SILVERRUSH. II. First Catalogs and Properties of ~2,000 Lya Emitters and Blobs at z~6-7 Identified over the 14-21 deg2 Sky
Takatoshi Shibuya, Masami Ouchi, Akira Konno, Ryo Higuchi, Yuichi Harikane, Yoshiaki Ono, Kazuhiro Shimasaku, Yoshiaki Taniguchi, Masakazu A. R. Kobayashi, Masaru Kajisawa, Tohru Nagao, Hisanori Furusawa, Tomotsugu Goto, Nobunari Kashikawa, Yutaka Komiyama, Haruka Kusakabe, Chien-Hsiu Lee, Rieko Momose, Kimihiko Nakajima, Masayuki Tanaka, Shiang-Yu Wang, Suraphong Yuma
aa r X i v : . [ a s t r o - ph . GA ] S e p Publ. Astron. Soc. Japan (2014) 00(0), 1–19doi: 10.1093/pasj/xxx000 SILVERRUSH. II. First Catalogs and Properties of ∼ , Ly α Emitters and Blobs at z ∼ − Identified over the − deg Sky
Takatoshi Shibuya , Masami Ouchi , Akira Konno , Ryo Higuchi ,Yuichi Harikane , Yoshiaki Ono , Kazuhiro Shimasaku , YoshiakiTaniguchi , Masakazu A. R. Kobayashi , Masaru Kajisawa , TohruNagao , Hisanori Furusawa , Tomotsugu Goto , Nobunari Kashikawa ,Yutaka Komiyama Haruka Kusakabe , Chien-Hsiu Lee , RiekoMomose , Kimihiko Nakajima , Masayuki Tanaka , Shiang-Yu Wang ,and Suraphong Yuma Institute for Cosmic Ray Research, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa,Chiba 277-8582, Japan Kavli Institute for the Physics and Mathematics of the Universe (Kavli IPMU, WPI), Universityof Tokyo, Kashiwa, Chiba 277-8583, Japan Department of Astronomy, Graduate School of Science, The University of Tokyo, 7-3-1Hongo, Bunkyo, Tokyo 113-0033, Japan Department of Physics, Graduate School of Science, The University of Tokyo, 7-3-1 Hongo,Bunkyo, Tokyo 113-0033, Japan Research Center for the Early Universe, Graduate School of Science, The University ofTokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan The Open University of Japan, Wakaba 2-11, Mihama-ku, Chiba 261-8586, Japan Faculty of Natural Sciences, National Institute of Technology, Kure College, 2-2-11Agaminami, Kure, Hiroshima 737-8506, Japan Research Center for Space and Cosmic Evolution, Ehime University, Bunkyo-cho 2-5,Matsuyama 790-8577, Japan National Astronomical Observatory, Mitaka, Tokyo 181-8588, Japan Institute of Astronomy, National Tsing Hua University, 101 Section 2, Kuang-Fu Road,Hsinchu 30013, Taiwan The Graduate University for Advanced Studies (SOKENDAI), 2-21-1 Osawa, Mitaka, Tokyo181-8588 Subaru Telescope, NAOJ, 650 N Aohoku Pl., Hilo, HI 96720, USA European Southern Observatory, Karl-Schwarzschild-Str. 2, D-85748, Garching beiMunchen, Germany Academia Sinica, Institute of Astronomy and Astrophysics, 11F of AS/NTUAstronomy-Mathematics Building, No.1, Sec. 4, Roosevelt Rd, Taipei 10617, Taiwan Department of Physics, Faculty of Science, Mahidol University, Bangkok 10400, Thailand ‡ Based on data obtained with the Subaru Telescope. The Subaru Telescope is operated by c (cid:13) Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 the National Astronomical Observatory of Japan. ∗ E-mail: [email protected]
Received h reception date i ; Accepted h acception date i Abstract
We present an unprecedentedly large catalog consisting of 2,230 > ∼ L ∗ Ly α emitters (LAEs)at z = 5 . and . on the . and . deg sky, respectively, that are identified by theSILVERRUSH program with the first narrowband imaging data of the Hyper Suprime-Cam(HSC) survey. We confirm that the LAE catalog is reliable on the basis of 96 LAEs whosespectroscopic redshifts are already determined by this program and the previous studies. Thiscatalogue is also available on-line. Based on this catalogue, we derive the rest-frame Ly α equivalent-width distributions of LAEs at z ≃ . − . that are reasonably explained by the ex-ponential profiles with the scale lengths of ≃ − ˚A, showing no significant evolution from z ≃ . to z ≃ . . We find that LAEs with a large equivalent width (LEW) of > ˚A are can-didates of young-metal poor galaxies and AGNs. We also find that the fraction of LEW LAEsto all ones is % and % at z ≃ . and z ≃ . , respectively. Our LAE catalog includes 11Ly α blobs (LABs) that are LAEs with spatially extended Ly α emission whose profile is clearlydistinguished from those of stellar objects at the > ∼ σ level. The number density of the LABs at z = 6 − is ∼ − − − Mpc − , being ∼ − times lower than those claimed for LABs at z ≃ − , suggestive of disappearing LABs at z > ∼ , albeit with the different selection methodsand criteria for the low and high- z LABs.
Key words: early universe — galaxies: formation — galaxies: high-redshift Ly α Emitters (LAEs) are one of important populations of high- z star-forming galaxies in the paradigm of the galaxy formationand evolution. Such galaxies are thought to be typically young(an order of Myr; e.g., Finkelstein et al. 2007; Gawiseret al. 2007; Finkelstein et al. 2007), compact (an effectiveradius of < kpc; e.g., Taniguchi et al. 2009; Bond et al.2012), less-massive (a stellar mass of − M ⊙ ; e.g., Onoet al. 2010; Guaita et al. 2011), metal-poor ( ≃ . of the so-lar metallicity; e.g., Nakajima et al. 2012; Nakajima et al.2013; Nakajima & Ouchi 2014; Kojima et al. 2016), less-dustythan Lyman break galaxies (e.g., Blanc et al. 2011; Kusakabeet al. 2015), and a possible progenitor of Milky Way massgalaxies (e.g., Dressler et al. 2011). In addition, LAEs areused to probe the cosmic reionizaiton, because ionizing pho-tons escaped from a large number of massive stars formedin LAEs contribute to the ionization of intergalactic medium(IGM; e.g., Rhoads & Malhotra 2001; Malhotra & Rhoads2006; Shimasaku et al. 2006; Kashikawa et al. 2006; Ouchiet al. 2008; Ouchi et al. 2010; Cowie et al. 2010; Hu et al.2010; Kashikawa et al. 2011; Shibuya et al. 2012; Konno et al.2014; Matthee et al. 2015; Matthee et al. 2015; Ota et al.2017; Zheng et al. 2017).LAEs have been surveyed by imaging observations withdedicated narrow-band (NB) filters for a prominent redshifted Ly α emission (e.g., Ajiki et al. 2002; Malhotra & Rhoads2004; Kodaira et al. 2003; Taniguchi et al. 2005; Gronwall et al.2007; Erb et al. 2011; Ciardullo et al. 2012). In large LAE sam-ple constructed by the NB observations, two rare Ly α -emittingpopulations have been identified: large equivalent width (LEW)LAEs, and spatially extended Ly α LAEs, Ly α blobs (LABs).LEW LAEs are objects with a large Ly α equivalent width(EW) of > ∼ ˚A which are not reproduced with the normalSalpeter stellar initial mass function (e.g., Malhotra & Rhoads2002). Such an LEW is expected to be originated from compli-cated physical processes such as (i) photoionization by youngand/or low-metallicity star-formation, (ii) photoionization byactive galactic nucleus (AGN), (iii) photoionization by exter-nal UV sources (QSO fluorescence), (iv) collisional excitationdue to strong outflows (shock heating), (v) collisional excitationdue to gas inflows (gravitational cooling), and (vi) clumpy ISM(see e.g., Hashimoto et al. 2017). The highly-complex radia-tive transfer of Ly α in the interstellar medium (ISM) makes itdifficult to understand the Ly α emitting mechanism (Neufeld1991; Hansen & Oh 2006; Finkelstein et al. 2008; Laursenet al. 2013; Laursen et al. 2009; Laursen & Sommer-Larsen2007; Zheng et al. 2010; Yajima et al. 2012; Duval et al.2013; Zheng & Wallace 2013).LABs are spatially extended Ly α gaseous nebulae in thehigh- z Universe (e.g., Steidel et al. 2000; Matsuda et al. ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 ≃ − kpc) are also explained byseveral mechanisms: (1) resonant scattering of Ly α photonsemitted from central sources in dense and extended neutral hy-drogen clouds (e.g., Hayes et al. 2011), (2) cooling radiationfrom gravitationally heated gas in collapsed halos (e.g., Haimanet al. 2000), (3) shock heating by galactic superwind originatedfrom starbursts and/or AGN activity (e.g., Taniguchi & Shioya2000), (4) galaxy major mergers (e.g., Yajima et al. 2013), and(5) photoionization by external UV sources (QSO fluorescence;e.g., Cantalupo et al. 2005). Moreover, LABs have been oftendiscovered in over-density regions at z ≃ − (e.g., Yang et al.2009; Yang et al. 2010; Matsuda et al. 2011). Thus, such LABscould be closely related to the galaxy environments, and mightbe liked to the formation mechanisms of central massive galax-ies in galaxy protoclusters.During the last decades, Suprime-Cam (SCam) on theSubaru telescope has led the world on identifying such rareLy α -emitting populations at z > ∼ (LEW LAEs; e.g., Nagaoet al. 2008; Kashikawa et al. 2012; LABs; e.g., Ouchi et al.2009; Sobral et al. 2015). However, the formation mechanismsof these rare Ly α -emitting populations are still controversialdue to the small statistics. While LEW LAEs and LABs at z ≃ − have been studied intensively with a sample of > ∼ sources, only a few sources have been found so far at z > ∼ .Large-area NB data are required to carry out a statistical studyon LEW LAEs and LABs at z > ∼ .In March 2014, the Subaru telescope has started a large-area NB survey using a new wide field of view (FoV) cam-era, Hyper Suprime-Cam (HSC) in a Subaru strategic program(SSP; Aihara et al. 2017b). In the five-year project, HSCequipped with four NB filters of NB , NB , NB ,and NB will survey for LAEs at z ≃ . , 5.7, 6.6, and7.3, respectively. The HSC SSP NB survey data consist oftwo layers; Ultradeep (UD), and Deep (D), covering 2 fields(UD-COSMOS, UD-SXDS), and 4 fields (D-COSMOS, D-SXDS, D-DEEP2-3, D-ELAIS-N1), respectively. The NB , NB , and NB images will be taken for the UD fields. The NB , NB , and NB observations will be conducted in15 HSC-pointing D fields.Using the large HSC NB data complemented by optical andNIR spectroscopic observations, we launch a research projectfor Ly α -emitting objects: Systematic Identification of LAEs forVisible Exploration and Reionization Research Using SubaruHSC (SILVERRUSH) . The large LAE samples provided bySILVERRUSH enable us to investigate e.g., LAE clustering(Ouchi et al. 2017), LEW LAEs and LABs (this work), spec- T h r o u g h p u t l [Å] z Ly a N B N B i z y Fig. 1.
Filter transmission curves of the NB and BB filters. The red and bluecurves represent the
NB921 and
NB816 filters, respectively. The red and blueticks show the NB central wavelengths with the same color coding as for theNB filter transmission curves. The black solid curves indicate the i, z , and y -band filters, from left to right. The gray line denotes the OH sky lines.The bandpass of these NB and BB filters corresponds to the area-weightedmean transmission curves . The transmission curves are derived by takinginto account 1) the quantum efficiency of CCD, the transmittance of 2) thedewar window and 3) the HSC primary focus unit (POpt2), 4) the reflectivityof the primary mirror, and 5) the sky transparency (see Aihara et al. 2017a).The upper x -axis corresponds to the redshift of Ly α . troscopic properties of bright LAEs (Shibuya et al. 2017b),Ly α luminosity functions (Konno et al. 2017), and LAE over-density (R. Higuchi et al. in preparation). The LAE sur-vey strategy is given by Ouchi et al. (2017). This program isone of the twin programs. Another program is the study fordropouts, Great Optically Luminous Dropout Research UsingSubaru HSC (GOLDRUSH), that is detailed in Ono et al.(2017), Harikane et al. (2017), and Toshikawa et al. (2017).This is the second paper in the SILVERRUSH project. Inthis paper, we present LAE selection processes and machine-readable catalogs of the LAE candidates at z ≃ . − . . Usingthe large LAE sample obtained with the first HSC NB data, weexamine the redshift evolutions of Ly α EW distributions andLAB number density. This paper has the following structure. InSection 2, we describe the details of the SSP HSC data. Section3 presents the LAE selection processes. In Section 4, we checkthe reliability of our LAE selection. Section 5 presents Ly α EWdistributions and LABs at z ≃ − . In Section 6, we discussthe physical origins of LEW LAEs and LABs. We summarizeour findings in Section 7.Throughout this page, we adopt the concordance cosmologywith (Ω m , Ω Λ , h ) = (0 . , . , . (Planck Collaboration et al.2016). All magnitudes are given in the AB system (Oke & Gunn1983). Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0
HSC S16A Data (1) SQL Selection
HSC LAE
Forced
Catalog
HSC LAE
ALL
Catalog
NB816 < NB8165sigAND i - NB816 > 1.2AND g > g3sigAND [(r < r3sig AND r - i > 1.0) OR (r > r3sig)] (2) Visual Inspection ( i ) Magnitude & Color Selection
AND detect_is_tract_inner = TrueAND detect_is_patch_inner = TrueAND NB_countinputs >= 3AND [grizy]flags_pixel_edge = FalseAND [grizy]flags_pixel_interpolated_center = FalseAND [grizy]flags_pixel_saturated_center = FalseAND [grizy NB]flags_pixel_cr_center = FalseAND [grizy NB]flags_pixel_bad = False ( ii ) Flag and other parameters
YES
Spurious sourcese.g., Cosmic rays Compact stellar objects Cross-talks Satellite trails Diffuse halo near bright stars (Object images) (4)
Forced
Selection (3) Rejection with multi-epoch images + z~5.7 (forced or unforced color) NB921 < NB9215sigAND z - NB921 > 1.0AND g > g3sigAND r > r3sigAND [(z < z3sig AND i - z > 1.3) OR (z > z3sig)] z~6.6 (forced or unforced color)
YESYESYES NO
AND [grizy NB]flags_pixel_bright_object_center = FalseAND [grizy NB]flags_pixel_bright_object_any = False NO Variables e.g. supernovaeMoving objects
NB816 < NB8165sigAND i - NB816 > 1.2AND g > g2sigAND [(r < r3sig AND r - i > 1.0) OR (r > r3sig)] z~5.7 (forced color)
NB921 < NB9215sigAND z - NB921 > 1.8AND g > g2sigAND r > r2sigAND [(z < z3sig AND i - z > 1.3) OR (z > z3sig)] z~6.6 (forced color)
Fig. 2.
Flow chart of the HSC LAE selection process. See Section 3 for more details.
We use the HSC SSP S16A data products of g, r, i, z , and y broadband (BB; Kawanomoto 2017), NB and NB (Ouchi et al. 2017) images that are obtained in 2014-2016. Itshould be noted that this HSC SSP S16A data is significantlylarger than the one of the first-data release in Aihara et al.(2017a). The NB ( NB ) filter has a central wavelength of λ c = 9215 ˚A ( ˚A) and an FWHM of ∆ λ = 135 ˚A (113 ˚A),all of which are the area-weighted mean values. The NB and NB filters trace the redshifted Ly α emission lines at z = 6 . ± . and z = 5 . ± . , respectively. The NBfilter transmission curves are shown in Figure 1. The centralwavelength, FWHM, and the bandpass shape for these NB fil-ters are almost uniform over the HSC FoV. The deviation of the λ c and FWHM values are typically within ≃ . % and ≃ %, ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 Fig. 3.
