damped Lyman-¦Á systems (DLA) that may dilute the signal, should be identi?ed. We therefore designed quality control of the data based on a visual inspection of the spectra of all BOSS objects that might be a quasar. During commissioning and the ?rst year of the survey this quality control was also very useful to report problems with the pipeline, which helped improve the overall quality of the data reductions. The catalog lists all the visually con?rmed quasars. About 10% of these quasars have been observed several times (Dawson et al. 2012), either because a particular plate has been reobserved (e.g. to increase the SNR for a particular scienti?c project), or because a particular region in the sky has been reobserved at di?erent epochs (e.g. Stripe 82), or, because plates overlap. Now, and throughout BOSS, overlapping plates are used as an opportunity to increase the SNR on a few objets (e.g. CORE objects). These repeat observations are often useful to con?rm the nature of objects with low SNR spectra. However we did not attempt to co-add these data mostly because they are often of quite di?erent SNR.
5

During the course of the ?rst two years of BOSS, di?erent versions of the SDSS spectroscopic pipeline were used after some systematic problems had been ?xed, thus improving the overall quality of the data. The visual inspection described below is performed on the ?y, within a few days after the data are obtained, quali?ed and reduced by the version of the pipeline that is available at the time the data are obtained. Once an object is positively identi?ed as a quasar, a galaxy, or a star from visual inspection, it keeps its identi?cation in our catalog unless an apparent mistake has been committed and is corrected in the course of some

MgII

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

3.1. A tool for the visual inspection

Immediately after the processing by the BOSS pipeline, the reduced data (spectra and pipeline classi?cation) are copied to the IN2P3 (Institut National de Physique Nucl? eaire et de Physique des Particules) computing center3 . A Java program gathers metainformation and saves it into an Oracle database. All spectra are matched to target objects, imaging and photometry information, and SDSS-II spectroscopy. They are processed by a Java program that computes basic statistics from the spectra and ?ts a power law continuum and individual emission lines to each spectrum. The spectra are then made available online through a collaborative web application, from which human scanners can ?ag objects and decide classi?cations. This tool and the visual inspection procedure described in the next Section evolved with time during commissioning and the ?rst six months of the survey. The whole procedure was repeated at the end of the ?rst two years to guarantee the homogeneity of the catalog.
3.2. Visual inspection procedure

The identi?cations provided by the BOSS pipeline are already very good. Nevertheless about 12% of all quasar targets have a non-zero ZWARNING ?ag, i.e. their redshift is not considered to be reliable by the pipeline. After visual inspection, 4% of all con?rmed quasars have a non-zero ZWARNING ?ag. Not surprisingly, the fraction of these objects increases with magnitude (see Fig. 6).

Normalized distribution of zWarning>0

0.08

0.06

0.04

0.02

0.00

redshift of the quasar. We manually con?rmed or modi?ed the identi?cation of the object and, when needed, corrected the redshift provided by the BOSS pipeline, i.e. when it was wrong (when e.g. an emission line is misidenti?ed or a bad feature was considered an emission line) or inaccurate (when emission lines are correctly identi?ed but not properly centered). Examples of misidenti?ed objects or inaccurate redshift estimates are displayed in Fig. 7. All the information on the objects is stored in a database which is updated in real time as new data arrives from the telescope. Modi?cations from the visual inspection are stored also in the database. For each plate, the objects classi?ed by the pipeline as star, QSO with z < 2, and QSO with z ¡Ý 2 are made available to the scanner in three di?erent lists. The cut in redshift corresponds to the Lyman-¦Á emission line entering the BOSS spectrum. It also corresponds to a strong gap in the BOSS quasar redshift distribution due to target selection (see Fig. 22). Most of the objects classi?ed as stars by the pipeline are indeed stars and most of the objects classi?ed as quasar with z < 2 are either quasars with z < 2 or stars (see below). The objects classi?ed as quasars at z ¡Ý 2 are ranked by decreasing SNR. This organizes the visual inspection and minimizes the risk of errors. Most of the quasars with z ¡Ý 2, the most valuable for the survey, are inspected by two di?erent individuals. Objects that cannot be ?rmly identi?ed by visual inspection are labeled in several categories. Some spectra cannot be recognized because either the SNR is too low, or the spectrum has been badly extracted; such objects are classi?ed as Bad. For others, the classi?cation is not considered to be robust, but there is some indication that they are stars (star ?) or quasars (QSO ?). For some objects both scanners were unable to give a ?rm identi?cation, such objects are labeled as ¡®?¡¯. Other objects are galaxies (Galaxy). Finally some objects are recognized as quasars but their redshifts are not certain (QSO Z?). The output of the visual classi?cation is provided as ?elds class person and z conf person in the specObjAll table of the SDSS Catalog Archive Server (CAS) or the specObjAll.?ts ?le from the Science Archive Server (SAS). The correspondence between the visual inspection classi?cation we describe in this paper (QSO, QSO BAL, QSO Z?, QSO ?, Star, Star ?, Galaxy, ?) and the values of z conf person and class person is given in Table 1. Each time a new version of the BOSS pipeline becomes available, the data are reprocessed and objects in the categories bad, ?, QSO ? and QSO Z? are inspected again. Examples of objects classi?ed as QSO Z? and QSO ? are displayed in Fig. 8. Only objects classi?ed as QSO or QSO BAL are listed in the o?cial DR9Q catalog. Objects classi?ed as QSO Z? are included in the supplemental list of quasars (see Section 8). Objects classi?ed as QSO ? are also given for information in a separate list. Of the 180,268 visually inspected targets corresponding to the DR9Q catalog, 87,822 were classi?ed as unique quasars, 81,307 as stars and 6,120 as galaxies. 1,362 objects are likely quasars (QSO ?), 112 are quasars with an uncertain redshift (QSO Z?) and 578 are likely stars (Star ?). 2,599 targets have bad spectra (Bad) while we were not able to identify 368 objects (?). Therefore 97.5% of the objects are successfully classi?ed. Only 27 true quasars were mis-identi?ed by the BOSS pipeline as Star, while 11,523 stars were classi?ed as QSO, most of them misidenti?ed, however, as low redshift quasars, and only 1,241 have ZWARNING = 0. Table 2 gives a summary of these numbers. Note that Palanque-Delabrouille et al. (2012) have obtained deeper MMT QSO data of some of the BOSS targets classi?ed

18

20

22

r?PSF magnitude

Fig. 6. Fraction of visually con?rmed quasars with a non-zero ZWARNING ?ag as a function of the r-PSF magnitude (after correcting for Galactic extinction). A positive ZWARNING means that the pipeline considers its redshift estimate to be unreliable. This fraction increases at faint magnitudes. We visually inspected all quasar candidates and objects from quasar ancillary programs (see Section 2.2) to (i) secure the identi?cation of the object and, (ii) reliably estimate the systemic
3

CC-IN2P3, http://cc.in2p3.fr

6

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

120 100
SDSS J154329.47+335908.7

30 20 10

SDSS J075713.63+540941.2

30 20 10 0

SDSS J011900.31?000156.7

f¦Ë (10?17 erg s?1 cm?2 ??1)

80 60 40 4000 6000 8000

0

4000

6000

8000

4000

6000

8000

SDSS J073223.59+433628.5

SDSS J235854.43+014953.6

4

SDSS J124112.26?012153.5

1.0

4

2

0.5

2

0

4000

6000

8000

0.0

4000

6000

8000

0

4000

6000

8000

Wavelength (?)
Fig. 7. First column: Examples of z > 2 quasars classi?ed as STAR by the BOSS pipeline. The overall shape of the spectrum is similar to the spectrum of F stars. Second column: Examples of stars identi?ed as QSO by the BOSS pipeline. Strong absorption lines or wiggles in the spectrum can mimic quasar features. Third colum: Examples of z > 2 quasars for which the BOSS pipeline provides an inaccurate redshift estimate that must be corrected during the visual inspection. The pipeline is confused by the strong absorption lines. The spectra were boxcar median smoothed over 5 pixels.
4 3
SDSS J154250.13+290041.1

1.5 1.0 0.5 0.0
SDSS J082704.94+295839.5

f¦Ë (10?17 erg s?1 cm?2 ??1)

2 1 0 4000 6000 8000

4000

6000

8000

z conf person / class person 0 1 3 4 30

0 Not inspected -

1 ? QSO ? -

2 Star ? QSO Z? -

3 Star QSO Galaxy QSO BAL

3
SDSS J121944.78+382749.1 SDSS J162648.74+125124.5

2 1 0

4 2 0

Table 1. The table gives the classi?cation from the visual inspection corresponding to the combination of class person (?rst column) and z conf person (?rst row) values provided in the headers of the SDSS-DR9 spectra available from the SDSS Catalog Archive Server.

4000

6000

8000

4000

6000

8000

Wavelength (?)

Fig. 8. Examples of QSO ? (top panels) and QSO Z? (lower panels). The spectra were boxcar median smoothed over 5 pixels.

as QSO ? and con?rmed that essentially all of these objects are true quasars. During the visual inspection, a redshift is determined that will be re?ned further by an automatic procedure (see Section 4). The redshift of identi?ed quasars provided by the visual inspection is obtained applying the following procedure:

¨C The ?rst guess for the redshift is given by the BOSS pipeline and is not modi?ed except if inaccurate or wrong. The redshift from the pipeline can be wrong in cases where an emission line is misidenti?ed. The presence of strong absorption at or near the emission, and especially a strong DLA, is also a source of error. Often the redshift is just inaccurate because either it misses the peak of the Mg ii emission line (and we consider that this line is the most robust indicator of the redshift) or it is de?ned by the maximum of the C iv emission line when we know that this line is often blueshifted compared to Mg ii (Gaskell 1982; McIntosh et al. 1999; Vanden Berk et al. 2001; Richards et al. 2002; Shen et al. 2008; Hewett & Wild 2010). ¨C If the Mg ii emission line is present in the spectrum, clearly detected, and not a?ected by sky subtraction, the visual in7

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

spection redshift is set by eye at the maximum of this line. The typical uncertainty is estimated to be ?z < 0.003. The redshift is re?ned further, as described below. ¨C In other cases and for z > 2.3 quasars, such that Mg ii is redshifted into the noisy part of the red spectrum where sky subtraction errors make it unreliable, the redshift is estimated using the positions of the red wing of the C iv emission line which is known to be often blueshifted compared to Mg ii and of the peak of the Lyman emission line. The precision is estimated to be ?z < 0.005. The visual redshift is not accurate to better than ?z ? 0.003. but can be used as a reliable guess for further automatic redshift determination (see Section 4). Fig. 9 displays the distribution of the velocity di?erence between the visual inspection redshift estimate and the redshift provided by the BOSS pipeline. At z ¡Ü 2 the pipeline estimate is usually good and does not require signi?cant adjustments. In the redshift range 2.0?2.3, about half of the redshifts are modi?ed because the Mg ii emission line is available and de?nes clearly the visual inspection redshift while the pipeline ?nds often a slightly lower redshift. At z 2.3, 10% of the redshifts are corrected. Only 1,116 quasars (?2%), regardless of ZWARNING ?ags, have a di?erence between the pipeline and visual redshifts larger than 0.1.

tion are ?agged as well whatever the identi?cation of the object is. These quality ?ags are pipeline-version dependent and are not meant to be released with the catalog. They are mainly useful for feedback to the pipeline team.
3.3. A note on Damped Lyman-¦Á systems

In the course of the visual inspection, we ?ag the spectra with strong H i absorption (DLAs) in the Lyman-¦Á forest. At this point we do not try to measure the column density or to determine the redshift of the DLA. Flagging these lines of sight can be useful to complement the search for DLAs by automatic procedures since this is a notoriously di?cult task. Fig. 10 shows the number of DLAs we ?ag along SDSS-DR7 lines of sight reobserved by BOSS, versus the N (H i) column density. It can be seen that we visually recover most of the DLAs (log N (H i) > 20.3) identi?ed in the SDSS-DR7 by Noterdaeme et al. (2009). Only 11 such DLAs are missed by the visual inspection out of 257. The detection and analysis of DLAs in BOSS spectra is beyond the scope of this paper and will be described in Noterdaeme et al. (2012).

40
0.45 < z < 1.1 1.1 < z < 1.6 1.6 < z < 2

Noterdaeme et al. (2009) Visual Inspection flag

Normalized distribution

Number of DLAs

0.4

2 < z < 2.25 2.25 < z < 2.5 2.5 < z < 3.5

30

20

0.2

10

0.0

?1000

0

1000

0

20.0

20.5

21.0

21.5

?vVI/Pipeline (km/s)

log NHI

Fig. 9. Normalized (to unit integral) distribution of the velocity di?erence between the pipeline and visual inspection redshift estimates for di?erent redshift bins. About half of the pipeline redshifts are corrected during the visual inspection. Most of the corrections are for quasars with 2 < z < 2.5 where the Mg ii emission line is available and where the pipeline redshift estimate does not correspond to the peak of the Mg ii emission line. In addition, peculiar spectral features are ?agged: ¨C When a damped Lyman-¦Á absorption line is present in the forest, the object is assigned a ?ag ¡°DLA¡±. This ?ag can be used to check automatic Damped Lyman-¦Á detections (see Noterdaeme et al. 2009, 2012). ¨C Broad absorption lines in C iv and/or Mg ii are also ?agged. At this point there is no estimate of the width of the lines and we stay conservative. This ?ag can be used to check automatic BAL detections (see Section 5.1). ¨C Problems such as the presence of arti?cial breaks in the spectrum, obviously wrong ?ux calibration, or bad sky subtrac8

Fig. 10. H i column density distribution for DLAs and subDLAs detected by Noterdaeme et al. (2009) in quasars observed both by SDSS-DR7 and BOSS (black histogram). The red histogram displays the same distribution but for DLAs ?agged after visual inspection of BOSS spectra. This shows that the visual inspection is robust for log N (H i) > 20.3, the standard de?nition of DLAs.