Multi-band cutout images of our example LAEs and spurious sources. (a) LAEs at z ≃ . (top) and z ≃ . (bottom) in the forced LAE catalog. (b)LAEs at z ≃ . (top) and z ≃ . (bottom) in the unforced catalog. In the rightmost cutout images, the yellow solid and cyan dashed circles represent thecentral positions of the unforced LAEs in the NB and BB images, respectively. The diameters of the yellow solid and dashed circles in the cutout images ofthe unforced
LAEs are ′′ and . ′′ , respectively. (c) Spurious sources with an NB magnitude-excess similar to that of LAE candidates (four panel sets at thetop), 1: variable (e.g., supernova); 2: cosmic ray; 3: cross-talk artifact; 4: moving object (e.g., asteroids) and corresponding multi-epoch images (four panelsets at the bottom). The image size is ′′ × ′′ for the LAEs and spurious sources. respectively. Thus, we use the area-weighted mean transmis-sion curves in this study. The detailed specifications of theseNB filters are given in Ouchi et al. (2017).Table 1 summarizes the survey areas, exposure time, anddepth of the HSC SSP S16A NB data. The current HSC SSPS16A NB data covers UD-COSMOS, UD-SXDS, D-COSMOS,D-DEEP2-3, D-ELAIS-N1 for z ≃ . , and UD-COSMOS,UD-SXDS, D-DEEP2-3, D-ELAIS-N1 for z ≃ . . The ef-fective survey areas of the NB and NB images are . and . arcmin , corresponding to the survey volumesof ≃ . × and ≃ . × Mpc , respectively. The areaof these HSC NB fields are covered by the observations of allthe BB filters. The typical limiting magnitudes of BB filters are g ≃ . , r ≃ . , r ≃ . , z ≃ . , and y ≃ . ( g ≃ . , r ≃ . , r ≃ . , z ≃ . , and y ≃ . ) in a . ′′ apertureat σ for the UD (D) fields. The FWHM size of point spreadfunction in the HSC images is typically ≃ . ′′ (Aihara et al.2017a).The HSC images were reduced with the HSC pipeline, hscPipe unforced and forced . The unforced photometry is a method to perform measurements ofcoordinates, shapes, and fluxes individually in each band imagefor an object. The forced photometry is a method to carry outphotometry by fixing centroid and shape determined in a refer-ence band and applying them to all the other bands. The algo-rithm of the forced detection and photometry is similar to the double-image mode of SExtractor (Bertin & Arnouts 1996)that are used in most of the previous studies for high- z galaxies.According to which depends on magnitudes, S/N , positions,and profiles for detected sources, one of the BB and NB filter isregarded as a reference band. For merging the catalogs of eachband, the object matching radius is not a specific value whichdepends on an area of regions with a > σ sky noise level. Werefer the detailed algorithm to choose the reference filter and Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 filter priority to Bosch et al. (2017).In the hscPipe detection and photometry, an NB filter isbasically chosen as a reference band for the NB-bright and BB-faint sources such as LAEs. However, a BB filter is used asa reference band in the case that sources are bright in the BBimage. The current version of hscPipe has not implementedthe NB-reference forced photometry for BB-bright sources. Inthis specification, there is a possibility that we miss BB-brightsources with a spatial offset between centroids of BB and NBby using only the forced photometry. Thus, we combine the unforced or forced photometry for BB − NB colors to iden-tify such BB-bright objects with a spatial offset between cen-troids of BB and NB (e.g., Shibuya et al. 2014a). See Section 3for details of the LAE selection criteria.We use cmodel magnitudes for estimating total magnitudesof sources. The cmodel magnitude is a weighted combinationof exponential and de Vaucouleurs fits to light profiles of eachobject. The detailed algorithm of the cmodel photometry arepresented in Bosch et al. (2017). To measure the S/N values forsource detections, we use . ′′ -diameter aperture magnitudes. Using the HSC data, we perform a selection for LAEs at z ≃ . and ≃ . . Basically, we select objects showing a significantflux excess in the NB images and a spectral break at the wave-length of redshifted Ly α emission. In this study, we createtwo LAE catalogs: HSC LAE
ALL (forced+unforced) cata-log and
HSC LAE forced catalog . The HSC LAE
ALL catalogis constructed in a combination of the forced and unforced photometry. We use this HSC LAE
ALL catalog for identify-ing objects with a spatial offset between centroids of BB andNB (see Section 2). On the other hand, the HSC LAE forced catalog consists of LAEs meeting only the selection criteria ofthe forced photometry. We use this HSC LAE forced catalogfor statistical studies for LAEs (e.g., Ly α LFs). The HSC LAE forced catalog is a subsample of the
ALL one. Figure 2 showsthe flow chart of the LAE selection process. We carry out thefollowing processes: (1) SQL selection, (2) visual inspectionsfor the object images, (3) rejections of variable and moving ob-jects with the multi-epoch images, and (4) forced selection.The details are described as below.(1)
SQL selection:
We retrieve detection and photometric cat-alogs from postgreSQL database tables. Using SQL scripts,we select objects meeting the following criteria of (i) mag-nitude and color selections and (ii) hscPipe parameters andflags.(i)
Magnitude and color selection:
To identify objectswith an NB magnitude excess in the HSC catalog, weapply the magnitude and color selection criteria that aresimilar to e.g., Ouchi et al. 2008; Ouchi et al. 2010: NB921 apfrc < NB921 σ && ( g frc > g σ || g apfrc > g σ )&& ( r frc > r σ || r apfrc > r σ )&& ( z frc − NB921 frc > . || z unf − NB921 unf > . { [( z frc < z σ || z apfrc < z σ )&& ( i frc − z frc > . || i unf − z unf > . || ( z frc > z σ || z apfrc > z σ ) } , (1)for z ≃ . , and, NB816 apfrc < NB816 σ && ( g frc > g σ || g apfrc > g σ )&& ( i frc − NB816 frc > . || i unf − NB816 unf > . { [( r frc < r σ || r apfrc < r σ )&& ( r frc − i frc > . || r unf − i unf > . || ( r frc > r σ || r apfrc > r σ ) } , (2)for z ≃ . , where the indices of frc and unf representthe forced and unforced photometry, respectively. Thesubscript of σ ( σ ) indicates the σ ( σ ) limiting mag-nitude for a given filter. The values with and without asuperscript of ap indicate the aperture and total magni-tudes, respectively. These magnitudes are derived withthe hscPipe software (see Section 2; Bosch et al. 2017).The limits of the i − NB and z − NB colors arethe same as those of Ouchi et al. (2008) and Ouchi et al.(2010), respectively. To exploit the survey capability ofHSC identifying rare objects, we use the σ g and r lim-iting magnitude (instead of the value of σ used in Ouchiet al. 2008) for the criteria of Lyman break off-band non-detection. In the process (4), we replace σ with σ forthe g and r magnitude criteria for the consistency withthe previous studies.Note that we do not apply the flags pixel bright object [center/any] mask-ing to the LAE ALL catalog in order to maximize LAEtargets for future follow-up observations (Aihara et al.2017a). These flags for the object masking are used inthe process (4).(ii)
Parameters and flags:
Similar to Ono et al. (2017),we set several hscPipe parameters and flags in the HSCcatalog to exclude e.g., blended sources, and objects af-fected by saturated pixels, and nearby bright source ha-los. We also mask regions where exposure times are rel-atively short by using the countinputs parameter, N c ,which denotes the number of exposures at a source posi-tion for a given filter. Table 2 summarizes the values andbrief explanations of the hscPipe parameters and flagsused for our LAE selection. The full details of these pa-rameters and flags are presented in Aihara et al. (2017a).To search for LAEs in large areas of the HSC fields, wedo not apply the countinputs parameter to the BB im-ages. ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 Table 1.
Properties of the HSC SSP S16A NB Data
Field R.A. Dec. Area T exp m lim (5 σ, . ′′ φ ) N LAE , ALL N LAE , F (J2000) (J2000) (deg ) (hour) (mag)(1) (2) (3) (4) (5) (6) (7) (8) NB ( z ≃ . )UD-COSMOS 10:00:28 + − + a a D-DEEP2-3 23:30:22 − + Total — — 21.2 24.00 — 1153 269 NB ( z ≃ . )UD-COSMOS 10:00:28 + − − + Total — — 13.8 11.25 — 1077 776(1) Field.(2) Right ascension.(3) Declination.(4) Survey area with the HSC SQL parameters in Table 2.(5) Total exposure time of the NB imaging observation.(6) Limiting magnitude of the NB image defined by a σ sky noise in a . ′′ diameter circular aperture.(7) Number of the LAE candidates in the ALL ( unforced + forced ) catalog.(8) Number of the LAE candidates in the forced catalog. a The value of N LAE , ALL ( N LAE , F ) includes 30 (7) LAEs selected in UD-COSMOS. Table 2.