4. Automatic redshift estimate
The visual inspection provides a reliable and secure redshift estimate for each quasar. Nevertheless, it is somewhat subjective and the accuracy of such an estimate is limited and cannot be better than 500 km s?1 . In principle, it is possible to estimate the redshift of a quasar using a linear combination of principal components to ?t the spectrum: the well known systematic shifts between emission lines are intrinsically imprinted in the components and the method can take into account the variations from

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release Classi?cation QSO QSO with z > 2 QSO ? QSO Z? Galaxy Star Star ? Bad ? Total # pipeline 102,696 (69,975) 10,563 67,009 180,268 # pipeline with ZWARNING=0 86,855 (64,004) 6,812 49,475 143,142 # visual inspection 87,822 (65,185) 1,362 112 6,120 81,307 578 2,599 368 180,268

Table 2. Number of objects identi?ed as such by the pipeline with any ZWARNING value (second column) and with ZWARNING = 0 (third column), and after the visual inspection (fourth column).

quasar to quasar (see e.g. P? aris et al. 2011). This should be a reliable procedure providing: ¨C The reference sample used to derive the principal components is representative of the whole quasar population; ¨C The redshift of each quasar in the reference sample is reliable. We will derive PCA components in order to reproduce the quasar spectrum between 1,410 and 2,900 ?, in the quasar rest frame, so that most of the prominent emission lines are covered, especially C iv and Mg ii. This will yield an automatic estimate of the quasar redshifts. These components will be also used to ?t emission lines individually to estimate a redshift for each emission line from the peak of the ?t model. To derive these PCA components, we will use a reference sample of quasars for which the two main emission lines are well observed. The redshifts of the quasars in the reference sample have also to be chosen carefully. The technique to derive PCA components of quasar spectra has been described in detail in several papers (e.g. Francis et al. 1992; Yip et al. 2004; Suzuki et al. 2005). We refer the reader in particular to Section 2.3 of P? aris et al. (2011).
4.1. Selection of the reference sample

4.2. Computing principal components

We now need an accurate redshift for each quasar before we calculate the PCA eigenvectors. We ?rst describe the use of Hewett & Wild (2010) redshifts and then an improved approach using the peak of the Mg ii emission line in individual spectra. Using Hewett & Wild (2010) redshifts: We ?rst consider the redshifts provided by Hewett & Wild (2010; HW10). They have performed a systematic investigation of the relationship between di?erent redshift estimation schemes and have derived empirical relationships between redshifts based on di?erent emission lines. They generated a high-SNR quasar template covering the UV and optical bands to be used to calculate cross-correlation redshifts. They estimate and correct for the quasar luminosity-dependence of systematic shifts between quasar emission lines. They are thus able to reduce systematic e?ects dramatically, correcting redshifts for the mean systematic shifts between emission lines. Note however that this does not fully account for intrinsic quasar-to-quasar variation among the population. Using these redshifts for the sample of representative quasars de?ned in Section 4.1, we derive the PCA eigenvectors. We then use the set of principal components to ?t a linear combination of 4 principal components to the whole spectrum of z ¡Ý 2.2 SDSSDR7 quasar spectra and estimate their redshifts. This number of components has been chosen after several trials in order to be able to derive a robust redshift for the maximum of objects. Note that the samples used to compute the principal components and to which we apply the procedure are disjoint. The median of the distribution of the velocity di?erences between the redshift given by HW10 and our redshift estimate is less than 30 km s?1 . However, the rms of this distribution is about 1,200 km s?1 which is undesirably large and is presumably due to quasar to quasar variations in emission-line shifts. We can try to overcome this drawback by using a redshift that is more representative of the individual characteristics of the quasars in the reference sample. This is why we will derive a redshift from the observed Mg ii emission line in each quasar spectrum. Indeed this line has been recognized as a reliable indicator of the actual redshift of the quasar (Shen et al. 2007, HW10). Using Mg ii emission line redshifts: Using the set of PCA components previously described, we ?t the Mg ii emission line of each quasar in the same SDSS-DR7 reference sample. From this ?t, we de?ne the Mg ii redshift using the peak-?ux position of the emission line ?t. Using a com9

To compute a set of principal components from a sample as representative as possible of the whole quasar population, we selected quasar spectra in SDSS-DR7 meeting the following requirements: ¨C The rest frame wavelength range 1,410 - 2,900 ? is redshifted into the observed wavelength range 3,900 - 9,100 ? (i.e. 1.77 < z < 2.13). This observed wavelength range is chosen to avoid the ?ux-excess issue in the very blue portion of the spectra (Section 2.4.1 and P? aris et al. 2011) and bad sky line subtraction at the red end. ¨C The median squared SNR per pixel over the full wavelength range is higher than 5. ¨C The spectra do not display BAL troughs as listed in the Allen et al. (2011) catalog. In SDSS-DR7, 8,986 quasar spectra meet these requirements. They all were visually inspected to remove spectra with obvious reduction issues (missing pixels, continuum breaks or very bad ?ux calibration). We ?nally used 8,632 spectra. The low-SNR cut we use here maximizes the number of quasars used for the PCA decomposition and makes our sample as representative as possible of the BOSS quasars.

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

bination of principal components to ?t an emission line avoids the need to assume a line pro?le (e.g. Gaussian, Lorentzian or Voigt). To estimate the quality of each emission line ?t: ¨C We compute the amplitude of the emission line (expressed in units of the median error pixel of the spectrum in the window we use to ?t the line) from the maximum ?ux relative to a ?tted power-law continuum. ¨C We measure the FWHM of the emission line in km s?1 . The amplitude-to-FWHM ratio (expressed in s km?1 ) provides an estimate of the prominence of the emission line. In particular, a weak and broad emission line will display a very low value of the amplitude-to-FWHM ratio. To con?rm the quality of the Mg ii line measurement, we also ?t C iv emission lines using the same procedure. The C iv emission line is easier to ?t since it is stronger and the region of the spectrum where it is redshifted is cleaner. If C iv could not be ?t, we also considered the Mg ii ?t to be unreliable. We then used the 7,193 spectra with both C iv and Mg ii amplitude-to-FWHM ratios larger than 8 ¡Á 10?4 s km?1 to compute the new PCA components to be applied to the whole spectra. We use the set of principal components derived with the Mg ii redshifts in the following. Fig. 11 displays the mean spectrum together with the ?rst ?ve principal components.
0.3 0.2 0.1 ?0.02 ?0.04
Principal Component 1 Mean spectrum

Flux (arbitrary unit)

0.0 ?0.1 0.00 ?0.05 0.0
Principal Component 4 Principal Component 3 Principal Component 2

due to the presence of strong absorption lines and especially BAL troughs, we ?rst obtain a ?t with only two principal components. This number is chosen because it provides a reasonable estimate of the amplitude of emission lines. Using this ?rst guess, we remove pixels below 2¦Ò and above 3¦Ò of the continuum where ¦Ò is de?ned as the median ?ux error in an 11 pixel window. We are thus able to remove broad absorption lines and badly subtracted sky emission lines (especially at the very red end of the spectra). We then increase the number of principal components iteratively to three and four, removing narrow absorption lines, keeping the same detection thresholds. Then, taking the visual inspection redshift estimate as an initial guess, it is possible to determine a redshift for each quasar by ?tting a linear combination of four principal components to the spectrum, in which the redshift becomes a free parameter. We call this redshift the PCA redshift. In addition, and in the same way as described in the previous subsection, we used ?ve principal components to ?t the Mg ii emission line in BOSS spectra when possible and derived a redshift from the peak ?ux of the ?t model. Using PCA allows to recover the line without a priori assumptions about the line pro?le in a region of the spectrum a?ected by sky subtraction. In the following we will call this redshift the PCA Mg ii redshift estimate. We compare in Fig. 12 the distributions of the velocity di?erence between PCA and PCA Mg ii redshift estimates. The PCA was applied to all BOSS quasars with 1.57 < zvisual < 2.3 so that both C iv and Mg ii emission lines are in the observed redshift range and are not strongly a?ected by sky subtraction. We considered three PCA estimates, varying the rest frame wavelength range over which the PCA was applied : (i) 1,410 ? 2,850 ? (full range); (ii) 1,410 ? 2,500 ? (Mg ii is not included) and (iii) 1410 ? 1800 ? (only C iv is in the range). There are 18,271 objects. It can be seen in Fig. 12 that the distributions are very similar. The median and rms of the distributions are (?35.3, 642), (?52.2, 780) and (?30.3, 851) km s?1 respectively for the three wavelength ranges. The rms is dominated by low SNR spectra and slightly increases when the amount of information decreases. The similarity of the distributions clearly shows that the PCA redshift estimate is consistent with the Mg ii estimate even when Mg ii is not included in the ?t.
4.4. Comparison to HW10

?0.2 0.0
Principal Component 5

?0.2

1600 1800 2000 2200 2400 2600 2800

Rest frame wavelength (?)
Fig. 11. Mean spectrum and the ?rst ?ve principal components derived in Section 4.2. A linear combination of the ?rst four principal components is used to estimate the global redshift of the quasar, while ?ve components are used to ?t emission lines locally.

4.3. Redshift estimates for BOSS quasars

For each quasar in the DR9Q catalog, we use four principal components to ?t the overall spectrum after having subtracted the mean spectrum. Four components are enough to reproduce the overall shape of the spectrum and derive the redshift (Yip et al. 2004; P? aris et al. 2011). However, in order to avoid poor ?tting
10

In order to compare the HW10 redshift estimates to ours, we selected SDSS-DR7 quasars re-observed by BOSS in the redshift range 2.00 < z < 2.30. We also restricted the sub-sample to quasars for which we were able to ?t the Mg ii emission line reliably and required the amplitude-to-FWHM of this line be larger than 8 ¡Á 10?4 s km?1 . Even though the Mg ii emission line is still detectable up to z = 2.5, we restrict the redshift range to below z = 2.3 to avoid the red end of the spectra where sky lines can be badly subtracted. 746 quasar spectra remain for the comparison. Fig. 13 displays the distributions of velocity di?erences between the PCA Mg ii redshift estimate and our PCA global estimate (black histogram) or HW10 redshift (red histogram). Both distributions were normalized in the same manner and we also took into account the di?erence in the rest frame wavelength used by the di?erent authors. The median HW10 redshift estimate is shifted by +136.9 km s?1 (with positive velocity indicating redshift) compared to our median Mg ii redshift with an rms of 467 km s?1 . Both

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release
0.08

800

PCA (1410?2850 ?) PCA (1410?2500 ?) PCA (1410?1800 ?)

PCA / MgII HW10 / MgII Pipeline / MgII

Number of Quasars

0.06

600

Frequency

0.04

400

0.02

200

0 ?2000

?1000

0

1000

2000

0.00 ?2000

?1000

0

1000

2000

?vPCA/MgII (km/s)
Fig. 12. Distributions of the velocity di?erence between PCA redshift estimates derived using di?erent rest frame wavelength ranges and the PCA Mg ii redshift estimate. HW10 and Shen et al. (2011) ?nd that the median shift of the Mg ii emission line relative to the [OIII] doublet is smaller than 30 km s?1 . These discrepancies in the median velocity shift may not be very signi?cant as di?erent ?tting recipes for any of these lines (Mg ii, [OIII], C iv) can potentially cause systematic velocity di?erences of this order. The median shift of our global estimate compared to the Mg ii redshift is ?49.9 km s?1 , with an rms of 389 km s?1 . The rms of the distribution is smaller than previously because we restrict our comparison here to spectra with high SNR. It is more peaked and the number of outliers is lower. This is not surprising as we use the same components to ?t the overall spectrum. However this illustrates the intrinsic dispersion between the results from the two methods. The overall conclusion is that our PCA estimate is very close to the Mg ii emission line redshift. And we are con?dent that the application of the procedure using PCA components to quasars for which the Mg ii emission line is redshifted beyond the observed wavelength range, will give robust redshift estimates.
4.5. Emission line redshifts

?v (km/s)

Fig. 13. Normalized distributions of the velocity di?erence between our global PCA redshift estimate (black histogram), the pipeline redshift estimate (blue histogram) or Hewett & Wild (2010) redshifts (red histogram) with the redshift derived from a PCA ?t of the Mg ii emission line (see text).
Transition C iv C iii] Mg ii Window 1450?1700 1800?2000 2600?2850 Rest frame wavelength (?) 1549.061 1908.734 2798.778

Table 3. Window and rest frame wavelength used to ?t each emission line.

Following the procedure described above, it is possible to reproduce the shape of each emission line with a linear combination of principal components. This combination can therefore ?t the individual lines without any a priori assumption about the line pro?le. In the case of individual lines we have more ?exibility to use more components because the ?t is more stable over a smaller wavelength range. We will use ?ve PCA components and de?ne the position (redshift) of the line as the position of the maximum of this ?t. Table 3 displays the de?nition of each window used to ?t emission lines together with the vacuum rest frame wavelengths taken from the NIST database4 used to compute the redshift. For multiplets (e.g. C iv and Mg ii), the rest frame wavelength used is the average wavelength over the transitions in the multiplet weighted by the oscillator strengths. Together with the redshift estimate of each line, we also retrieve
4

information on the symmetry of the line. We compute the blue (red) HWHM (half width at half maximum) from the PCA ?t, bluewards (redwards) of its maximum. The total FWHM is the sum of the blue and the red HWHMs. The continuum is provided by the ?t of a power law over the rest frame wavelength windows 1450 ? 1500, 1700 ? 1850 and 1950 ? 2750 ?. In Fig. 14 we plot the velocity of C iv relative to Mg ii versus the absolute magnitude of the quasar. The more luminous the quasar, the more blueshifted is the C iv emission line. Errors in the ?t are less than 200 km ?1 . These measurements can be useful to understand the relative shifts between di?erent emission lines and discuss the structure of the broad line region (see Shen et al. 2007; Shang et al. 2007). The C iii]¦Ë1909 line is blended with Si iii]¦Ë1892 and to a lesser extent with Al iii¦Ë1857. We do not attempt to deblend these lines. This means that the redshift and red HWHM derived for this blend should correspond to C iii]¦Ë1909, but the blue HWHM is obviously a?ected by the blend.

Broad absorption troughs are ?agged as BAL during the visual inspection. This ?ag means that an absorption feature broader than a usual intervening absorption (those arising in galaxies lying along the line of sight to the quasar) is seen. These BALs may a?ect the Lyman-¦Á forest and should be removed from its analysis. We ?ag mostly C iv BALs but also Mg ii BALs. Since during the visual inspection we do not measure the width of the trough, there is no a priori limit on the strength of the absorption.
11

http://physics.nist.gov/PhysRefData/ASD/lines form.html

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release
1000
90%

0
50%

70%

30%

?1000
10%

?2000 ?28

?26

?24

where f (v) is the ?ux normalized to the continuum as a function of velocity displacement from the line center. C (v) is initially set to 0 and can take only two discrete values, 0 or 1. It is set to 1 whenever the quantity 1 ? f (v)/0.9 is continuously positive over an interval of at least 2,000 km/s. It is reset to zero whenever the quantity in brackets becomes negative. Therefore BI = 0 does not mean that no trough is present. It means that, if a trough is present, the absorption does not reach 0.9 times the estimated continuum over a continuous window of 2,000 km s?1 . We will also de?ne a detection index, DI, giving C a value 1 over the whole trough if the criterion of a continuous trough over 2000 km s?1 is ful?lled. This index has the advantage of measuring the strength over the whole trough. This index will be useful to apply cuts in the analyses of the Lyman-¦Á forest. Indeed these analyses need an estimate of the total strength of the trough in order to avoid lines of sight spoiled by a strong BAL. To study weaker troughs, Hall et al. (2002) introduced the AI measurement de?ned as
0

?vCIV/MgII (km/s)

Mi[z=2]

Fig. 14. Velocity di?erence between C iv and Mg ii emission line redshifts as a function of the absolute i magnitude of the quasar. The solid black line shows the median velocity shift in 0.2 mag bins. Blue and cyan histograms display the 10th , 30th , 70th and 90th percentiles. The mean shift between the two emission lines increases with the quasar luminosity.