HSC SQL Parameters and Flags for Our LAE Selection
Parameter or Flag Value Band Comment detect is tract inner True — Object is in an inner region of a tract andnot in the overlapping region with adjacent tracts detect is patch inner True — Object is in an inner region of a patch andnot in the overlapping region with adjacent patches countinputs > = 3 NB Number of visits at a source position for a given filter. flags pixel edge False grizy , NB Locate within images flags pixel interpolated center False grizy , NB None of the central × pixels of an object is interpolated flags pixel saturated center False grizy , NB None of the central × pixels of an object is saturated flags pixel cr center False grizy , NB None of the central × pixels of an object is masked as cosmic ray flags pixel bad False grizy , NB None of the pixels in the footprint of an object is labelled as badThe number of objects selected in this process is n SQL ≃ , .(2) Visual inspections for object images:
To exclude cosmicrays, cross-talks, compact stellar objects, and artificial dif-fuse objects, we perform visual inspections for the BB andNB images of all the objects selected in the process (1). Mostspurious sources are diffuse components near bright starsand extended nearby galaxies. The hscPipe software con-ducts the cmodel fit to broad light profiles of such diffusesources in the NB images, which enhances the BB − NB colors. For this reason, the samples constructed in the cur-rent SQL selection are contaminated by many diffuse com-ponents. Due to the clear difference of the appearance be-tween LAE candidates and diffuse components, such spu-rious sources can be easily excluded through the visual in-spections. The number of objects selected in this process is n vis ≃ , .The visual inspection processes are mainly conducted by oneof the authors. For the reliability check, four authors in thispaper have individually carried out such visual inspections Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 z − N B [ m a g ] NB921 [mag] z~6.6
UD024 22 23 24 25 D
Forced Unforced
22 23 24 25 i − N B [ m a g ] NB816 [mag] z~5.7
UD024 22 23 24 25 D
Forced Unforced
22 23 24 25
Fig. 4. ( Top ) Color of z − NB as a function of NB magnitude for the LAEs at z ≃ . in the UD (left) and D (right) fields. The filled red and openmagenta circles denote the forced and unforced LAEs, respectively. For the LAEs undetected in the z -band images, the z -band magnitudes are replacedwith the σ limiting magnitudes. The x-axis denotes the forced ( unforced ) z − NB colors for the forced ( unforced ) LAEs. The horizontal dashed anddotted line shows the color criteria of z − NB > . and z − NB > . , respectively. The gray dots present objects detected in the NB images.The solid curves show the σ error tracks of z − NB color for each field. The σ error tracks are derived by Equation 3. ( Bottom ) Color of i − NB asa function of NB magnitude for the LAEs at z ≃ . . The definitions of symbols, curves, and lines are the same as those of the top panels. for ≃ , objects in the UD-COSMOS NB fields, andcompare the results of the LAE selection. The difference inthe number of selected LAEs is within ± objects. Thus, wedo not find a large difference in our visual inspection results.(3) Rejection of variable and moving objects with multi- epoch images:
We exclude variable and moving objectssuch as supernovae, AGNs, satellite trails, and asteroids us-ing multi-epoch NB images. The NB images were typicallytaken a few months - years after the BB imaging observa-tions. For this reason, there is a possibility that sources with ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 ③(cid:0)✁✂✁❯✄☎✆✝✄✆✲✞✲✟✲✠✲✡✲☛ ❯✄☎☞✌✆✍✌✆❧✎✏ S ❬✑✒✓✔✕✖✗✘✒✗✙✚✖✛✜✷ ❪ ✄☎❉✢✣✤✆✥✦✲✞✲✟✲✠✲✡✲☛ ✄☎✄❉❉✧★✩✡✠ ✡✟ ✡✞◆✪✫✬✭ ✮✯✰✱✳✄☎☞✌✆✍✌✆✲✞✲✟✲✠✲✡✲☛ ✡✠ ✡✟ ✡✞ ③(cid:0)✴✂✵❯✄☎✆✝✄✆✲✞✲✟✲✠✲✡✲☛ ❯✄☎☞✌✆✍✌✆◆✪✶✭✸ ✮✯✰✱✳✄☎❉✢✣✤✆✥✦✲✞✲✟✲✠✲✡✲☛ ✡✠ ✡✟ ✡✞ ✄☎✄❉❉✧★✩❋✹✺✻✼✽❆✾✾❖✿❀❁❂❃❄❅❇❈❄✡✠ ✡✟ ✡✞ Fig. 5.
Surface number density (SND) of the HSC LAEs at z ≃ . (five panels at the left) and ≃ . (four panels at the right) in each UD and D field. The filledred and open magenta circles indicate the LAEs in the forced and ALL catalog, respectively. The error bars are given by Poisson statistics from the numberof LAEs. The gray crosses represent the LAEs in Ouchi et al. (2010) for z ≃ . and Ouchi et al. (2008) for z ≃ . . The data points of the gray crosses areidentical in all the fields for each redshift. The SND slight declines in the HSC LAEs at NB > ∼ . mag would be originated from the incompleteness of theLAE detection and selection. The completeness-corrected SNDs are presented in Konno et al. (2017). The data points of the HSC LAEs are slightly shiftedalong x-axis for clarity. an NB flux excess are variable or moving objects which hap-pened to enhance the luminosities during the NB imagingobservations.The NB images are created by coadding ≃ − and ≃ − frames of minute exposures for the current HSC UDand D data, respectively. Using the multi-epoch images, weautomatically remove the variable and moving objects as fol-lows. First, we measure the flux for individual epoch images, f , for each object. Next, we obtain an average, f ave ,and a standard deviation, σ epoch , from a set of the f values after a σ flux clipping. Finally, we discard an objecthaving at least a multi-epoch image with a significantly large f value of f ≥ f ave + A epoch × σ epoch . Here wetune the A epoch factor based on the depth of the NB fields.The A epoch value is typically ≃ . − . . Figure 3 showsexamples of the spurious sources.We also perform visual inspections for multi-epoch images to remove contaminants which are not excluded in the auto-matic rejection above. We refer the remaining objects afterthis process as the LAE ALL catalog.(4)
Forced selection:
In the selection criteria of Equations (1) and (2), the HSCLAE
ALL catalog is obtained in the combination of the forced and unforced colors. In this process, we se-lect LAEs only with the forced color excess to create the forced
LAE subsamples from the HSC LAE
ALL catalog.In addition, the σ limit is replaced with σ for the criteriaof g and r band non-detections.Here we also adopt a new stringent color criterion of z − NB > . for z ≃ . LAEs. Due to the difference ofthe z band transmission curves between SCam and HSC,the criterion of z − NB > . in Equation (1) do not al-low us to select LAEs whose EW , Ly α is similar to thoseof previous SCam studies. The BB − NB color crite- Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 lo g S [ N/ . m a g / a m in ] NB921 [mag] z~6.6 -5-4-3-2-1 23 24 25
NB816 [mag] z~5.7
UD-SXDSUD-COSMOSD-ELAISN1D-DEEP23D-COSMOSOuchi+08,10
23 24 25
Fig. 6.
Surface number density (SND) as a function of NB magnitude for the LAEs at z ∼ . (left) and ∼ . (right) in the HSC LAE forced catalog. Thecolored symbols denote the LAEs in each UD and D field (green circles: UD-SXDS; magenta squares: UD-COSMOS; cyan triangles: D-ELAIS-N1; light-redinverse-triangles: D-DEEP2-3; orange diamonds: D-COSMOS). The error bars are given by Poisson statistics from the number of LAEs. The gray crossesrepresent the LAEs in Ouchi et al. (2010) for z ≃ . and Ouchi et al. (2008) for z ≃ . . The SND slight declines in the HSC LAEs at NB > ∼ . mag wouldbe originated from the incompleteness of the LAE selection. The completeness-corrected SNDs are presented in Konno et al. (2017). The data points of theHSC LAEs are slightly shifted along x-axis for clarity. Table 3.
Photometric properties of example LAE candidates
Object ID NB g r i z y (mag) (mag) (mag) (mag) (mag) (mag)(1) (2) (3) (4) (5) (6) (7)UD-SXDS ( NB )HSC J021601 − . ± .
10 26 . ± .
45 27 . ± .
62 26 . ± .
63 25 . ± .
31 25 . ± . HSC J021754 − . ± . > . > . > . . ± .
57 25 . ± . HSC J021702 − . ± . > . > . > . > . > . HSC J021638 − . ± . > . > . > . . ± . > . HSC J021609 − . ± .
26 27 . ± .
72 27 . ± . > . . ± . > . UD-COSMOS ( NB )HSC J100243 + . ± . > . > . . ± . > . > . HSC J100239 + . ± . > . > . . ± .