AI = ?
25,000

1?

f ( v) C ( v) d v, 0 .9

(2)

We also implemented an automatic detection of C iv BALs. We describe in Section 5.1 the method used to detect BALs and estimate their properties automatically. We then test the robustness of the visual inspection in Section 5.2 and the results of the automatic detection in Section 5.3. In the following subsections, we will concentrate on C iv BALs with z > 1.57. The quasar redshift limit is chosen so that the Si iv emission line is included in the spectra. This ensures that C iv BALs can be measured across the full range of velocities in balnicity index, e.g. up to 25,000 km s?1 .
5.1. Method used to estimate BAL properties automatically

where f (v) is the normalized ?ux and C (v) has the same de?nition as for the DI except that the threshold to set C to 1 is reduced to 450 km s?1 . The AI index was introduced in order to take into account weaker troughs and to measure troughs that are located close to the quasar rest velocity. It is however more sensitive to the continuum placement than the BI. Note that Trump et al. (2006) used a modi?ed version of the AI wherein the factor of 0.9 was removed from the integral to make the AI an equivalent width measured in km s?1 , where 1,000 km s?1 was the threshold instead of 450 km s?1 , and where the integral extended to 29,000 km s?1 . In this work we use the original Hall et al. (2002) de?nition of the AI. Following Trump et al. (2006), we calculate the reduced ¦Ö2 for each trough: ¦Ö2 trough = 1 1 ? f ( v) N ¦Ò
2

,

(3)

In order to detect BALs and to characterize the strength of the troughs using an objective procedure, we compute the balnicity (BI, Weymann et al. 1991) and the absorption indices (AI, Hall et al. 2002) of the C iv troughs. In addition, we introduce a new index, the detection index, DI, which is a slight modi?cation of BI. In Section 5.3, we will measure these indices for all quasars regardless of visual inspection. The continuum has to be estimated ?rst. For this, we use the same linear combination of four principal components described in Section 4. The resulting continuum covers the region from the Si iv to the Mg ii emission lines (see examples in Fig. 15). As described in Section 4.5, the procedure iteratively avoids absorption features and especially the BALs. During the automatic procedure, we smoothed the data with a ?ve pixel boxcar median. With this continuum, we compute the balnicity index (BI) in the blue of the C iv emission line using the de?nition introduced by Weymann et al. (1991):
3,000

where N is the number of pixels in the trough, f (v) is the normalized ?ux and ¦Ò the rms of the pixel noise. The greater the value of ¦Ö2 trough , the more likely the trough is not due to noise. We apply the automatic detection to all quasars in the DR9Q catalog and provide values of DI, AI and BI. We estimate also an error on the indexes. The error squared is obtained by applying the same formula as for the indexes replacing (1 ? f /0.9) by (¦Ò/0.9)2 with ¦Ò the rms of the noise in each pixel. Note however that the error on the strength of the trough is most of the time dominated by the placement of the continuum. To estimate the latter we have displaced the ?tted continuum by 5% and applied Eq.(2) of Kaspi et al. (2002).
5.2. Robustness of the visual detection of BALs

BI = ?
25,000

1?

f ( v) C ( v) d v, 0 .9

(1)

During the visual inspection, we are conservative and ?ag a BAL only if the trough is apparent. In addition, the automatic detections rely on the position of the continuum while the visual inspection lacks this problem. This means that the BAL sample from the visual inspection is purer than those from automatic detection. It is however unavoidable that, as the strength of the absorption or the spectrum SNR decreases, the visual inspection will start to be subjective. On the other hand the fraction of

12

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

6

f¦Ë (10?17 erg s?1 cm?2 ??1)

4 2

SDSS J170945+230907.5 BI = 566 km/s SNR = 4.8

20 15 10 5

SDSS J231306.8+050042.8 BI = 671 km/s SNR = 10

20 10 0

SDSS J013812.07?003002.5 BI = 991 km/s SNR = 22.5

0

4000 4500 5000 5500 6000

0

4000

5000

6000

7000

4000

5000

6000

7000

60 4
SDSS J004906.79?004621.3 BI = 3191 km/s SNR = 4.9

6 4

SDSS J085549.82+035735.2 BI = 4017 km/s SNR = 10.1

40 20 0

SDSS J221558.15?005521.7 BI = 2301 km/s SNR = 25

2 2 0 0

4000

5000

6000

7000

4000 5000 6000 7000 8000

4000

5000

6000

7000

Wavelength (?)
Fig. 15. Examples of high-redshift BAL quasar spectra in di?erent ranges of signal-to-noise ratio and balnicity indices. The ?t of the continuum is overplotted. false BALs detected by the automatic procedure will be higher. Fig. 16 shows the ratio of the number of visual BALs to that of the automatic detections as a function of SNR per pixel at ¦Ërest = 1700 ? for BI > 500 km s?1 .
1.0

0.5

0.0

0

10

20

Signal?to?noise ratio
Fig. 16. Ratio of the numbers of BAL visual and automatic detections as a function of spectral SNR per pixel at ¦Ërest = 1,700 ? for BI > 500 km s?1 . As expected, this ratio decreases with decreasing SNR. Out of the whole DR9Q catalog, 7,533 quasars have been ?agged visually as BAL. Out of the 69,674 quasars with z > 1.57, 7,228 are ?agged as BAL by visual inspection. If we re-

strict the latter sample to quasars with SNR>10 at 1,700 ? in the rest frame we have ?agged 1,408 BALs out of 7,317 quasars, a fraction of 19.2% which compares well with what was found by Gibson et al. (2009). Trump et al. (2006) measured BAL troughs (BI and AI) in the SDSS-DR3 release. We compare their detections and BI measurements with ours for quasars in common between BOSS and SDSS-DR3. Out of the 477 BALs (BI > 0) that are detected by Trump et al. (2006), we ?ag 425. We checked the BOSS spectra of these quasars individually. About half of them are not BALs and a handful, all with BI < 500 km s?1 , are real BALs that were missed by the visual inspection. For the rest, it is hard to decide if they are real or not because of poor SNR. Note that, in general, BOSS spectra are of higher SNR than previous SDSS spectra. There are an additional 296 quasars in Trump et al. (2006) that have C iv troughs that we do not ?ag as BALs. These all have AI > 0 but BI = 0. The histogram of AI measurements from Trump et al. (2006) for these objects is plotted as well in Fig. 17 (black histogram). Most of the missing troughs have AI smaller than 1,000 km s?1 . A visual inspection of the BOSS spectra reveals that most of the AIs have been overestimated and about half are not real mainly because the continuum in the red side of the C iv emission line has been overestimated. Allen et al. (2011) searched for BALs in SDSS-DR6; they measured only BI. Out of the 7,223 quasars with z > 1.57 in common with BOSS, they ?nd 722 quasars with BI > 0. Of these 7,223 objects, we ?ag 1,259 as BALs of which 853 have BI>0. We checked the 131 objects for which we measured BI > 0 but Allen et al. (2011) found BI = 0 individually. Some of the additional BALs are identi?ed because of better SNR in BOSS, some were missed by Allen et al. (2011) because of di?culties in ?tting the emission line correctly, a handful are explained by the disappearance of the BAL between the two
13

nVI/nAuto

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

AI ? Trump et al. (2006)

Number of objects

40

BI ? Allen et al. (2011)

20

0

0

500

1000

1500

2000

Trough strength on CIV (km/s)

Normalized distribution

Fig. 17. Distribution of BI and AI for quasars detected as BAL by automatic procedures in previous SDSS releases, and that were not ?agged by the visual inspection of the BOSS spectra. The black histogram shows the distribution of AI as measured by Trump et al. (2006) for 296 such quasars (all have BI = 0). About half of them are not real BALs (see text). From the automatic detection by Allen et al. (2011) (blue histogram), 57 quasars were missed by the visual inspection. Here again, only a handful of these objects actually display BAL troughs.

We conclude from this comparison that our catalog of BAL quasars ?agged by visual inspection is pure at the 95% level, but is probably incomplete below BI?500 km s?1 . This results from the conservative approach we adopted when ?agging the troughs implying that the number of detections in the visual inspection is decreasing with decreasing SNR. It is di?cult to estimate the incompleteness especially at low SNR because none of the previously published samples is reliable at small BI values. Therefore we caution the reader against blind uses of the catalog. SNR at rest wavelength 1700 ? (SNR 1700) is provided in the catalog. This can be used to identify the more reliable spectra. The BI distributions normalized by the total number of quasars with BI>500 km s?1 in each sample for visually ?agged BALs in DR9 (this work) and DR6 (Allen et al. 2011) are compared in Fig. 19. We ?nd 3,130 BALs with BI>500 km s?1 out of 69,674 BOSS quasars with z > 1.57 (4.5 %). If we restrict ourselves to quasars with SNR > 10, these numbers are 813 BALs out of 7,317 quasars, corresponding to a rate of 11.1%. This compares well with the ?10% uncorrected observed fraction of BAL found by Allen et al. (2011) at z ? 2.5.

0.10 SDSS?DR9 (this work, visual inspection) SDSS?DR6 (Allen et al. 2011)

epochs (Filiz Ak et al. 2012) and also by some appearances (see Fig. 18). We also ?nd 57 objects that are detected by Allen et al. (2011) and are missing in our visual detection. The BI distribution of these objects is shown as the blue histogram in Fig. 17. About half of them are not BALs upon re-inspection and a handful are real BALs missed by the visual inspection. The nature of the rest of the objects is unclear.

0.05

0.00
Normalized Flux
1.0

1

2

3

4

SDSS?DR7
0.004

log BICIV (km/s)
Fig. 19. Normalized distributions of the logarithm of balnicity indices measured from C iv troughs. The BI distribution from the present catalog (black histogram) computed from 7,227 visually ?agged BAL quasars is very similar to the distribution from Trump et al. (2006, red histogram) obtained from 1,102 BAL quasars from the SDSS-DR3 quasar catalog (Schneider et al. 2005). The distribution is also very similar to the BI distribution from Allen et al. (2011, blue histogram) based on the SDSS-DR6 quasar catalog.

SDSS?DR9

0.5

Flux (arbitrary unit)

0.0 ?20 ?10 0

Velocity (103 km/s)

0.002

SDSS J115122.14+020426.3
0.000 4000

5000

6000

7000

Wavelength (?)

5.3. Automatic detection

Fig. 18. Example of appearing BAL troughs. This quasar has been observed in SDSS-DR7 (red curve) and in SDSS-DR9 (black curve). The two spectra have been scaled to have a surface unity between 5,600 and 6,200?. The normalized ?ux in the C iv region expressed in velocity is displayed in the inset. This quasar had was not detected as BAL in SDSS-DR7 (i.e. BI = 0) while it has BI = 1, 826 km s?1 in the SDSS-DR9 spectrum.
14

We also performed an automatic detection of the C iv troughs using the continua in the wavelength range between the Si iv to C iv emission lines computed as described in Section 4. BIs, AIs and DIs are calculated for all quasars with z > 1.57 using Eq. 1 and Eq. 2. The values are given in the catalog together with the number of troughs both with width > 2000 and 450 km s?1 . We also give, for quasars with BI > 0, the minimum and maximum

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

velocities relative to zem , vmin and vmax , spanned by the whole absorption ?ow. Out of the 69,674 (resp. 7,317) quasars with z > 1.57 (resp. and SNR 1700>10), 8,124 (resp. 3,499) BALs, with ¦Ö2 trough > 10, have AI > 0 km s?1 . A visual inspection of spectra with small values of AI indicates that a number are due to inadequate continuum ?tting. We advise to be careful with AI values smaller than 300 km s?1 (see also below). Out of the 69,674 (resp. 7,317) quasars with z > 1.57 (resp. and SNR 1700>10), 4855 (resp. 1,196) BALs have BI > 0 km s?1 . This corresponds to 7% (resp. 16.3%). 821 BALs (11.2%) have BI>500 km s?1 . While the overall detection rate is larger than for the visual inspection, it is important to note that the automatic detection ?nds only 8 more objects with BI> 500 km s?1 in spectra with SNR>10 than the visual inspection. Upon reinspection, we found that half of them are not real and are due either to a peculiar continuum or to the presence of strong metal lines from a DLA at zabs ? zem . Three are real, but shallow BALs. This shows that the automatic and visual detections give nearly identical results for BI>500 km s?1 . At lower BI and lower SNR, and consistently with what was found by comparison with previous surveys, the number of unreliable detections is large.

We have shown here that BI measurements provided in the catalog are robust for SNR > 5 (see Fig. 16) and BI > 500 km s?1 . Any statistical analysis should be restricted to the corresponding sample. The catalog gives a few properties of detected C iv troughs and of Si iv and Al iii troughs but only in cases where BI(C iv) > 500 km s?1 and SNR>5. These troughs have been measured by Gibson et al. (2009) in SDSS-DR5.