64 26 . ± . > . HSC J100243 + . ± . > . > . > . > . > . HSC J095936 + . ± . > . > . > . > . > . HSC J100245 + . ± . > . > . > . > . > . (1) Object ID.(2)-(7) Total magnitude of NB -, g -, r -, i -, z , and y -bands.The 2 σ limits of the total magnitudes for the undetected bands.(The complete machine-readable catalogs will be available on our project webpage athttp://cos.icrr.u-tokyo.ac.jp/rush.html.)ria in in the forced selection correspond to the rest-frameLy α EW of EW , Ly α > ˚A and > ˚A for z ≃ . and z ≃ . LAEs, respectively. These EW , Ly α limits are com-parable to those of the previous SCam studies (e.g., Ouchiet al. 2010). The relation between EW , Ly α and BB − NB colors is described in Konno et al. (2017) in details.Moreover, we remove the objects in masked regions de-fined by the flags pixel bright object [center/any] parameters (Aihara et al. 2017a).We refer the set of the remaining objects after this process as ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 the forced LAE catalog . This forced
LAE catalog is usedfor studies on LAE statistics such as measurements of Ly α EW scale lengths.The LAE candidates selected in this forced selection arereferred to as the forced
LAEs. On the other hand, we re-fer to the remaining LAE candidates in the HSC LAE
ALL catalog as the unforced
LAEs. The examples of forced and unforced
LAEs are shown in Figure 3. As shown inthe top-right panels of Figure 3, the unforced
LAEs have a ≃ ′′ − ′′ spatial offset between centroids in NB and BB.In total, we identify , and , LAE candidates inthe HSC LAE
ALL and forced catalogs, respectively. Table1 presents the numbers of LAE candidates in each field.The machine-readable catalogs of all the LAE candidateswill be provided on our project webpage at http://cos.icrr.u-tokyo.ac.jp/rush.html. The photometric properties of exampleLAE candidates are shown in Table 3.As shown in Table 1, the number of z ≃ . LAEs in D-DEEP2-3 appears to be large compared to that of the other z ≃ . fields. This may be because the seeing of the NB images of D-DEEP2-3 is better than that of the other z ≃ . fields. Similarly, the small number of z ≃ . LAEs in UD-SXDS may be affected by the seeing size. The number densityof LAEs is discussed in the next section. Note that edge re-gions of UD-COSMOS is overlapped with a flanking field, D-COSMOS (Aihara et al. 2017b). We find that 30 (7) LAEs inUD-COSMOS are also selected in the HSC LAE
ALL ( forced )sample of D-COSMOS. To analyze the D field independently inthe following sections, we include the overlapped LAEs in theD-COSMOS sample.Figure 4 shows the color-magnitude diagrams for the LAEcandidates. The solid curves in the color magnitude diagramsindicate the σ errors of BB − NB color as a function of theNB flux, f NB , given by ± σ BB − NB = − . (cid:0) ∓ p f σ NB + f σ BB f NB (cid:1) , (3)where f σ NB and f σ BB are the σ flux error in the z and NB ( i and NB ) bands for z ≃ . ( z ≃ . ), respec-tively. As shown in Figure 4, the LAE candidates have a signif-icant NB magnitude excess. Here we check the reliability of our LAE selection.
We have conducted optical spectroscopic observations withSubaru/FOCAS and Magellan/LDSS3 for 18 bright LAE can-didates with NB < ∼ mag. In these observations, we have confirmed 13 LAEs. By investigating our spectroscopic cata-log of Magellan/IMACS, we also spectroscopically identify 8LAEs with NB < ∼ mag. In addition, we find that 75 LAEsare spectroscopically confirmed in literature (Murayama et al.2007; Ouchi et al. 2008; Taniguchi et al. 2009; Ouchi et al.2010; Mallery et al. 2012; Sobral et al. 2015; Higuchi et al.in preparation). In total, 96 LAEs have been confirmed in ourspectroscopy and previous studies. Using the spectroscopicsample whose number of observed LAEs is known, we esti-mate the contamination rate to be ≃ − %. The details of thespectroscopic observations and contamination rates are given byShibuya et al. (2017b). Figure 5 shows the surface number density (SND) of our LAEcandidates and LAEs identified in previous Subaru/SCam NBsurveys, SCam LAEs (e.g., Ouchi et al. 2008; Ouchi et al.2010). We find that the SNDs of the forced
LAEs are compa-rable to those of SCam LAEs. On the other hand, the SNDs of unforced
LAEs at z ≃ . are higher than that of SCam LAEs.The high SND of the unforced LAEs is mainly caused by thecolor criterion for the HSC LAE
ALL catalog of z − NB > . that is less stringent than z − NB > . (see Section 3).We also identify SND humps of our forced LAEs at z ≃ . atthe bright-end of NB ≃ mag in UD-COSMOS. The presenceof such a SND hump has been reported by z ≃ . LAE studies(e.g., Matthee et al. 2015). The significance of the bright-endhump existence in Ly α LFs is ≃ σ , which are discussed inKonno et al. (2017). The slight declines in SNDs at a faint NBmagnitude of NB > ∼ . mag would be originated from the in-completeness of the LAE detection and selection. Konno et al.(2017) present the SND corrected for the incompleteness.Figure 6 compiles the SNDs of all the HSC UD and D fields.We find that our SNDs show a small field-to-field variation, buttypically follow those of the SCam LAEs. The UD-SXDS field has been observed previously by SCamequipped with the NB and NB filters (Ouchi et al.2008; Ouchi et al. 2010). We compare the catalogs of our se-lected HSC LAE candidates and SCam LAEs, and calculate theobject matching rates as a function of NB magnitudes. The ob-ject matching radius is ′′ . The object matching rate betweenthe HSC LAEs and SCam LAEs is ≃ % at a bright NB mag-nitudes of < ∼ mag. The high object matching rate indicatesthat we adequately identify LAEs in our selection processes.However, the matching rate decreases to ≃ % at a faint mag-nitude of ≃ . mag. This is due to the shallow depth of theHSC NB fields compared to the SCam ones. Konno et al. (2017)discuss the detection completeness of faint LAEs. Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 N u m b e r EW
0, Ly a [Å] D W e = +14−16 [Å] s g = +4−24 [Å] UD W e = +4−4 [Å] s g = +8−8 [Å] z~6.6 D W e = +18−18 [Å] s g = +36−12 [Å] UD W e = +6−4 [Å] s g = +2−2 [Å] z~5.7 Fig. 7. Ly α EW distribution for the HSC LAEs at z ≃ . (left) and z ≃ . (right). The top and bottom panels show the UD and D fields, respectively. Thethin gray histograms with error bars denote the Ly α EW distributions for the forced
LAEs. The error bars are given by Poisson statistics from the number ofsample LAEs. The red solid and blue dashed lines present the best-fit exponential and Gaussian functions of Equations (4) and (5), respectively, which areobtained from MC simulations with the EW , Ly α uncertainties (see Section 5.1 for more details). Here we present the Ly α EW distributions (Section 5.1) andLABs selected with the HSC data (Section 5.2). For the con-sistency with previous LAE studies, we use the forced
LAEsample in the following analyses, if not specified. α EW Distribution
We present the Ly α EW distributions for LAEs at z ≃ . − . .In a method described in Section 8, we calculate the rest-frameLy α EW, EW , Ly α , for the LAEs. The y ( z ) band magnitudesare used for the rest-frame UV continuum emission of z ≃ . ( z ≃ . ) LAEs. Figure 7 shows the observed Ly α EW dis-tributions at z ≃ . − . in the UD and D fields. To quan-tify these Ly α EW distributions we perform Monte Calro (MC)simulations. The procedure of the MC simulations is similarto that of e.g., Shimasaku et al. (2006), Ouchi et al. (2008)and Zheng et al. (2014). First, we generate artificial LAEs ina Ly α luminosity range of log L Ly α / erg s − = 42 − accord-ing to z ≃ . − . Ly α LFs of Konno et al. (2017). Next, we assign Ly α EW and BB magnitudes to each LAE by as-suming that the Ly α EW distributions are the exponential andGaussian functions (e.g., Gronwall et al. 2007; Kashikawa et al.2011; Oyarz´un et al. 2016): d N d EW = N exp (cid:16) − EWW e (cid:17) , (4)and, d N d EW = N p πσ exp (cid:16) − EW σ (cid:17) , (5)where N is the galaxy number, W e and σ g are the Ly α EW scalelengths of the exponential and Gaussian functions, respectively.By changing the intrinsic W e and σ g values, we make samplesof artificial Ly α EW distributions. We then select LAEs basedon NB and BB limiting magnitudes and BB − NB colors cor-responding to Ly α EW limits which are the same as those ofour LAE selection criteria (Section 3). Finally, the best-fit Ly α EW scale lengths are obtained by fitting to the artificial Ly α EWdistribution to the observed ones.Figure 7 presents the Ly α EW distributions obtained in theMC simulations. As shown in Figure 7, we find that the Ly α ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 ③(cid:0)✁✂✁ ❯✄❙☎✆✝✞ ▲✟✠✞✡☛☞✌❈✍✎ ❍✏✑✏✒✓✷✺ ③(cid:0)✔✂✕ ❯✄❆ ✐✖✗❬✘✙✚✛✜✚✢ ❪ ◆✣ ✤✥✦✧★✄✸✶ ✷✸ ✷✸✩✺ ✷✪ ✄✷✸ ✷✸✩✺ ✷✪ Fig. 8.
Isophotal area, A iso , as a function of NB magnitude to select LABs at z ≃ . (left) and z ≃ . (right). The top and bottom panels show the UD andD fields, respectively. The green diamonds denote the LABs. The filled red circles indicate the LAEs in the forced catalog. The gray dots represent star-likepoint sources selected in the HSC NB images. The diagonal and vertical lines denote the LAB selection criteria of A iso and NB magnitude. The diagonal linesare defined by the . σ deviation from the A iso -NB magnitude distribution for the star-like point sources. The filled red circles with a cross indicate unreliableLAB candidates which are affected by e.g., diffuse halos of nearby bright stars. The z ≃ . LABs in the UD fields are CR7 (Sobral et al. 2015) and Himiko(Ouchi et al. 2009).
EW distributions are reasonably explained by the exponentialand Gaussian profiles. The best-fit scale lengths are summa-rized in Table 5. The best-fit exponential (Gaussian) Ly α scalelengths are, on average of the UD and D fields, ± ˚A and ± ˚A ( ± ˚A and ± ˚A) at z ≃ . and z ≃ . ,respectively. As show in Table 5, there is no large difference inthe Ly α EW scale lengths for the UD and D fields. This no large EW , Ly α difference indicates that the results of our best-fit Ly α EW scale lengths does not highly depend on the image depthsand the detection incompleteness. In Section 6.1, we discussthe redshift evolution of the Ly α EW scale lengths.We investigate LEW LAEs whose intrinsic Ly α EW value, EW int0 , Ly α , exceeds ˚A (e.g., Malhotra & Rhoads 2002;Dawson et al. 2004). To obtain EW int0 , Ly α , we correct for the IGM attenuation for Ly α using the prescriptions of Madau(1995). In the HSC LAE ALL sample, we find that 45 and 230LAEs have a LEW of EW int0 , Ly α > ˚A, for z ≃ . and z ≃ . LAEs, respectively. These LEW LAEs are candidates of young-metal poor galaxies and AGNs. The fraction of the LEW LAEsin the sample is % for z ≃ . LAEs. The fraction of LEWLAEs at z ≃ . is comparable to that of previous studies on z ≃ . LAEs (e.g., ≃ % at z ≃ . in Ouchi et al. 2008; ≃ − % at z ≃ . in Shimasaku et al. 2006). In contrast,the fraction of LEW LAEs at z ≃ . is % which is lower thanthat at z ≃ . . The low fraction at z ≃ . might be due tothe neutral hydrogen IGM absorbing the Ly α emission. Out ofthe LEW LAEs, 32 and 150 LAEs at z ≃ . and z ≃ . ex-ceed EW int0 , Ly α = 240 beyond the σ uncertainty of EW int0 , Ly α , Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0
Table 4.
Properties of the LABs selected in the HSC NB Data.
Object ID α (J2000) δ (J2000) NB tot UV tot log L Ly α EW , Ly α z spec (mag) (mag) (erg s − ) ( ˚A)(1) (2) (3) (4) (5) (6) (7) (8) NB ( z ≃ . )HSC J100058 + a + e ± e a HSC J021757 − b − e +8 − e b HSC J100334 + c + e ± e c NB ( z ≃ . )HSC J100129 + + +40 − d HSC J100109 + + +172 − d HSC J100123 + + +70 − d HSC J095946 + + +25 − —HSC J100139 + + +66 − —HSC J161927 + + +33 − —HSC J161403 + + +23 − —HSC J232924 + + +45 − —(1) Object ID.(2) Right ascension.(3) Declination.(4) Total magnitudes of NB - and NB -bands for z ≃ . and z ≃ . , respectively.(5) Total magnitudes of y - and z -bands for z ≃ . and z ≃ . , respectively.(6) Ly α luminosity.(7) Rest-frame equivalent width of Ly α emission line.(8) Spectroscopic redshift. a CR7 in Sobral et al. (2015). b Himiko in Ouchi et al. (2009). c Spectroscopically confirmed in Shibuya et al. (2017b). d Spectroscopically confirmed in Mallery et al. (2012). e Spectroscopic measurements from the literature.respectively. z ≃ . − . We search for LABs with spatially-extended Ly α emission. Toidentify LABs, we measure the NB isophotal areas, A iso , for the forced LAEs. In this process, we include an unforced
LAE,Himiko, which is an LAB identified in a previous SCam NB sur-vey (Ouchi et al. 2009). First, we estimate the sky backgroundlevel of the NB cutout images. Next, we run the
SExtractor with the sky background level, and obtain the A iso values aspixels with fluxes brighter than the σ sky fluctuation. Notethat the NB magnitudes include both fluxes of Ly α and the rest-frame UV continuum emission. Instead of creating Ly α imagesby subtracting the flux contribution of the rest-frame UV con-tinuum emission, we here simply use the NB images for consis-tency with previous studies (e.g., Ouchi et al. 2009).Using A iso and NB magnitude diagrams, we select LABswhich are significantly extended compared to point sources.This selection is similar to that of Yang et al. (2010). Figure 8 presents A iso as a function of total NB magnitude. We alsoplot star-like point sources which are randomly selected in HSCNB fields. The A iso and NB magnitude selection window isdefined by a . σ deviation from the A iso -NB magnitude dis-tribution for the star-like point sources. The value of . σ isapplied for fair comparisons with previous studies of e.g., Yanget al. (2009) and Yang et al. (2010) who have used ≃ − σ .We perform visual inspections for the NB cutout images to re-move unreliable LABs which are significantly affected by e.g.,diffuse halos of nearby bright stars.In total, we identify 11 LABs at z ≃ . − . . Figure 9and Table 4 present multi-band cutout images and propertiesfor the LABs, respectively. As shown in Figure 9, these LABsare spatially extended in NB. Our HSC LAB selection confirmsthat CR7 and Himiko have a spatially extended Ly α emission.Six out of our 11 LABs have been confirmed by our spectro-scopic follow-up observations (Shibuya et al. 2017b) and previ-ous studies (Ouchi et al. 2009; Mallery et al. 2012; Sobral et al.2015). In Section 6.2, we discuss the redshift evolution of theLAB number density. ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 Fig. 9.
Postage stamps of the LABs selected with the HSC NB data. Theyellow contours indicate isophotal apertures with a threshold of σ sky back-ground noise level. The size of the cutout images is ′′ × ′′ . Table 5.
Best-fit Ly α EW ScaleLengths
Redshift W e σ g ( ˚A) ( ˚A)(1) (2) (3)6.6 (UD) +4 − +8 − +6 − +2 − +14 − +4 − +18 − +36 − (1) Redshift of the LAE sample. Theparenthesis indicates the UD or D fields.(2) Best-fit Ly α EW scale length of theexponential form. (3) Best-fit Ly α EWscale length of the Gaussian form. α EW Distribution
We discuss the redshift evolution of the Ly α EW scale lengthsin a compilation of the results from literature (Zheng et al.2014; Ouchi et al. 2008; Nilsson et al. 2009; Hu et al. 2010; Kashikawa et al. 2011; Cowie et al. 2011; Ciardullo et al. 2012).Figure 10 shows the redshift evolution of the Ly α EW scalelengths at z ≃ − . Our best-fit Ly α scale lengths are com-parable to that of Kashikawa et al. (2011) and/or Zheng et al.(2014) at z ≃ . − . . The high Ly α EW scale lengths at high- z would indicate that metal-poor and/or less-dusty galaxies witha strong Ly α emission is more abundant at higher- z (e.g., Starket al. 2011). In addition, Zheng et al. (2014) have found thatthe Ly α EW scale length increases towards high- z following a (1 + z ) -form. Our W e and σ g values for z ≃ . − . are alsoroughly comparable to Zheng et al.’s (1 + z ) -form evolution.However, no significant evolution in the Ly α EW scale lengthsfrom z ≃ . to z ≃ . is identified in our HSC LAE data, al-though a possible decline in σ g in the UD fields is found. Aslight decrease both in W e and σ g from z ≃ . to z ≃ . hasbeen found by Kashikawa et al. (2011). This sudden declinein the Ly α scale lengths at z ≃ . may be caused by the in-creasing hydrogen neutral fraction in the epoch of the cosmicreionization at z > ∼ . Note that the Ly α EW scale length mea-surements would largely depend on BB and NB depths and Ly α EW cuts. Using deeper NB and BB images from the future HSCdata release, we will examine the redshift evolution of Ly α scalelengths accurately. We discuss the redshift evolution of the LAB number density, N LAB . Figure 11 shows N LAB at z ≃ − measured by thisstudy and the literature (Keel et al. 2009; Yang et al. 2009; Yanget al. 2010; Matsuda et al. 2009; Saito et al. 2006). For the plotof the N LAB , Yang et al. (2010) have compiled N LAB measure-ments down to an NB surface brightness (SB) limit of × − erg s − cm − arcsec − . The SB limits of our HSC NB dataare ≃ × − and ≃ × − erg s − cm − for the UDand D fields, respectively. Our HSC NB images at least for theUD fields are comparably deep, allowing for fair comparisonswith Yang et al.’s N LAB plot. Our N LAB values are . × − and . × − Mpc − ( . × − and . × − Mpc − ) at z ≃ . and z ≃ . in the UD (D) fields, respectively. The num-ber density at z ≃ − is ≃ − times lower than thoseclaimed for LABs at z ≃ − (e.g., Matsuda et al. 2004; Yanget al. 2009; Yang et al. 2010). As shown in Figure 11, there isan evolutional trend that N LAB increases from z ≃ to ≃ andsubsequently decreases from z ≃ to ≃ . This trend of theLAB number density evolution is similar to the Madau-Lillyplot of the cosmic SFR density (SFRD) evolution (e.g., Madauet al. 1996; Lilly et al. 1996). Similar to Shibuya et al. (2016),we fit the Madau-Lilly plot-type formula, N LAB ( z ) = a × (1 + z ) b z ) /c ] d , (6)where a, b, c , and d are free parameters (Madau & Dickinson Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 s g [ Å ] z
100 0 2 4 630050 W e [ Å ] HSC (UD)HSC (D)Co11Ci12Ni09Ou08Ka11Zh14
Fig. 10.