6. Description of the DR9Q catalog
The DR9Q catalog is available both as a standard ASCII ?le and a binary FITS table ?le at the SDSS public website http://www.sdss3.org/dr9/algorithms/qso catalog.php. The ?les contain the same number of columns, the FITS headers contain all of the required documentation (format, name, unit of each column). The following description applies to the standard ASCII ?le. Table 4 provides a summary of the information contained in each of the columns in the ASCII catalog. The supplemental list of quasars (see Section 8) together with the list of objects classi?ed as QSO ? are also available at the same SDSS public website. Notes on the catalog columns: 1. The DR9 object designation, given by the format SDSS Jhhmmss.ss+ddmmss.s; only the ?nal 18 characters are listed in the catalog (i.e., the ¡°SDSS J¡± for each entry is dropped). The coordinates in the object name follow IAU convention and are truncated, not rounded. 2-3. The J2000 coordinates (Right Ascension and Declination) in decimal degrees. The astrometry is from DR9 (see Ahn et al. 2012). 4. The 64-bit integer that uniquely describes the spectroscopic observation that is listed in the catalog (Thing ID). 5-7. Information about the spectroscopic observation (Spectroscopic plate number, Modi?ed Julian Date, and spectroscopic ?ber number) used to determine the characteristics of the spectrum. These three numbers are unique for each spectrum, and can be used to retrieve the digital spectra from the public SDSS database. 8. Redshift from the visual inspection (see Section 3.2). 9. Redshift from the BOSS pipeline (see Section 2 and Bolton et al. 2012). 10. Error on the BOSS pipeline redshift estimate. 11. ZWARNING ?ag from the pipeline. ZWARNING > 0 indicates bad ?ts in the redshift-?tting code. 12. Automatic redshift estimate from the ?t of the quasar continuum over the rest frame wavelength range 1,410?2,000 ? with a linear combination of four principal components (see Section 4). When the velocity di?erence between automatic PCA and visual inspection redshift estimates is larger than 3000 km s?1 , this PCA redshift is set to ?1. The inaccuracy in the PCA estimate is often due to di?culties in the ?t of the continuum. In that case no automatic measurements are made on these objects and BI is set to ?1. 13. Error on the automatic PCA redshift estimate. If the PCA redshift is set to ?1, the associated error is also set to ?1. 14. Estimator of the PCA continuum quality (between 0 and 1). See Eq.(11) of P? aris et al. (2011). 15-17. Redshifts measured from C iv, C iii] complex and Mg ii emission lines from a linear combination of ?ve principal components (see Section 4). 18. Morphological information. If the SDSS photometric pipeline classi?ed the image of the quasar as a point source, the catalog entry is 0; if the quasar is extended, the catalog entry is 1.
15

1.0

BI (Isa)

10+4

0.5

0.0 0.0

0.5

10+4

1.0

BI (Allen)

Fig. 20. Balnicity index (BI) from this work against BI measured by Allen et al. (2011) for SDSS-DR6 objects re-observed by BOSS. We compare in Fig. 20 the BI values measured by Allen et al. (2011) for SDSS-DR6 spectra with BI values measured by our automatic procedure using BOSS spectra for the same quasars. Although the scatter is large, the median di?erence is only ?30 km s?1 . Note that part of the scatter is probably due to BAL variability (see Gibson et al. 2008, 2010; Filiz Ak et al. 2012). In Fig. 21 left (resp. right) panel, we compare the frequency distribution of AI (resp. BI) values in our BAL sample detected automatically with that of previous studies. The distributions are normalized in the same manner. It can be seen that the shape of the distributions are very similar. They peak around AI = 300 km s?1 which is the lower limit we set for robust detection.

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

Normalized distribution

SDSS?DR3 (Trump et al. 2006)

Normalized distribution

0.2

SDSS?DR9 (this work)

0.06

SDSS?DR9 (this work) SDSS?DR6 (Allen et al. 2011)

0.04

0.1

0.02

0.0

1

2

3

4

0.00

1

2

3

4

log AICIV (km/s)

log BICIV (km/s)

Fig. 21. Left panel: Distribution of absorption indices (AI) from our automatic detection (black histogram) and from SDSS-DR3 (red histogram, Trump et al. 2006). The distributions are normalized for log AI > 3. The di?erence between the two results at low AI is a consequence of slightly di?erent formula used to measure AI. Right panel: Distribution of balnicity indices (BI) from our automatic detection (black histogram) and from SDSS-DR6 (blue histogram, Allen et al. 2011). 19-21. Quasars targeted by BOSS are tracked with the BOSS TARGET1 ?ag bits (19; see details of selection method in Ross et al. 2012). In addition, 5% of ?bers on each plate are dedicated to ancillary programs tracked with the ANCILLARY TARGET1 (20) and ANCILLARY TARGET2 (21) ?ag bits. The bit values and the corresponding program names are listed in Dawson et al. (2012). 22. A quasar known from SDSS-DR7 has an entry equal to 1, and 0 otherwise 23-25. Spectroscopic plate number, Modi?ed Julian Date, and spectroscopic ?ber number in SDSS-DR7. 26. Uniform ?ag. See Section 7.4. 27. The absolute magnitude in the i band at z = 2 calculated after correction for Galactic extinction and assuming H0 = 70 km s?1 Mpc?1 , ?M = 0.3, ?¦« = 0.7, and a power-law (frequency) continuum index of ?0.5. The K-correction is computed using Table 4 from Richards et al. (2006). 28. The ?(g ? i) color, which is the di?erence in the Galactic extinction corrected (g ? i) for the quasar and that of the mean of the quasars at that redshift . If ?(g ? i) is not de?ned for the quasar, which occurs for objects at either z < 0.12 or z > 5.12 the column will contain ¡°?9.000¡±. See Section 7 for a description of this quantity. 29. Spectral index ¦Á¦Í (see Section 7.2). 30. Median signal-to-noise ratio computed over the whole spectrum. 31. Median signal-to-noise ratio computed over the window 1,650-1,750 ? in the quasar rest frame. 32. Median signal-to-noise ratio computed over the window 2,950-3,050 ? in the quasar rest frame. 33. Median signal-to-noise ratio computed over the window 5,100-5,250 ? in the quasar rest frame. 34-37. FWHM (km s?1 ), blue and red half widths at halfmaximum (HWHM; the sum of the latter two equals FWHM),
16

and amplitude (in units of the median rms pixel noise, see Section 4) of the C iv emission line. 38-39. Rest frame equivalent width and corresponding uncertainty in ? of the C iv emission line. 40-43. Same as 34-37 for the C iii] emission complex. It is well known that C iii]¦Ë1909 is blended with Si iii]¦Ë1892 and to a lesser extend with Al iii¦Ë1857. We do not attempt to deblend these lines. Therefore the redshift and red HFHM derived for this blend correspond to C iii]¦Ë1909. The blue HFWM is obviously a?ected by the blend 44-45. Rest frame equivalent width and corresponding uncertainty in ? of the C iii] emission complex. 46-49. Same as 34-37 for the Mg ii emission line. 50-51. Rest frame equivalent width and corresponding uncertainty in ? of the Mg ii emission line. 52. BAL ?ag from the visual inspection. It is set to 1 if a BAL feature was seen during the visual inspection. It is set to 0 otherwise. Note that BAL quasars are ?agged during the visual inspection at any redshift. 53-54. Balnicity index (BI) for C iv troughs, and its error, expressed in km s?1 . See de?nition in Section 5.1. The Balnicity index is measured for quasars with z > 1.57 only. If the BAL ?ag from the visual inspection is set to 1 and the BI is equal to 0, this means either that there is no C iv trough (but a trough is seen in another transition) or that the trough seen during the visual inspection does not meet the formal requirement of the BAL de?nition. In cases with bad ?ts to the continuum, the balnicity index and its error are set to -1. 55-56. Absorption index, and its error, for C iv troughs expressed in km s?1 . See de?nition in Section 5.1. In cases with bad continuum ?t, the absorption index and its error are set to -1. 57-58. Detection index, and its error, for C iv troughs expressed in km s?1 . See de?nition in Section 5.1. In cases with bad continuum ?t, the detection index and its error are set to -1.

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

59. Following Trump et al. (2006), we calculate the reduced ¦Ö2 for each trough from Eq. 3. We require that troughs have ¦Ö2 trough > 10 to be considered as true troughs (see Section 5.1). 60. Number of troughs of width larger than 2,000 km s?1 61-62. Full velocity range over which C iv troughs are at least 10% below the continuum for troughs of width larger than 2,000 km s?1 . 63. Number of troughs of width larger than 450 km s?1 64-65. Full velocity range over which C iv troughs are at least 10% below the continuum for troughs of width larger than 450 km s?1 . 66-68. Rest frame equivalent width in ? of Si iv, C iv and Al iii troughs detected in BAL quasars with BI > 500 km s?1 and SNR 1700 > 5. They are set to 0 otherwise or in cases where no trough is detected and to -1 if the continuum is not reliable. 69-70 The SDSS Imaging Run number and the Modi?ed Julian Date (MJD) of the photometric observation used in the catalog. The MJD is given as an integer; all observations on a given night have the same integer MJD (and, because of the observatory¡¯s location, the same UT date). For example, imaging run 94 has an MJD of 51075; this observation was taken on 1998 September 19 (UT). 71-74 Additional SDSS processing information: the photometric processing rerun number; the camera column (1¨C6) containing the image of the object, the ?eld number of the run containing the object, and the object identi?cation number (see Stoughton et al. 2002, for descriptions of these parameters). 75-84. DR9 ?ux and errors (not corrected for Galactic extinction) in the ?ve SDSS ?lters. 85-89. TARGET photometric ?ux in the ?ve SDSS ?lters. 90. Galactic extinction in the u band based on the maps of Schlegel et al. (1998). For an RV = 3.1 absorbing medium, the extinctions in the SDSS bands can be expressed as A x = C x Au where x is the ?lter (ugriz), and values of Cg,r,i,z are 0.736, 0.534, 0.405, and 0.287. See Schla?y & Finkbeiner (2011) however. 91. The logarithm of the Galactic neutral hydrogen column density along the line of sight to the quasar. These values were estimated via interpolation of the 21-cm data from Stark et al. (1992), using the COLDEN software provided by the Chandra X-ray Center. Errors associated with the interpolation are typically expected to be less than ¡Ö 1 ¡Á 1020 cm?2 (e.g., see ¡ì5 of Elvis et al. 1994). 92. The logarithm of the vignetting-corrected count rate (photons s?1 ) in the broad energy band (0.1¨C2.4 keV) in the ROSAT All-Sky Survey Faint Source Catalog (Voges et al. 2000) and the ROSAT All-Sky Survey Bright Source Catalog (Voges et al. 1999). The matching radius was set to 30¡ä¡ä (see Section 7.5.1); 93. The SNR of the ROSAT measurement. 94. Angular Separation between the SDSS and ROSAT All-Sky Survey locations (in arcseconds). 95-98. UV ?uxes and errors from GALEX, aperturephotometered from the original GALEX images in the two bands FUV and NUV (see Section 7.5.2). 99-100. The J magnitude and error from the Two Micron All Sky Survey All-Sky Data Release Point Source Catalog (Cutri et al. 2003) using a matching radius of 2.0¡ä¡ä (see Section 7.5.3). A non-detection by 2MASS is indicated by a ¡°0.000¡± in these columns. Note that the 2MASS measurements are Vega-based, not AB, magnitudes. 101-102. SNR in the J band and corresponding 2MASS jr d ?ag. 103-106. Same as 98-101 for the H -band.

107-110. Same as 98-101 for the K -band. 111. Angular separation between the SDSS and 2MASS positions (in arcseconds). 112-113. The w1 magnitude and error from the Wide-?eld Infrared Survey Explorer (WISE; Wright et al. 2010) All-Sky Data Release Point Source Catalog using a matching radius of 2¡± (see Section 7.5.4). 114-115 SNR and ¦Ö2 in the WISE w1 band. 116-119. Same as 111-114 for the w2-band. 120-123. Same as 111-114 for the w3-band. 124-127. Same as 111-114 for the w4-band. 128. Angular separation between SDSS and WISE positions (in arcseconds). 129. If there is a source in the FIRST catalog (version July 2008) within 2.0¡ä¡ä of the quasar position, this column contains the FIRST peak ?ux density (see Section 7.5.5). An entry of ¡°0.000¡± indicates no match to a FIRST source; an entry of ¡°?1.000¡± indicates that the object does not lie in the region covered by the ?nal catalog of the FIRST survey. 130. The SNR of the FIRST source whose ?ux is given in column 128. 131. Angular separation between the SDSS and FIRST positions (in arcseconds).

7. Summary of sample

The DR9Q catalog contains 87,822 unique, visually con?rmed quasars, of which 65,205 and 61,931 have, respectively, z ¡Ý 2 and z > 2.15. 91% of these quasars were discovered by BOSS. The ?rst two years of operations cover an area of approximately 3,275 deg2 leading to a mean density of >15 quasars with z > 2.15 per square degree. In the following, we describe the properties of the quasar population drawn from the whole sample. However, we also provide a uniform ?ag (see Section 7.4). A sample of quasars with uniform > 0 is a su?ciently statistical sample for, e.g., clustering measurements on some scales (e.g., White et al. 2012) and luminosity function demographics. Quasars from the present catalog span a range of redshift from z = 0.058 to z = 5.855. The redshift distribution is given in Fig. 22 together with that from SDSS-DR7 (red histogram, Schneider et al. 2010). It is apparent from the ?gure that BOSS primarily targets z > 2.15 quasars as it was designed. Only 7,932 of those quasars were previously known, e.g. detected by previous surveys and the majority of those were previous SDSS discoveries. The DR9Q catalog thus contains about 2.6 times more high-redshift quasars than the whole SDSS-I/II survey. The two peaks in the redshift distribution at z ? 0.8 and z ? 1.6 are due to known degeneracies in the SDSS color space. Six objects have z < 0.1. These are Seyfert galaxies that were classi?ed as quasars in order to di?erentiate them from normal galaxies. Fig. 23 displays the redshift distributions in the redshift range of interest for BOSS for the whole sample (black histogram), the CORE sample (red histogram) and the BONUS sample (blue histogram). The CORE sample is selected via uniform target selection (see details in Ross et al. 2012), and is designed for statistical studies of the quasar population (see Section 7.4). On the other hand, the BONUS sample is the result of the combination of four target selection algorithms. This sample was designed to maximize the number of high-redshift quasars. Typical spectra are shown in Fig. 24. Table 5 gives the number of objects targeted by the various selection methods and visually inspected (column #2) as de17

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

SDSS?DR7

Full DR9 sample 4000 CORE sample BONUS sample

Number of quasars

SDSS?DR9 (new quasars)

Number of quasars

4000

SDSS?DR9 (all)

2000

2000

0

2

4

0 2.0

2.5

3.0

3.5

4.0

Redshift

Redshift
Fig. 23. Redshift distribution of SDSS-DR9 quasars in the range 2.00-4.00 for the whole distribution (black histogram), the CORE sample (red histogram) and the BONUS sample (blue histogram). The CORE sample was uniformly selected through the likelihood method (Kirkpatrick et al. 2011) during most of the ?rst year of operation and the XDQSO method (Bovy et al. 2011) for the second year. The BONUS sample was selected through a combination of four target selection algorithms to maximize the number of high-redshift quasars in the sample. Fig. 28 shows the SDSS (u-g), (g-r), (r-i), and (i-z) colors as a function of redshift for the DR9Q catalog. Also shown are the mean color in redshift bins (thin red solid line), and the models described in Ross et al. (2012, in prep., thick colored lines).This model is systematically bluer than the data at low redshift; BOSS target selection systematically excludes UVexcess quasars. The trends with redshift are due to various emission lines moving in and out of the SDSS broadband ?lters, and the onset of the Lyman-¦Á forest and Lyman-limit systems (e.g., Fan 1999; Hennawi et al. 2010; Richards et al. 2002, 2003; Bovy et al. 2012; Peth et al. 2011); see also Prochaska et al. (2009) and Worseck & Prochaska (2011) for biases in the SDSS target selection. Fig. 29 shows the SDSS color-color diagrams for the quasars in the DR9Q catalog. This ?gure illustrates the redshift dependence of quasar colors (see also Fig. 28; Fan 1999). The quasars at z ? 2.7 are located in the stellar locus (black contours).
7.2. Spectral index and composite spectra
¦Í The quasar continuum can be expressed as fcont ¡Ø ¦Í¦Á rest , where ¦Á¦Í is the spectral index. This index is obtained by ?tting a power law over wavelength ranges outside the Lyman-¦Á forest and devoid of strong emission lines. The regions of the ?ts are 1450-1500, 1700-1850 and 1950-2750 ? in the rest frame. The continuum is iteratively ?tted to remove absorption lines and to limit the impact of the iron emission blends on the ¦Á¦Í measurement. The distribution of the quasar spectral index of SDSS-DR7 quasars re-observed by BOSS is shown in Fig. 31. The median spectral index measured for BOSS spectra (black histogram) is ¦Á¦Í,DR9 = ?0.517 while the median value measured with SDSSDR7 spectra is ¦Á¦Í,DR7 = ?0.862. This discrepancy is mainly the

Fig. 22. The redshift distribution of quasars from this catalog is displayed in black. The same distribution is shown for newly discovered quasars only (dashed blue histogram). Most of the SDSS-DR9 quasars have a redshift greater than 2. The redshift distribution of quasars from the SDSS-DR7 catalog (Schneider et al. 2010) is shown for comparison in red. The latter is dominated by quasars at low redshift. The present catalog contains 2.6 times more quasars at z > 2.15 than the DR7 catalog.

scribed in Ross et al. (2012) together with the number of objects classi?ed by visual inspection as quasars (column #3), quasars with z > 2.15 (column #5), stars (column #6) or galaxies (column #7). Column #8 and #9 give, respectively, the number of objects with good spectra but uncertain identi?cation and the number of objects with data of too low SNR to allow for identi?cation (see Section 3 for a detailed description of the di?erent categories). Note that a single object can be selected by several methods. BOSS targets fainter objects than SDSS-I/II. The r-PSF magnitude distribution (corrected for Galactic extinction) of SDSS-DR9 quasars is shown in Fig. 25 (top panel), and peaks at ?20.8. The median signal-to-noise ratio computed over the whole spectrum versus the r-PSF magnitude is shown in the bottom panel of Fig. 25. Percentiles are indicated in grey. The i-PSF magnitude distribution of quasar candidates, spectroscopically con?rmed quasars and z > 2.15 con?rmed quasars are shown in Fig. 26. There is no drop of the success rate at high magnitude, indicating again that the SNR threshold used to de?ne a survey quality plate is well chosen. Fig. 27 shows the distribution of objects in the redshiftluminosity (L versus z) plane for the BOSS survey (black contours and points) together with the same quantities for the SDSSDR7 (red contours and points; Schneider et al. 2010). We calculate the absolute i-band (at z = 2) magnitudes, Mi , using the observed i-band PSF magnitudes and the K-corrections given in Table 4 of Richards et al. (2006). This shows the coverage available for calculating the evolution of the faint end of the quasar luminosity function, and for placing constraints on the luminosity dependence of quasar clustering (White et al. 2012).
18

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release Selection QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO QSO CORE BONUS KNOWN MIDZ KNOWN LOHIZ NN UKIDSS KDE COADD LIKE FIRST BOSS KDE CORE MAIN BONUS MAIN CORE ED CORE LIKE KNOWN SUPPZ AAL AALS IAL RADIO RADIO AAL RADIO IAL NOAALS GRI HIZ RIZ Maskbits BOSS TARGET1 10 11 12 13 14 15 16 17 18 19 40 41 42 43 44 ANCILLARY TARGET1 22 23 24 25 26 27 28 29 30 31 ANCILLARY TARGET2 0 1 2 3 4 5 7 8 # Objects 3,468 4,259 9,927 24 72,365 48 1,362 90,762 3,348 92,203 67,677 148,085 22,715 23,951 24 174 281 80 72 58 31 32 1,117 335 728 62 28 183 1,380 576 332 208 1,887 # QSO 1,3 803 9,775 24 45,319 27 305 54,313 2,507 47,564 41,817 76,660 15,019 17,635 24 172 277 80 71 58 30 31 354 0 47 2 0 81 856 549 149 208 568 # QSO z > 2.15 1,084 437 9,121 0 34,864 20 202 35,869 1,629 34,252 32,355 53,000 12,387 12,522 0 1 2 0 0 0 0 0 343 0 42 2 0 39 296 263 15 0 166 # STAR 1,975 3,319 36 0 25,541 19 921 32,909 433 42,248 23,930 65,320 7,055 5,591 0 0 0 0 0 0 0 0 373 272 545 55 25 89 431 6 131 0 1,185 # GALAXY 64 89 3 0 569 2 56 1,699 142 1,021 799 2,931 198 319 0 1 0 0 0 0 0 0 177 4 78 1 0 4 85 3 29 0 39 #? 41 30 56 0 407 0 50 938 174 647 461 1,467 169 188 0 0 1 0 1 0 0 1 26 3 11 0 0 9 5 14 7 0 35 # BAD 11 18 57 0 529 0 30 903 92 723 670 1705 274 218 0 1 3 0 0 0 1 0 187 56 47 4 3 0 3 4 16 0 60

HIZQSO82 HIZQSOIR KQSO BOSS QSO VAR QSO VAR FPG RADIO 2LOBE QSO QSO SUPPZ QSO VAR SDSS

Table 5. Number of visually inspected DR9 BOSS quasar targets (third column) and identi?cations in the DR9Q catalog for each target selection method (?rst column; see Table 4 of Ross et al. 2012, and Tables 6 and 7 in the Appendix of Dawson et al., 2012). These categories overlap because many objects are selected by multiple algorithms.

consequence of the inaccuracy of the BOSS ?ux calibration in the blue (see Fig. 5). This may explain as well the fact that the distribution is more symmetric than previously measured (e.g. Richards et al. 2003) lacking the red tail. Therefore the reader should be careful of this measurement using BOSS quasar spectra. Although the absolute ?ux calibration is in error, it is interesting to compare the composite spectra in di?erent absolute magnitude bins. They are displayed in Fig. 32 for the absolute magnitude bins ?25.0 < Mi < ?23.5 (magenta), ?26.5 < Mi < ?25.0 (blue) and Mi < ?26.5 (black). The equivalent widths of the emission lines decreases with increasing luminosity. This is the well-known Baldwin e?ect (Baldwin 1977). The rest equivalent widths of the most important equivalent emission lines is given for di?erent absolute magnitude bins in Table 6.
7.3. Rest equivalent widths in individual spectra

Mi [z = 2] ?25.0 < Mi < ?23.5 ?26.5 < Mi < ?25.0 Mi < ?26.5

Restframe equivalent width (?) Si iv C iv C iii] Mg ii 10.6 65.8 31.3 44.2 9.5 48.6 27.4 37.0 8.3 34.0 23.4 29.8

Table 6. Rest frame equivalent widths measured on the composite spectra displayed in Fig. 32
.

As explained in Section 4.5, we used ?ve PCA components to ?t the emission lines and derive their redshift. We used these ?ts to measure also the rest equivalent width and widths (FWHM and half widths at half maximum) of the emission lines. The continuum is ?tted as a power law to the best PCA component ?t over the windows 1,450-1,470 ? and 1,650-1,820 ?. We modi-

?ed the windows used by Shen et al. (2011) (1,445-1,465 ? and 1,700-1,705 ?) to minimize the fraction of bad ?ts, especially for emission lines narrower than the mean. Fig. 30 shows the comparison between the rest equivalent width measured on SDSS DR7 spectra by Shen et al. (2011) and that measured on BOSS spectra of the same quasars. Our rest equivalent widths are about 10 % smaller on average. This systematic shift is likely related to a di?erence in the rest frame wavelength range used to compute the rest frame equivalent width. While we strictly limited the equivalent width computation to the 1,500-1,600 ? range, Shen et al. (2011) used this range to ?t the line but accounted for the extra wings to estimate the rest frame equivalent width. The rms scatter is about 33 %.
19

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

6 4 2 0
SDSS J000552.36?000656.1 z = 5.855

300 200 100 0 8
SDSS J155152.46+191104.1 z = 2.85 Mi[z=2] = ?30.91

f¦Ë (10?17 erg s?1 cm?2 ??1)

4000

6000

8000

4000

6000

8000

4 2

SDSS J213510.14+063137.9 z = 2.298 SNR = 2.18

6 4 2

SDSS J123410.46+352313.5 z = 2.505 SNR = 2.29

0

4000

6000

8000

0 8

4000

6000

8000

10 6 5
022833.24?030029.8 z = 2.704 SNR = 2.35

SDSS J132402.27+001911.6 z = 3.5 SNR = 3.56

4 2

0

4000

6000

8000

0

4000

6000

8000

Wavelength (?)
Fig. 24. First row, left: spectrum of the highest redshift quasar (z = 5.855) observed by BOSS ; this quasar was discovered by Fan et al. (2004); the quasar with highest redshift discovered by BOSS is SDSS J222018.50?010147.0 at z = 5.605. First row, right: spectrum of the most luminous ( Mi [z = 2] = ?30.91) quasar available in this catalog. Middle and bottom rows: Four typical quasar spectra selected to be representative in terms of SNR at di?erent redshifts (z ? 2.3, 2.5, 2.7, 3.5). The SNR listed is the median SNR per pixel over the whole spectrum. All the spectra were boxcar median smoothed over 5 pixels. We checked by hand some of the largest discrepancies and found that our procedure seems to behave well. We applied our procedure to both SDSS-DR7 data and BOSS data from the same quasars. The mean di?erence is 4% and rms 25%. Statistical errors are of the order of 15%, and variability can account for another 15% (see e.g. Bentz et al. 2009; Wilhite et al. 2006). BOSS spectra are also of better quality. The rest equivalent widths are listed in the catalog for C iv, the C iii] complex and Mg ii. A value of ?1 indicates that the PCA failed to ?t the emission line. The variance was computed as the integral over the width of the line of the variance in each pixel. Note however that errors are mostly due to the position of the continuum.
7.4. Uniform sample

gets that were selected with the XDQSO technique (Bovy et al. 2011) after XDQSO became the CORE targeting algorithm of choice for BOSS (e.g., in or after Chunk 12; Ross et al. 2012). XDQSO will remain the BOSS quasar target algorithm for the rest of the survey, so this uniform = 1 sample will grow signi?cantly in subsequent releases.

We provide a similar uniform ?ag in our catalog to previous versions of the SDSS quasar catalogs (e.g., Schneider et al. 2007). Quasars in our catalog with uniform = 1 are CORE tar20

Quasars with uniform = 2 would have been selected by XDQSO if it had been the CORE algorithm prior to Chunk 12. uniform = 2 objects are quite complete to what XDQSO would have selected (e.g., Ross et al. 2012), so uniform > 0 is a su?ciently statistical sample for, e.g., clustering measurements on some scales (e.g., White et al. 2012). Quasars in our catalog with uniform = 0 are not homogeneously selected CORE targets. Finally, the very few (30) quasars in our catalog with uniform = -1 have no chunk information, but are a su?ciently small sample to be discarded for the purposes of statistical analyses.

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

4 3 4

(u?g)

(g?r)
0 2 4

2 1 0

2

0 0 2 4

Redshift
2 1

Redshift

(i?z)
0 2 4

1

(r?i)

0

0 ?1 0 2 4

Redshift

Redshift

Fig. 28. SDSS colors vs. redshift for quasars in the DR9Q catalog. The thin solid red line is the median color in bins of redshift. The thick color lines are models from simulations used to determine the BOSS quasar completeness (McGreer et al., in prep.; see also Ross et al. 2012, in prep.) for three di?erent quasar luminosities: Mi [z = 2] = ?22.49 (cyan), Mi [z = 2] = ?24.99 (green) and Mi [z = 2] = ?27.49 (orange); and empirical tracks for the DR7 quasars (blue Bovy et al. 2011). These simulations include the Baldwin e?ect (Baldwin 1977). Therefore the colors depend on the quasar luminosity. The model is systematically bluer than the data at low redshift because BOSS systematically excludes UV-excess sources.
7.5. Multiwavelength matching
7.5.1. ROSAT all sky survey

sible detection quality issues and therefore are not included in the present quasar catalog.
7.5.2. The Galaxy Evolution Explorer (GALEX)

We cross-correlate the DR9Q catalog with the ROSAT all sky survey catalogues listing the sources detected in the energy band 0.1?2.4 keV. The matching radius is set to 30¡±. We report the logarithm of the vignetting-corrected count rate (photons s?1 ) from the ROSAT All-Sky Survey Faint Source Catalog (Voges et al. 2000) and the ROSAT All-Sky Survey Bright Source Catalog (Voges et al. 1999). An entry of ¡±?9.000¡± in the column RASS COUNTS indicates no X-ray detection. We also report the SNR at the position of the quasar and the separation between the quasar and the X-ray source. There are 16 matches with the Bright Source Catalog and 298 with the Faint Source Catalog. It never happened to ?nd more than one source within the matching radius. No DR9 quasar is detected both in the Bright and Faint Source catalogues. Only the most reliable detections were included in our catalog: X-ray counterparts for 13 quasars were ?agged for pos-

The GALEX space mission (Martin et al. 2005) has performed an all-sky imaging survey in two UV bands (FUV: 1350 to 1750 ? ; NUV: 1750 to 2750 ?) down to mAB ? 20.5 and a mediumdeep imaging survey that reaches mAB ? 23 (e.g., Bianchi et al. (2011)). Both surveys are used here. GALEX images are force photometering GALEX images (from GALEX Data Release 5) at the SDSS-DR8 centroids (Aihara et al. 2011), such that low signal-to-noise point-spread function (PSF) ?uxes of objects not detected by GALEX is obtained. A total of 77,236 quasars lie in the GALEX FUV footprint, 78,062 lie in the NUV footprint, and 77,197 are covered by both bandpasses.
21