Redshift evolution of the best-fit Ly α EW scale lengths of the expo-nential (top) and Gaussian (bottom) functions. The red squares and circlesindicate our HSC LAEs in the UD and D fields, respectively. The black sym-bols are taken from the data points in literature which have been compiledin Zheng et al. (2014) (cross: Cowie et al. 2011; asterisks: Ciardullo et al.2012; filled triangle: Nilsson et al. 2009; filled inverse triangles: Ouchi et al.2008; filled diamonds: Kashikawa et al. 2011; open circles: results of Monte-Carlo simulations using data of Zheng et al. 2014 and Hu et al. 2010). Thegray curves indicate the best-fit (1 + z ) -form functions obtained in Zhenget al. (2014). N LAB evolution. For the fitting, we excludeMatsuda et al. (2009)’s data point which has been obtainedin a overdense region, SSA22. The best-fit parameters are a = 9 . × − , b = 2 . , c = 5 . , and d = 11 . .The similarity of the cosmic SFRD and LAB evolution mightindicate that the origin of LABs are related to the star formationactivity. As described in Section 1, LABs are thought to beformed in physical mechanisms that are connected with the starformation, e.g., the cold gas accretion and the galactic super-winds. The cold gas accretion could produce the extended Ly α emission powered by the gravitational energy (e.g., Momoseet al. 2016; Mas-Ribas & Dijkstra 2016; Mas-Ribas et al. 2017).On the other hand, the superwinds induced by the starburstsin the central galaxies would blow out the surrounding neutralgas, and form extended Ly α nebulae (e.g., Mori & Umemura2006). The cold gas accretion rate and the strength of galacticsuperwinds are predicted to evolve with physical quantities re-lated to the cosmic SFRD (e.g., Dekel et al. 2009; Kereˇs et al.2009). The comparisons of the cosmic SFRD and LAB evolu- -7-6-5-4 0 2 4 6 l o g N LAB [ Mpc - ] z LAB Evolution
HSC UDHSC DKe09Ya09Ya10Ma04Sa06SSA22
Fig. 11.
Redshift evolution of the LAB number density. The filled red squaresand filled red circles denote the LABs selected in the HSC UD and D fields,respectively. The error bars are given by Poisson statistics from the LABnumber counts. The black symbols show LABs in the literature (filled dia-mond: Keel et al. 2009; filled circle: Yang et al. 2009; open circle: Yanget al. 2010; filled inverse-triangle: Matsuda et al. 2004; pentagon: Saitoet al. 2006). All the measurements are based on LABs identified down to thesurface brightness limit of ≃ × − erg s − cm − arcsec − . The graysolid curve represents the best-fit formula of Equation 6 to the data pointsexpect for the measurement in the SSA22 proto-cluster region. tions would provide useful hints that LABs are formed in thesescenarios.However, it should be noted that the LAB selection methodis not homogeneous in our comparison of N LAB at z ≃ − .There is a possibility that the N LAB evolution from z ≃ to z ≃ is caused by the cosmological surface brightness dimmingeffect at high- z . The cosmological surface brightness dimmingwould significantly affect the detection and selection complete-ness for LABs at high- z . To confirm the N LAB evolution andquantitatively compare with the cosmic SFRD, we need to ho-mogenize the selection method for LABs at z ≃ − in thefuture HSC NB data. We develop an unprecedentedly large catalog consisting ofLAEs at z = 5 . and . that are identified by the SILVERRUSHprogram with the first NB imaging data of the Subaru/HSC sur-vey. The NB imaging data is about an order of magnitude largerthan any other surveys for z ≃ − LAEs conducted to date.Our findings are as follows: • We identify 2,230 > ∼ L ∗ LAEs at z = 5 . and . on the . and . deg sky, respectively. We confirm that the LAE cat-alog is reliable on the basis of 96 LAEs whose spectroscopicredshifts are already determined by this program (Shibuyaet al. 2017b) and the previous studies (e.g., Mallery et al.2012). The LAE catalog is presented in this work, and pub-lished online. ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0 • With the large LAE catalog, we derive the rest-frame Ly α EW distributions of LAEs at z ≃ . and ≃ . that are rea-sonably explained by the exponential profile. The best-fit ex-ponential (Gaussian) Ly α scale lengths are, on average ofthe Ultradeep and Deep fields, ± ˚A and ± ˚A( ± ˚A and ± ˚A) at z ≃ . and z ≃ . , re-spectively, showing no significant evolution from z ≃ . to z ≃ . . We find 45 and 230 LAEs at z ≃ . and z ≃ . witha LEW of EW int0 , Ly α > ˚A corrected for the IGM attenu-ation for Ly α . The fraction of the LEW LAEs to all LAEs is ≃ % and ≃ % at z ≃ . and z ≃ . , respectively. TheseLEW LAEs are candidates of young-metal poor galaxies andAGNs. • We search for LABs that are LAEs with spatially extendedLy α emission whose profile is clearly distinguished fromthose of stellar objects at the > ∼ σ level. In the search, weidentify 11 LABs in the HSC NB images down to a surfacebrightness limit of ≃ − × − erg s − cm − which is asdeep as data of previous studies. The number density of theLABs at z ≃ − is ∼ − − − Mpc − that is ∼ − times lower than those claimed for LABs at z ≃ − , sug-gestive of disappearing LABs at z > ∼ , although the selectionmethods are different in the low and high- z LABs.It should be noted that Ly α EW scale length derivation meth-ods and the LAB selections are not homogeneous in a redshiftrange of z ≃ − . Using the future z ≃ . , . , . , and . HSC NB data, we will systematically investigate the redshiftevolution of Ly α EW scale lengths and N LAB at z ≃ − inhomogeneous methods. α EW In this section, we describe the method to calculate the EW , Ly α values. The procedures and the assumption of thismethod are similar to those of e.g., Malhotra & Rhoads (2002),Dawson et al. (2004), Gronwall et al. (2007), Kashikawa et al.(2011). For the calculation of EW , Ly α , we assume that LAEshave a δ function-shaped Ly α line and the flat rest-frame UVcontinuum emission (i.e. β ν = 0 , where β ν is the UV spectralslope per unit frequency). In such an LAE spectrum, the mag-nitude, m , for a waveband filter with a transmission curve, T ν ,is described as follows: . m = − . × log R ∞ ( f c + f l δ ( ν − ν α )) T ν d ν R ∞ T ν d ν , (7)where f l , f c , δ ( ν ) , and ν α is a Ly α line flux, the flux density ofthe rest-frame UV continuum emission, the δ function, and theobserved frequency of Ly α , respectively. Here we also assumethat the Ly α line is located at ˚A ( ˚A) which is thecentral wavelength of the NB ( NB ) filter, for z ≃ . ( z ≃ . ) LAEs. In this study, we do not take into account theIGM transmission for Ly α , if not specified. This is because the IGM transmission for Ly α highly depends on the Ly α linevelocity offset from the systemic redshift (e.g., Hashimoto et al.2013; Shibuya et al. 2014b). The numerator of the logarithm inEquation (7) corresponds to f c Z ∞ ν c exp ( − τ eff ) T ν d ν + f c Z ν c T ν d ν + f l T ν ( ν α )= f c B + f c R + f l T ν ( ν α ) . (8)In Equation (8), we use B , R , and A that are defined by equa-tions of B ≡ Z ∞ ν c exp ( − τ eff ) T ν d ν, (9) R ≡ Z ν c T ν d ν, (10) A ≡ Z ∞ T ν d ν, (11)where τ eff is the IGM optical depth calculated from analyticalmodels of Madau (1995). Using Equations (7) and (8), we de-rive the flux density of the NB and BB filters, f NB and f BB , asfollows: f NB = 10 − . m NB +48 . = f c ( B NB + R NB ) + f l T NB ( ν α ) A NB , (12) f BB = 10 − . m BB +48 . = f c ( B BB + R BB ) A BB . (13)The B , R , and A values with the subscripts of NB (BB) arecalculated with the transmission curves of the NB (BB) filters, T NB ( T BB ). In this study, we use magnitudes of the y and z bandfilters which do not cover the wavelength of Ly α for z ≃ . and z ≃ . LAEs, respectively, indicating T BB ( ν α ) = 0 . Inthe case that m BB is fainter than the σ limit, we use the σ limiting magnitude for the EW , Ly α calculation. By combiningthe equations of f NB and f BB , we obtain f c and f l , f c = A BB f BB B BB + R BB = f BB , (14) f l = A NB ( B BB + R BB ) f NB − A BB ( B NB + R NB ) f BB ( B BB + R BB ) T NB ( ν α ) (15) = A NB f NB − ( B NB + R NB ) f BB T NB ( ν α ) (16) = a × f NB − b × f BB . (17)Note that B BB + R BB = A BB due to the negligible IGM absorp-tion at the wavelengths of the BB filters. Here we define a and b as a ≡ A NB T NB ( ν α ) , (18) b ≡ B NB + R NB T NB ( ν α ) . (19) Publications of the Astronomical Society of Japan , (2014), Vol. 00, No. 0