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

3 3 2 2

(g?r)

1 0 ?1

(r?i)
0 2 4

1 0 ?1 ?1

z < 1.5 1.5 < z < 2.0 2.0 < z < 2.2 2.2 < z < 2.5 2.5 < z < 2.8 2.8 < z < 3.0 3.0 < z < 3.5 3.5 < z < 4.0 4.0 < z

(u?g)

0

(g?r)

1

2

3

2 1

i?PSF magnitude
0 1 2

22 20 18 16 ?1

(i?z)

0 ?1 ?1

(r?i)

0

(g?r)

1

2

3

Fig. 29. Color-color diagrams for all quasars in the DR9Q catalog. Colors of points encode their redshifts (see top right panel). The stellar locus is represented with black contours.
7.5.3. The two micron all sky survey (2MASS)

We cross-correlate the DR9Q catalog with the All-Sky Data Release Point Source Vatalog (Cutri et al., 2003) using a matching radius of 2.0¡±. Together with the Vega magnitudes in the J, H and K-bands (xMAG with x = J, H or K) and their errors (ERR xMAG), we report the SNR (xSNR) since the errors on the magnitude do not di?erentiate the 2¦Ò upper limits (in a 4¡± radius aperture) from detections. We also give for each band the value of the 2MASS ?ag rd ?g[1] (entry xRDFLAG) which gives the meaning of the peculiar values of xMAG and ERR xMAG (see http://www.ipac.caltech.edu/2mass/releases/allsky/doc/explsup.html) There are 1,441 matches in the catalog.
7.5.4. The Wide-Field Infrared Survey (WISE)

for 3.4, 4.6, 12 ?m, and ¡Ö12¡± for 22 ?m (Wright et al. 2010). After testing for various matching radii (1¡±, 2¡±, 6¡±, 12¡±, 18¡±), we use a matching radius of 2.0¡±, and a total of 45,987 rows of WISE photometry data are returned, along with the separation in arcseconds between the SDSS and WISE source (stored in SDSSWISE SEP). In the DR9Q catalog, we report the photometric quantities, w xmpro, w xsigmpro, w xsnr, w xrchi2, where x = 1 ? 4 and represents the four WISE bands centered at wavelengths of 3.4, 4.6, 12 and 22 ?m. These magnitudes are in the Vega system, and are measured with pro?le-?tting photometry (see e.g. http://wise2.ipac.caltech.edu/docs/release/allsky/expsup/sec2 2a.html and http://wise2.ipac.caltech.edu/docs/release/allsky/expsup/sec4 4c.html#wpro). Formulae for converting WISE Vega magnitudes to ?ux density units (in Janskys) and AB magnitudes are given in Wright et al. (2010) and Jarrett et al. (2011) and also here: http://wise2.ipac.caltech.edu/docs/release/- allsky/expsup/sec4 4h.html#conv2?ux Although the MIR WISE properties of the BOSS quasars will be valuable for many scienti?c questions, we strongly urge the user to not only consider the various ¡°health warnings¡± associated with using the BOSS quasar dataset (as given in Section

We take the DR9Q catalog, and match to the Wide-Field Infrared Survey (WISE; Wright et al. 2010) All-Sky Data Release5 , asking for all quasars in the DR9Q catalog that are in the All-Sky Source Catalog. WISE photometry covers four bands, 3.4, 4.6, 12 and 22 ?m, where the angular resolution of WISE is ¡Ö6¡±
5

http://wise2.ipac.caltech.edu/docs/release/allsky/

22

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release

Number of quasars (x103)

4 2 0

Fraction of quasars

6

0.6

0.4

18 20 22 r?PSF magnitude
90%

0.2

18

19

20

21

22

i?PSF magnitude

40 30 SNR 20 10 0

Fig. 26. Fraction of quasar candidates con?rmed as quasars (red histogram) and z > 2.15 quasars (blue histogram) versus the iband PSF magnitude (corrected for Galactic extinction).
?30

70% 30% 10%

Mi[z = 2]

?25

18 20 22 r?PSF magnitude
?20 0 2

SDSS?DR9 SDSS?DR7

Fig. 25. Top panel: Distribution of r magnitude of SDSS-DR9 quasars (PSF; corrected for Galactic extinction). Bottom panel: Median SNR per pixel over the whole spectrum with respect to the r-PSF magnitude (black histogram). Percentiles are indicated in grey.

4

redshift
Fig. 27. L ? z plane for SDSS-DR9 quasars (black contours and points) and SDSS-DR7 quasars (red contours and points; Schneider et al. 2010). The luminosity assumes H0 = 70 km s?1 Mpc?1 and the K-correction is given by Richards et al. (2006) who consider K (z = 2) = 0. Contours are drawn at constant point density. Note that, as in SDSS-I/II, FIRST sources are automatically included in the target selection. An additional cut in color (ug > 0.4) is added to avoid as much as possible low-redshift sources (Ross et al. 2012). The catalog contains 3,283 FIRST matches.

2) but also those connected to the WISE All-Sky Release Data Products6 . We do not investigate any of the 2MASS, or UKIDSS properties associated with WISE matches here, but these are being investigated in Ross et al. (2012, in prep.).
7.5.5. FIRST

We cross-correlate the DR9Q catalog with objects that are detected in the FIRST radio survey (Becker et al. 1995). We use the version of July 2008. If there is a source in the FIRST catalog within 2.0¡ä¡ä of the quasar position, we indicate the FIRST peak ?ux density and the SNR. Note that extended radio sources may be missed by this matching.
6

We provide a supplemental list of 949 quasars, of which 318 at z > 2.15, that have been identi?ed among quasar targets after DR9 was ¡°frozen¡± (Section 8.1) or among galaxy targets http://wise2.ipac.caltech.edu/docs/release/allsky/expsup/sec1 4.html (Section 8.2). This supplemental list of quasars is provided in
23

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release
8 HI?Ly?¦Á

2.5

Normalized near 1450? (rest) flux

2.0

6

?25.0 < Mi < ?23.5 ?26.5 < Mi < ?25.0 Mi < ?26.5

log EWCIV,DR7 (?)

HI?Ly?¦Â

4

1.5

CIII]

2

SiIV

CIV

1.0

0 1000 1500 2000 2500 3000

Restframe wavelength (?)

1.0

1.5

2.0

2.5

log rEWCIV,DR9 (?)

Fig. 30. Rest frame equivalent width of the C iv emission line of SDSS-DR7 quasars, measured from the SDSS-DR7 spectra (Shen et al. 2011) and spectra obtained by BOSS (this work).

Fig. 32. Composite spectra of BOSS quasars in di?erent ranges of absolute magnitude: ?25.0 < Mi < ?23.5 (magenta), ?26.5 < Mi < ?25.0 (blue) and Mi < ?26.5 (black). All the spectra were normalized to have a ?ux unity near 1450 ? in the quasar rest frame. The Baldwin e?ect is apparent (see Table 6).

8.1. Additional quasars from the quasar target list
300
SDSS?DR9

200

SDSS?DR7

100

0 ?4

?2

0

Spectral index ¦Á¦Í

The quasar catalog was frozen7 in February 2012, but we subsequently identi?ed an additional 301 quasars (294 with z > 2.15) that have been targeted as quasar candidates. Some of these were identi?ed with improvements of the pipeline. Others are identi?ed from a good spectrum taken on a plate which was not survey quality so was not included in the ?rst inspection. A handful are objects that have been misidenti?ed during the ?rst inspection but were corrected during the checks. In addition, a few ancillary programs were not included in the ?rst inspection. Therefore the supplemental list contains the quasars that have been targeted only by these programs. These comprise 500 quasars of which 20 have z > 2.15. Finally, only objects classi?ed as QSO or QSO BAL are listed in the o?cial DR9Q catalog. Objects classi?ed as QSO Z? (126 in total; 122 corresponding to the DR9Q inspection) are also included in the supplemental list. Most of the latter are very peculiar BAL quasars.
8.2. Galaxy targets

Fig. 31. Distribution of the spectral index ¦Á¦Í of z > 2 of SDSSDR7 quasars (red histogram) re-observed by BOSS (black histogram). The spectral index was measured using the rest frame wavelength ranges 1450-1500, 1700-1850 and 1950-2750 ?. As seen already from the composite spectrum shown in Fig. 5, the spectral indices measured using SDSS-DR9 quasar spectra are bluer than those obtained using DR7 spectra (see Section 7.2).

Number of objects

the same format as the DR9Q catalog but in a separate ?le and is meant to be merged with the whole catalog for DR10. Fig. 33 gives the redshift distribution of these additional quasars. The list is available together with the DR9Q catalog and the list of objects classi?ed as QSO ? at the SDSS public website http://www.sdss3.org/dr9/algorithms/qso catalog.php.
24

In order to be as complete as possible we also tried to identify serendipitous quasars. For this, we visually inspected all objects from the BOSS galaxy target list that the pipeline reliably classi?es (ZWARNING = 0) as quasars with z > 2, and all objects classi?ed as GALAXY/BROADLINE. We also visually inspected 10% of the galaxy targets classi?ed as quasars with ZWARNING not equal to zero; none were in fact quasars, so we did not inspect the remaining such objects. A large fraction (65%) of the unclassi?able objects are attributed to the QSO class, but with low signi?cance. They include all sorts of unidenti?ed objects and spectra with calibration problems but probably very few real quasars, if any. In total we identi?ed 22 additional quasars, 4 of which were at z > 2.15, out of more than 3,000 targets. There is one quasar classi?ed as QSO Z?.
7 By which we mean no additional quasar or change in the identi?cations were intended to be included in the catalog

MgII

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release
80

Number of Quasars

60

40

20

Funding for SDSS-III has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, and the U.S. Department of Energy O?ce of Science. The SDSS-III web site is http://www.sdss3.org/. SDSS-III is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS-III Collaboration including the University of Arizona, the Brazilian Participation Group, Brookhaven National Laboratory, University of Cambridge, Carnegie Mellon University, University of Florida, the French Participation Group, the German Participation Group, Harvard University, the Instituto de Astro?sica de Canarias, the Michigan State/Notre Dame/JINA Participation Group, Johns Hopkins University, Lawrence Berkeley National Laboratory, Max Planck Institute for Astrophysics, Max Planck Institute for Extraterrestrial Physics, New Mexico State University, New York University, Ohio State University, Pennsylvania State University, University of Portsmouth, Princeton University, the Spanish Participation Group, University of Tokyo, University of Utah, Vanderbilt University, University of Virginia, University of Washington, and Yale University.

References
0 0 2 4

Redshift
Fig. 33. Redshift distribution of the 949 additional quasars described in Section 8.

9. Conclusion
The quasar catalog presented here contains 87,822 quasars, 61,931 having z > 2.15, with robust identi?cation from visual inspection and redshift derived from the ?t of PCA components to the spectra. This catalog has been gathered during the ?rst two years of the BOSS operation covering 3,275 deg2 . It will be the basis for studies of the luminosity function and the spatial distribution of quasars as well as studies of the clustering properties of the Lyman-¦Á forest. In particular it will be used to measure for the ?rst time the BAO clustering signal in the IGM at z ? 2.3 from the Lyman-¦Á forest. For quasars with zem > 1.57, the catalog identi?es 7,228 broad absorption line quasars from visual inspection, of which 3,130 have BI > 500 km s?1 . In the 7,317 spectra with SNR > 10, we ?nd 813 BALs with BI > 500 km s?1 corresponding to a fraction of 11.1%. We implement a procedure to identify BALs automatically, ?tting the quasar continuum with PCAs. 3,330 BALs with BI > 500 km s?1 have been identi?ed in this way, of which 821 in spectra of SNR > 10. The catalog gives their characteristics, balnicity and absorption indices. The list of BALs will be used for statistical analysis of this population of quasars. Since SDSS-DR7 z > 2.15 quasars are reobserved by BOSS, this will be a unique opportunity to study the variability of these troughs. High redshift (z > 2) quasar continua together with pixel masks, improved noise estimates, and other products designed to aid in the BAO-Lyman-¦Á clustering analysis will be released in Lee et al. (2012, in prep.). BOSS is a ?ve year program and the next version of our quasar catalog, to be released as a part of SDSS-DR10 in July 2013, should contain about two times as many quasars as the DR9Q catalog. Improvements in the pipeline will allow us to achieve identi?cation of more objects. We will also perform multiple checks and improve our procedures in order to place better constraints on the characterisitcs of the quasar spectra.
Acknowledgements. I.P. received partial support from Center of Excellence in Astrophysics and Associated Technologies (PFB 06). The French Participation Group to SDSS-III was supported by the Agence Nationale de la Recherche under contract ANR-08-BLAN-0222. W.N.B. and N.F.-A. gratefully acknowledge support from NSF AST-1108604. A.D.M. is a research fellow of the Alexander von Humboldt Foundation of Germany.

Abazajian, K. N., Adelman-McCarthy, J. K., Ag¡§ ueros, M. A., et al. 2009, ApJS, 182, 543 Ahn, C. P., Alexandro?, R., Allende Prieto, C., et al. 2012, ApJS, ArXiv:1207.7137 Aihara, H., Allende Prieto, C., An, D., et al. 2011, ApJS, 193, 29 Allen, J. T., Hewett, P. C., Maddox, N., Richards, G. T., & Belokurov, V. 2011, MNRAS, 410, 860 Baldwin, J. A. 1977, ApJ, 214, 679 Becker, R. H., White, R. L., & Helfand, D. J. 1995, ApJ, 450, 559 Bentz, M. C., Peterson, B. M., Netzer, H., Pogge, R. W., & Vestergaard, M. 2009, ApJ, 697, 160 Bianchi, L., Efremova, B., Herald, J., & et al. 2011, MNRAS, 411, 2770 Bolton, A. S., Schlegel, D. J., Aubourg, E., et al. 2012, AJ, ArXiv:1207.7326 Bovy, J., Hennawi, J. F., Hogg, D. W., et al. 2011, ApJ, 729, 141 Bovy, J., Myers, A. D., Hennawi, J. F., et al. 2012, ApJ, 749, 41 Boyle, B. J., Shanks, T., Croom, S. M., et al. 2000, MNRAS, 317, 1014 Croom, S. M., Richards, F. T., Shanks, T., & et al. 2009, MNRAS, 392, 19 Croom, S. M., Shanks, T., Boyle, B. J., et al. 2001, MNRAS, 325, 483 Croom, S. M., Smith, R. J., Boyle, B. J., et al. 2004, MNRAS, 349, 1397 Cutri, R. M., Skrutskie, M. F., van Dyk, S., et al. 2003, VizieR Online Data Catalog, 2246, 0 Dawson, K. S., Schlegel, D. J., Ahn, C. P., et al. 2012, AJ, ArXiv:1208.0022 Diamond-Stanic, A. M., Fan, X., Brandt, W. N., & et al. 2009, ApJ, 699, 782 Eisenstein, D. J., Weinberg, D. H., Agol, E., et al. 2011, AJ, 142, 72 Elvis, M., Lockman, F. J., & Fassnacht, C. 1994, ApJS, 95, 413 Fan, X. 1999, AJ, 117, 2528 Fan, X., Hennawi, J. F., Richards, G. T., et al. 2004, AJ, 128, 515 Filiz Ak, N., Brandt, W. N., Hall, P. B., et al. 2012, ApJ, 757, 114 Font-Ribera, A., Miralda-Escud? e, J., Arnau, E., et al. 2012, J. Cosmology Astropart. Phys., ArXiv:1209.4596 Francis, P. J., Hewett, P. C., Foltz, C. B., & Cha?ee, F. H. 1992, ApJ, 398, 476 Fukugita, M., Ichikawa, T., Gunn, J. E., et al. 1996, AJ, 111, 1748 Gaskell, C. M. 1982, ApJ, 263, 79 Gibson, R. R., Brandt, W. N., Gallagher, S. C., Hewett, P. C., & Schneider, D. P. 2010, ApJ, 713, 220 Gibson, R. R., Brandt, W. N., Schneider, D. P., & Gallagher, S. C. 2008, ApJ, 675, 985 Gibson, R. R., Jiang, L., Brandt, W. N., et al. 2009, ApJ, 692, 758 Gunn, J. E., Carr, M., Rockosi, C., et al. 1998, AJ, 116, 3040 Gunn, J. E., Siegmund, W. A., Mannery, E. J., et al. 2006, AJ, 131, 2332 Hall, P. B., Anderson, S. F., Strauss, M. A., et al. 2002, ApJS, 141, 267 Hennawi, J. F., Myers, A. D., Shen, Y., et al. 2010, ApJ, 719, 1672 Hewett, P. C., Foltz, C. B., & Cha?ee, F. H. 1995, AJ, 109, 1498 Hewett, P. C. & Wild, V. 2010, MNRAS, 405, 2302(HW10) Hewitt, A. & Burbidge, G. 1993, ApJS, 87, 451 Hogg, D. W., Finkbeiner, D. P., Schlegel, D. J., & Gunn, J. E. 2001, AJ, 122, 2129 ¡¦ Lupton, R. H., Schlegel, D., et al. 2004, Astronomische Nachrichten, Ivezi? c, Z., 325, 583 Jarrett, T. H., Cohen, M., Masci, F., & et al. 2011, ApJ, 735, 112 Kaspi, S., Brandt, W. N., George, I. M., et al. 2002, ApJ, 574, 643 Kirkpatrick, J. A., Schlegel, D. J., Ross, N. P., et al. 2011, ApJ, 743, 125 Lawrence, A., Warren, S. J., Almaini, O., et al. 2007, MNRAS, 379, 1599 Lee, K.-G., Bailey, S., Carithers, W., et al. 2012, in prep. Lupton, R., Gunn, J. E., Ivezi? c, Z., et al. 2001, in Astronomical Society of the Paci?c Conference Series, Vol. 238, Astronomical Data Analysis Software and Systems X, ed. F. R. Harnden Jr., F. A. Primini, & H. E. Payne, 269

25

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release
MacLeod, C., Butler, N., Anderson, S. F., et al. 2012, in American Astronomical Society Meeting Abstracts, Vol. 219, American Astronomical Society Meeting Abstracts, 243.28 Martin, D. C., Fanson, J., Schiminovich, D., et al. 2005, ApJ, 619, L1 McDonald, P. & Eisenstein, D. J. 2007, Phys. Rev. D, 76, 063009 McIntosh, D. H., Rix, H.-W., Rieke, M. J., & Foltz, C. B. 1999, ApJ, 517, L73 Morris, S. L., Weymann, R. J., Anderson, S. F., et al. 1991, AJ, 102, 1627 Noterdaeme, P., Petitjean, P., Carithers, W. C., et al. 2012, ArXiv:1210.1213 Noterdaeme, P., Petitjean, P., Ledoux, C., & Srianand, R. 2009, A&A, 505, 1087 Padmanabhan, N., Schlegel, D. J., Finkbeiner, D. P., et al. 2008, ApJ, 674, 1217 Palanque-Delabrouille, N., Magneville, C., Yeche, C., et al. 2012, A&A, ArXiv:1209.3968 Palanque-Delabrouille, N., Y eche, C., Myers, A. D., et al. 2011, A&A, 530, A122 P? aris, I., Petitjean, P., Rollinde, E., et al. 2011, A&A, 530, A50 Peth, M. A., Ross, N. P., & Schneider, D. P. 2011, AJ, 141, 105 Pier, J. R., Munn, J. A., Hindsley, R. B., et al. 2003, AJ, 125, 1559 Prochaska, J. X., Worseck, G., & O¡¯Meara, J. M. 2009, ApJ, 705, L113 Richards, G. T., Fan, X., Newberg, H. J., et al. 2002, AJ, 123, 2945 Richards, G. T., Hall, P. B., Vanden Berk, D. E., et al. 2003, AJ, 126, 1131 Richards, G. T., Myers, A. D., Gray, A. G., et al. 2009, ApJS, 180, 67 Richards, G. T., Nichol, R. C., Gray, A. G., et al. 2004, ApJS, 155, 257 Richards, G. T., Strauss, M. A., Fan, X., et al. 2006, AJ, 131, 2766 Ross, N. P., McGreer, I. D., White, M., et al. 2012, in prep. Ross, N. P., Myers, A. D., Sheldon, E. S., et al. 2012, ApJS, 199, 3 Schla?y, E. F. & Finkbeiner, D. P. 2011, ApJ, 737, 103 Schlegel, D. J., Blanton, M., Eisenstein, D., et al. 2007, in Bulletin of the American Astronomical Society, Vol. 38, 966 Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525 Schmidt, M. 1963, Nature, 197, 1040 Schneider, D. P., Fan, X., Hall, P. B., et al. 2003, AJ, 126, 2579 Schneider, D. P., Hall, P. B., Richards, G. T., et al. 2007, AJ, 134, 102 Schneider, D. P., Hall, P. B., Richards, G. T., et al. 2005, AJ, 130, 367 Schneider, D. P., Richards, G. T., Fan, X., et al. 2002, AJ, 123, 567 Schneider, D. P., Richards, G. T., Hall, P. B., et al. 2010, AJ, 139, 2360 Shang, Z., Wills, B. J., Wills, D., & Brotherton, M. S. 2007, AJ, 134, 294 Shen, Y., Greene, J. E., Strauss, M. A., Richards, G. T., & Schneider, D. P. 2008, ApJ, 680, 169 Shen, Y., Richards, G. T., Strauss, M. A., et al. 2011, ApJS, 194, 45 Shen, Y., Strauss, M. A., Oguri, M., et al. 2007, AJ, 133, 2222 Smee, S., Gunn, J. E., Uomoto, A., et al. 2012, ArXiv:1208.2233 Smith, J. A., Tucker, D. L., Kent, S., et al. 2002, AJ, 123, 2121 Spergel, D. N., Verde, L., Peiris, H. V., & et al. 2003, ApJS, 148, 175 Stark, A. A., Gammie, C. F., Wilson, R. W., et al. 1992, ApJS, 79, 77 Stoughton, C., Lupton, R. H., Bernardi, M., et al. 2002, AJ, 123, 485 Suzuki, N., Tytler, D., Kirkman, D., O¡¯Meara, J. M., & Lubin, D. 2005, ApJ, 618, 592 Trump, J. R., Hall, P. B., Reichard, T. A., et al. 2006, ApJS, 165, 1 Tucker, D. L., Kent, S., Richmond, M. W., et al. 2006, Astronomische Nachrichten, 327, 821 Vanden Berk, D. E., Richards, G. T., Bauer, A., et al. 2001, AJ, 122, 549 V? eron-Cetty, M.-P. & V? eron, P. 2006, A&A, 455, 773 Voges, W., Aschenbach, B., Boller, T., et al. 1999, A&A, 349, 389 Voges, W., Aschenbach, B., Boller, T., et al. 2000, IAU Circ., 7432, 3 Weymann, R. J., Morris, S. L., Foltz, C. B., & Hewett, P. C. 1991, ApJ, 373, 23 White, M., Myers, A. D., Ross, N. P., et al. 2012, MNRAS, 424, 933 Wilhite, B. C., Vanden Berk, D. E., Brunner, R. J., & Brinkmann, J. V. 2006, ApJ, 641, 78 Worseck, G. & Prochaska, J. X. 2011, ApJ, 728, 23 Wright, E. L., Eisenhardt, P. R. M., Mainzer, A. K., et al. 2010, AJ, 140, 1868 Y eche, C., Petitjean, P., Rich, J., et al. 2010, A&A, 523, A14 Yip, C. W., Connolly, A. J., Vanden Berk, D. E., et al. 2004, AJ, 128, 2603 York, D. G., Adelman, J., Anderson, Jr., J. E., et al. 2000, AJ, 120, 1579

26

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release Table 4. DR9Q catalog format Column 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Name SDSS NAME RA DEC THING ID PLATE MJD FIBERID Z VI Z PIPE ERR ZPIPE ZWARNING Z PCA ERR ZPCA PCA QUAL Z CIV Z CIII Z MGII SDSS MORPHO BOSS TARGET1 ANCILLARY TARGET1 ANCILLARY TARGET2 SDSS DR7 PLATE DR7 MJD DR7 FIBERID DR7 UNIFORM MI DGMI ALPHA NU SNR SPEC SNR 1700 SNR 3000 SNR 5150 FWHM CIV BHWHM CIV RHWHM CIV AMP CIV REWE CIV ERR REWE CIV FWHM CIII BHWHM CIII RHWHM CIII AMP CIII REWE CIII ERR REWE CIII FWHM MGII BHWHM MGII RHWHM MGII AMP MGII REWE MGII ERR REWE MGII BAL FLAG VI BI CIV ERR BI CIV AI CIV ERR AI CIV DI CIV ERR DI CIV CHI2THROUGH NCIV 2000 VMIN CIV 2000 VMAX CIV 2000 NCIV 450 VMIN CIV 450 VMAX CIV 450 Format A19 F11.6 F11.6 I10 I5 I6 I5 F9.4 F9.4 F9.4 I4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 I2 I20 I20 I20 I2 I5 I6 I4 I2 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 I2 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 I3 F9.4 F9.4 I3 F9.4 F9.4 Descriptiona SDSS-DR9 designation hhmmss.ss+ddmmss.s (J2000) Right Ascension in decimal degrees (J2000) Declination in decimal degrees (J2000) Thing ID Spectroscopic Plate number Spectroscopic MJD Spectroscopic Fiber number Redshift from visual inspection Redshift from BOSS pipeline Error on BOSS pipeline redshift ZWARNING ?ag Re?ned PCA redshift Error on re?ned PCA redshift Estimator of the PCA continuum quality Redshift of C iv emission line Redshift of C iii] emission complex Redshift of Mg ii emission line SDSS morphology ?ag 0 = point source 1 = extended BOSS target ?ag for main survey BOSS target ?ag for ancillary programs BOSS target ?ag for ancillary programs 1 if the quasar is known from DR7 SDSS-DR7 spectroscopic Plate number if the quasar is known from DR7 SDSS-DR7 spectroscopic MJD if the quasar is known from DR7 SDSS-FR7 spectroscopic Fiber number if the quasar is known from DR7 Uniform sample ?ag Mi [z = 2] H0 = 70km s?1 Mpc?1 , ? M = 0.3, ?¦« = 0.7, ¦Á¦Í = ?0.5 ?(g ? i) = (g ? i) ? (g ? i) redshift (Galactic extinction corrected) Spectral index measurement ¦Á¦Í Median signal-to-noise ratio over the whole spectrum Median signal-to-noise ratio in the window 1,650 - 1,750? (rest frame) Median signal-to-noise ratio in the window 2,950 - 3,050? (rest frame) Median signal-to-noise ratio in the window 5,100 - 5,250? (rest frame) FWHM of C iv emission line in km s?1 Blue HWHM of C iv emission line in km s?1 Red HWHM of C iv emission line in km s?1 Amplitude of C iv emission line in units of median rms pixel noise Rest frame equivalent width of C iv emission line in ? Uncertainty on the rest frame equivalent width of C iv emission line in ? FWHM of C iii] emission complex in km s?1 Blue HWHM of C iii] emission line in km s?1 Red HWHM of C iii] emission line in km s?1 Amplitude of C iii] emission complex in units of median rms pixel noise Rest frame equivalent width of C iii] emission line in ? Uncertainty on the rest frame equivalent width of C iii] emission complex in ? FWHM of Mg ii emission line in km s?1 Blue HWHM of Mg ii emission line in km s?1 Red HWHM of Mg ii emission line in km s?1 Amplitude of Mg ii emission line in units of median rms pixel noise Rest frame equivalent width of Mg ii emission line in ? Uncertainty on the rest frame equivalent width of Mg ii emission in ? BAL ?ag from visual inspection Balnicity index of C iv trough in km s?1 Error on the Balnicity index of C iv trough in km s?1 Absorption index of C iv trough in km s?1 Error on the absorption index of C iv trough in km s?1 Detection index of C iv trough in km s?1 Error on the detection index of C iv trough in km s?1 ¦Ö2 of the trough from Eq. 3 Number of distinct C iv troughs of width larger than 2,000 km s?1 Minimum velocity of the C iv troughs de?ned in row 60 km s?1 Maximum velocity of the C iv troughs de?ned in row 60 in km s?1 Number of distinct C iv troughs of width larger than 450 km s?1 Minimum velocity of the C iv troughs de?ned in row 63 in km s?1 Maximum velocity of the C iv troughs de?ned in row 63 in km s?1