For the HSC NB and NB filters, the sets of the val-ues are calculated to be ( a, b ) ≃ (4 . , . × and ( a, b ) ≃ (5 . , . × , respectively. Using f c and f l , we calculate the EW , Ly α values via EW , Ly α = f l f c cν
11 + z . (20)To obtain the median values and uncertainties for EW , Ly α ,we perform Monte Carlo (MC) simulations in a method simi-lar to that of e.g., Shimasaku et al. (2006). In the simulation,we randomly generate a flux density value, f MC , following aGaussian probability distribution with an average of f and adispersion of the σ sky background noise, f σ , for the NBand BB bands. Here we also randomize β ν and ν α in Gaussianprobability distributions with σ dispersions of ∆ β = 0 . and ∆ ν α = FWHM NB / . , respectively, where FWHM NB is theFWHM of the NB filters. The dispersion of ∆ β = 0 . is typ-ical for high- z galaxies (Bouwens et al. 2014). In the mannerthat are the same as described in this section, we calculate a EW , Ly α value using f MC for NB and BB. In this process, neg-ative values of f c , f l , and EW , Ly α are forced to be zero. Sucha process is performed 1,000 times for each object. During theiteration, a simulated EW , Ly α value is discarded in the casethat a BB − NB color does not meet the selection criteria ofEquations (1) and (2). Using the set of EW , Ly α values ob-tained from the MC simulations, we calculate the median valuesand the 16- and 84-percentile errors for EW , Ly α . Acknowledgments
We would like to thank James Bosch, Richard S. Ellis, Masao Hayashi,Robert H. Lupton, Michael A. Strauss for useful discussion and com-ments. We thank the anonymous referee for constructive comments andsuggestions. This work is based on observations taken by the SubaruTelescope and the Keck telescope which are operated by the NationalObservatory of Japan. This work was supported by World PremierInternational Research Center Initiative (WPI Initiative), MEXT, Japan,KAKENHI (23244025) and (21244013) Grant-in-Aid for ScientificResearch (A) through Japan Society for the Promotion of Science (JSPS),and an Advanced Leading Graduate Course for Photon Science grant. TheNB816 filter was supported by Ehime University (PI: Y. Taniguchi). TheNB921 filter was supported by KAKENHI (23244025) Grant-in-Aid forScientific Research (A) through the Japan Society for the Promotion ofScience (PI: M. Ouchi). NK is supported by the JSPS grant 15H03645.SY is supported by Faculty of Science, Mahidol University, Thailandand the Thailand Research Fund (TRF) through a research grant for newscholar (MRG5980153).The Hyper Suprime-Cam (HSC) collaboration includes the astro-nomical communities of Japan and Taiwan, and Princeton University.The HSC instrumentation and software were developed by the NationalAstronomical Observatory of Japan (NAOJ), the Kavli Institute for thePhysics and Mathematics of the Universe (Kavli IPMU), the Universityof Tokyo, the High Energy Accelerator Research Organization (KEK),the Academia Sinica Institute for Astronomy and Astrophysics in Taiwan(ASIAA), and Princeton University. Funding was contributed by theFIRST program from Japanese Cabinet Office, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), the Japan Society forthe Promotion of Science (JSPS), Japan Science and Technology Agency(JST), the Toray Science Foundation, NAOJ, Kavli IPMU, KEK, ASIAA,and Princeton University.This paper makes use of software developed for the Large SynopticSurvey Telescope. We thank the LSST Project for making their codeavailable as free software at http:dm.lsst.orgThe Pan-STARRS1 Surveys (PS1) have been made possible throughcontributions of the Institute for Astronomy, the University of Hawaii,the Pan-STARRS Project Office, the Max-Planck Society and its partic-ipating institutes, the Max Planck Institute for Astronomy, Heidelbergand the Max Planck Institute for Extraterrestrial Physics, Garching,The Johns Hopkins University, Durham University, the Universityof Edinburgh, Queen’s University Belfast, the Harvard-SmithsonianCenter for Astrophysics, the Las Cumbres Observatory Global TelescopeNetwork Incorporated, the National Central University of Taiwan, theSpace Telescope Science Institute, the National Aeronautics and SpaceAdministration under Grant No. NNX08AR22G issued through thePlanetary Science Division of the NASA Science Mission Directorate,the National Science Foundation under Grant No. AST-1238877, theUniversity of Maryland, and Eotvos Lorand University (ELTE) and theLos Alamos National Laboratory.Based on data collected at the Subaru Telescope and retrieved fromthe HSC data archive system, which is operated by Subaru Telescope andAstronomy Data Center at National Astronomical Observatory of Japan.
References
Aihara, H., et al. 2017a, arXiv:1702.08449—. 2017b, arXiv:1704.05858Ajiki, M., et al. 2002, ApJL, 576, L25Arrigoni Battaia, F., Hennawi, J. F., Prochaska, J. X., & Cantalupo, S.2015a, ApJ, 809, 163Arrigoni Battaia, F., Yang, Y., Hennawi, J. F., Prochaska, J. X., Matsuda,Y., Yamada, T., & Hayashino, T. 2015b, ApJ, 804, 26Axelrod, T., Kantor, J., Lupton, R. H., & Pierfederici, F. 2010, An opensource application framework for astronomical imaging pipelinesBertin, E., & Arnouts, S. 1996, A&As, 117, 393Blanc, G. A., et al. 2011, ApJ, 736, 31Bond, N. A., Gawiser, E., Guaita, L., Padilla, N., Gronwall, C., Ciardullo,R., & Lai, K. 2012, ApJ, 753, 95Bosch, J., et al. 2017, ArXiv e-printsBouwens, R. J., et al. 2014, ApJ, 793, 115Cai, Z., et al. 2017, ApJ, 837, 71Cantalupo, S., Arrigoni-Battaia, F., Prochaska, J. X., Hennawi, J. F., &Madau, P. 2014, Nature, 506, 63Cantalupo, S., Porciani, C., Lilly, S. J., & Miniati, F. 2005, ApJ, 628, 61Ciardullo, R., et al. 2012, ApJ, 744, 110Cowie, L. L., Barger, A. J., & Hu, E. M. 2010, ApJ, 711, 928Cowie, L. L., Hu, E. M., & Songaila, A. 2011, ApJL, 735, L38Dawson, S., et al. 2004, ApJ, 617, 707Dekel, A., et al. 2009, Nature, 457, 451Dressler, A., Martin, C. L., Henry, A., Sawicki, M., & McCarthy, P. 2011,ApJ, 740, 71Duval, F., Schaerer, D., ¨Ostlin, G., & Laursen, P. 2013, ArXiv e-printsErb, D. K., Bogosavljevi´c, M., & Steidel, C. C. 2011, ApJL, 740, L31Finkelstein, S. L., Rhoads, J. E., Malhotra, S., Grogin, N., & Wang, J.2008, ApJ, 678, 655 ublications of the Astronomical Society of Japan , (2014), Vol. 00, No. 019