27

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release Table 4. continued. Column Name 66 REW SIIV 67 REW CIV 68 REW ALIII 69 RUN NUMBER 70 PHOTO MJD 71 RERUN NUMBER 72 COL NUMBER 73 FIELD NUMBER 74 OBJ ID 75 UFLUX 76 ERR UFLUX 77 GFLUX 78 ERR GFLUX 79 RFLUX 80 ERR RFLUX 81 IFLUX 82 ERR IFLUX 83 ZFLUX 84 ERR ZFLUX 85 TARGET UFLUX 86 TARGET GFLUX 87 TARGET RFLUX 88 TARGET IFLUX 89 TARGET ZFLUX 90 U EXT 91 HI GAL 92 RASS COUNTS 93 RASS COUNTS SNR 94 SDSS2ROSAT SEP 95 NUVFLUX 96 ERR NUVFLUX 97 FUVFLUX 98 ERR FUVFLUX 99 JMAG 100 ERR JMAG 101 JSNR 102 JRDFLAG 103 HMAG 104 ERR HMAG 105 HSNR 106 HRDFLAG 107 KMAG 108 ERR KMAG 109 KSNR 110 KRDFLAG 111 SDSS2MASS SEP 112 W1MAG 113 ERR W1MAG 114 W1SNR 115 W1CHI2 116 W2MAG 117 ERR W2MAG 118 W2SNR 119 W2CHI2 120 W3MAG 121 ERR W3MAG 122 W3SNR 123 W3CHI2 124 W4MAG 125 ERR W4MAG 126 W4SNR 127 W4CHI2 128 SDSS2WISE SEP 129 FIRST FLUX 130 FIRST SNR 131 SDSS2FIRST SEP a All magnitudes are PSF magnitudes Format F9.4 F9.4 F9.4 I6 I6 A4 I2 I5 A20 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 F9.4 I2 F9.4 F9.4 F9.4 I2 F10.6 F10.6 F10.6 I2 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 F10.6 Description rest frame equivalent width of the Si iv trough rest frame equivalent width of the C iv trough rest frame equivalent width of the Al iii trough SDSS Imaging Run Number of photometric measurements Modi?ed Julian Date of imaging observation SDSS Photometric Processing Rerun Number SDSS Camera Column Number (1-6) SDSS Field Number SDSS Object Identi?cation Number ?ux in the u-band (not corrected for Galactic extinction) Error in u ?ux ?ux in the g-band (not corrected for Galactic extinction) Error in g ?ux ?ux in the r-band (not corrected for Galactic extinction) Error in r ?ux ?ux in the i-band (not corrected for Galactic extinction) Error in i ?ux ?ux in the z-band (not corrected for Galactic extinction) Error in z ?ux TARGET ?ux in the u-band (not corrected for galactic extinction) TARGET ?ux in the g-band (not corrected for galactic extinction) TARGET ?ux in the r-band (not corrected for galactic extinction) TARGET ?ux in the i-band (not corrected for galactic extinction) TARGET ?ux in the z-band (not corrected for galactic extinction) u band Galactic extinction (from (Schlegel et al. 1998)) log NH (logarithm of Galactic H i column density in cm?2 ) log RASS full band count rate (counts s?1 ) SNR of the RASS count rate SDSS-RASS separation in arcsec nuv ?ux (GALEX) Error in nuv ?ux f uv ?ux (GALEX) Error in f uv ?ux J magnitude (Vega, 2MASS) Error in J magnitude J-band SNR J-band photometry ?ag H magnitude (Vega, 2MASS) Error in H magnitude H-band SNR H-band photometry ?ag K magnitude (Vega, 2MASS) Error in K magnitude K-band SNR K-band photometry ?ag SDSS-2MASS separation in arcsec w1 magnitude (Vega, WISE) Error in w1 magnitude SNR in w1 band ¦Ö2 in w1 band w2 magnitude (Vega, WISE) Error in w2 magnitude SNR in w1 band ¦Ö2 in w1 band w3 magnitude (Vega, WISE) Error in w3 magnitude SNR in w1 band ¦Ö2 in w1 band w4 magnitude (Vega, WISE) Error in w4 magnitude SNR in w1 band ¦Ö2 in w1 band SDSS-WISE separation in arcsec FIRST peak ?ux density at 20 cm expressed in mJy SNR of the FIRST ?ux density SDSS-FIRST separation in arcsec

28

Isabelle P? aris et al.: The Sloan Digital Sky Survey quasar catalog: ninth data release Cit? e, 10, rue Alice Domon & L? eonie Duquet, 75205 Paris Cedex 13, France Lawrence Berkeley National Lab, 1 Cyclotron Rd, Berkeley CA, 94720, USA Department of Physics and Astronomy, University of Wyoming, Laramie, WY 82071, USA Max-Planck-Institut f¡§ ur Astronomie, K¡§ onigstuhl 17, D-69117 Heidelberg, Germany Princeton University Observatory, Peyton Hall, Princeton, NJ 08544, USA University of Washington, Dept. of Astronomy, Box 351580, Seattle, WA 98195, USA Institut de Ci` encies del Cosmos (IEEC/UB), Barcelona, Catalonia Apache Point Observatory, P.O. Box 59, Sunspot, NM 88349-0059, USA Department of Physics and Astronomy, University of Utah,UT, USA Institute for Advanced Study, Einstein Drive, Princeton, NJ 08540, USA Department of Astronomy and 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 Department of Astronomy, University of Florida, Gainesville, FL 32611-2055, USA School of Physics and Astronomy, Tel Aviv University, Tel Aviv 69978, Israel Carnegie Mellon University, Physics Department, 5000 Forbes Ave, Pittsburgh, PA 15213, USA CEA, Centre de Saclay, Irfu/SPP, 91191, Gif-sur-Yvette, France Harvard-Smithsonian Center for Astrophysics, 60 Garden St., MS#20, Cambridge, MA 02138 USA Department of Astronomy, Yale University, New Haven, CT06511, USA Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721 Faculty of Sciences, Department of Astronomy and Space Sciences, Erciyes University, 38039, Kayseri, Turkey Institute of Theoretical Physics, University of Zurich, 8057 Zurich, Switzerland Department of Physics and Astronomy, York University, Toronto, ON M3J1P3, Canada Center for Cosmology and Particle Physics, Department of Physics, New York University, 4 Washington Place, New York, NY 10003, USA National Radio Astronomy Observatory, 520 Edgemont Rd., Charlottesville, VA, 22903, USA Department of Physics and Astronomy, UC Irvine, 4129 Frederick Reines Hall, Irvine, CA 92697-4575, USA Department of Astrophysical sciences, Princeton university, Princeton 08544, USA Institute of Cosmology and Gravitation, University of Portsmouth, Dennis Sciama building, Portsmouth P01 3FX, UK Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA, UK Instituci? o Catalana de Recerca i Estudis Avanc ? ats, Catalonia Center for Cosmology and Particle Physics, Department of Physics, New York University, New York, NY 10003 USA Instituto de Astrofsica de Canarias (IAC), E-38200 La Laguna, Tenerife, Spain Departamento de Astro?sica, Universidad de La Laguna (ULL), E38205 La Laguna, Tenerife, Spain Department of Physics, Drexel University, Philadelphia, PA 19104, USA 5 Brookhaven National Laboratory, Blgd 510, Upton, NY 11375, USA Department of Physics, University of California Berkeley, Berkeley, CA 94720, USA INAF - Osservatorio Astronomico di Trieste, Via G.B. Tiepolo 11 INFN/National Institute for Nuclear Physics, Via Valerio 2, I-34127 Trieste, Italy Astronomy Department and Center for Cosmology and AstroParticle Physics, Ohio State University, 140 West 18th Avenue, Columbus, OH 43210, USA 29 PITT PACC, Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, PA 15260, USA

1
2 3

UPMC-CNRS, UMR7095, Institut d¡¯Astrophysique de Paris, F75014, Paris, France, e-mail: paris@iap.fr Departamento de Astronom? ?a, Universidad de Chile, Casilla 36-D, Santiago, Chile APC, Astroparticule et Cosmologie, Uniiversit? e Paris Diderot, CNRS/IN2P3, CEA/Irfu, Observatoire de Paris, Sorbonne Paris

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

41

ÔÞÖúÉÌÁ´½Ó
Ïà¹ØÎÄÕÂ:
LatexÖÐÎÄÅÅ°æ·½·¨
LatexÖÐÎÄÅÅ°æ·½·¨_IT/¼ÆËã»ú_×¨Òµ×ÊÁÏ¡£Latex ÖÐÎÄ ÅÅ°æ·½·¨×Ü½áÁËÏÂ Latex ¶ÔÖÐÎÄµÄÖ§³Ö,ÎÄÕÂ¶àÎªÒýÓÃÇ°ÈËµÄ³É¹û,Ëæ×ÅÒÔºóÓöµ½ÐÂµÄÎÊÌâ, ËæÊ±Ìí¼Ó´ËÎÄÄÚÈÝ Ä£°å...
latex¹«Ê½ÅÅ°æ
latex¹«Ê½ÅÅ°æ_¼ÆËã»úÈí¼þ¼°Ó¦ÓÃ_IT/¼ÆËã»ú_×¨Òµ×ÊÁÏ¡£latexÅÅ³öÀ´µÄ¹«Ê½Òª±ÈwordµÈÒªºÃ¿´µÃ¶à,µ«ÒòÎªÒªÐ´´úÂë,²»ÊÇ¿É¼û¼´¿ÉµÃµÄ·½Ê½,ËùÒÔÈëÃÅÆðÀ´Ò²±È½ÏÄÑ,ÕâÊÇ...
latex¿Æ¼¼ÂÛÎÄÅÅ°æ½Ì³Ì
latex¿Æ¼¼ÂÛÎÄÅÅ°æ½Ì³Ì_¼ÆËã»úÈí¼þ¼°Ó¦ÓÃ_IT/¼ÆËã»ú_×¨Òµ×ÊÁÏ¡£LATEX¿Æ¼¼ÅÅ°æ 1 ...\end{document} Õâ¸öÀý×Ó¿ÉÒÔÏóÓ¢ÎÄÎÄµµÒ»ÑùµÄ±àÒëµÃµ½ÕýÈ·µÄÊä³ö¡£ ? TY: ...
LaTexÅÅ°æ¼¼ÇÉ(1)
LaTexÅÅ°æ¼¼ÇÉ(1)_ÀíÑ§_¸ßµÈ½ÌÓý_½ÌÓý×¨Çø¡£ÏêÏ¸½éÉÜÁËLaTexµÄÅÅ°æ¼¼ÇÉµ¥ÐÐÅÅÓ¡,¹«Ê½±àºÅ°´ÕÂ½ÚÅÅÁÐ \documentclass[11pt,onecolumn,journal]{article} \usepackage...
Latex ±í¸ñÅÅ°æ¡ª¡ª¶àÐÐ¶àÁÐºÏ²¢
Latex ±í¸ñÅÅ°æ¡ª¡ª¶àÐÐ¶àÁÐºÏ²¢_¼ÆËã»úÈí¼þ¼°Ó¦ÓÃ_IT/¼ÆËã»ú_×¨Òµ×ÊÁÏ¡£ÊÊÓÃÓÚlatex±í¸ñÅÅ°æÖÐ¶àÐÐºÏ²¢ºÍ¶àÁÐºÏ²¢,Ö±½ÓÕ³Ìù´úÂë¼È¿ÉÒÔÊµÏÖ,ÄÚ¸½Ð§¹ûÍ¼¡£...
latexÂÛÎÄ²åÍ¼¼¼ÇÉ×Ü½á
latex »á¾¡Á¿Âú×ãÅÅÔÚÇ°ÃæµÄ¸¡¶¯¸ñÊ½,¾ÍÊÇ h-t-b Õâ¸öË³Ðò,ÈÃÅÅ°æµÄÐ§¹û¾¡Á¿...ÖªµÀµÄ·½·¨ÊÇÓÃ Psfrag ºê°ü,ÏêÏ¸·½ ·¨»¹ÊÇ²Î¼ûÄÇ±¾Êé,µ«ÊÇÎÒ¾õµÃÓÃÓ¢ÎÄ¾ÍÐÐ...
LaTeXÐÂÈË½Ì³Ì,Ò»Ð¡Ê±´ÓÍêÈ«Ä°Éúµ½»ù±¾ÈëÃÅ(Ò»ÖÖÅÅ°æÈí¼þ)
LaTeXÐÂÈË½Ì³Ì,Ò»Ð¡Ê±´ÓÍêÈ«Ä°Éúµ½»ù±¾ÈëÃÅ(Ò»ÖÖÅÅ°æÈí¼þ)_¼ÆËã»úÈí¼þ¼°Ó¦ÓÃ_IT...±ðÈËÏëÑ§ LaTeX,ºÃ,ÄãËæÊÖ¸øÈË ¼Ò¶ªÒ»¸ö¼¸°ÙÒ³µÄÓ¢ÎÄ Manual,ÏÔµÃ×Ô¼ººÜ¸ß¶Ë...
LatexÅÅ°æ¡ª¡ª¸´ÔÓ±í¸ñÖÆ×÷
LatexÅÅ°æ¡ª¡ª¸´ÔÓ±í¸ñÖÆ×÷_¼ÆËã»úÈí¼þ¼°Ó¦ÓÃ_IT/¼ÆËã»ú_×¨Òµ×ÊÁÏ ÔÝÎÞÆÀ¼Û|0ÈËÔÄ¶Á|0´ÎÏÂÔØ|¾Ù±¨ÎÄµµ LatexÅÅ°æ¡ª¡ª¸´ÔÓ±í¸ñÖÆ×÷_¼ÆËã»úÈí¼þ¼°Ó¦ÓÃ_IT/¼ÆËã»ú_×¨Òµ...
LaTeXÒ³Ãæ²¼¾Ö×¨Ìâ¡ª¡ª°æÃæÉèÖÃ
o LaTeX Ò³Ãæ²¼¾Ö×¨Ìâ¡ª¡ª°æÃæÉèÖÃ LaTeX Ò³Ãæ²¼¾Ö×¨Ìâ¡ª¡ªÒ³Ã¼ºÍÒ³½Å LaTeX Ò³Ãæ²¼¾Ö×¨Ìâ¡ª¡ª×°¶©Ïß LaTeX Ò³Ãæ²¼¾Ö×¨Ìâ¡ª¡ªÆäËüÍ¨Æª·¶Î§ÄÚµÄÅÅ°æ (Ò»)°æÃæÉèÖÃ...
LaTeXÒ³Ãæ²¼¾Ö×¨Ìâ
ÓÐÁ½¸öÑ¡Ïî:oneside,twoside ·Ö±ð±íÊ¾µ¥ÃæÅÅ°æºÍË«ÃæÅÅ°æ¡£ (1.3)°æÐ¾ÉèÖÃ¡£ (A)ÔÚ LaTeX ÖÐÓÐÈçÏÂ³¤¶ÈÀ´¿ØÖÆ°æÐ¾ÉèÖÃ: \paperwidth,\paperheight ·Ö±ð±íÊ¾...