Subaru Spectroscopy of SDSS-II Supernovae
Kohki Konishi, Naoki Yasuda, Kouichi Tokita, Mamoru Doi, Yutaka Ihara, Tomoki Morokuma, Naohiro Takanashi, Jakob Nordin, John Marriner, Linda Östman, Michael Richmond, Masao Sako, Donald P. Schneider, J. Craig Wheeler
aa r X i v : . [ a s t r o - ph . C O ] J a n Subaru Spectroscopy of SDSS-II Supernovae Kohki Konishi , , Naoki Yasuda , , Kouichi Tokita , Mamoru Doi , Yutaka Ihara , TomokiMorokuma , Naohiro Takanashi , Jakob Nordin , , John Marriner , Linda ¨Ostman ,Michael Richmond , Masao Sako , Donald P. Schneider , J. Craig Wheeler [email protected] ABSTRACT
The Sloan Digital Sky Survey II (SDSS-II) Supernova Survey discovered TypeIa supernovae (SNe Ia) in an almost unexplored intermediate redshift range of0 . < z < . Based in part on data collected at Subaru Telescope, which is operated by the National AstronomicalObservatory of Japan. Department of Physics, Graduate School of Science, University of Tokyo, Tokyo 113-0033, Japan Institute for Cosmic Ray Research, University of Tokyo, Kashiwa 277-8582, Japan Institute for the Physics and Mathematics of the Universe, University of Tokyo, Kashiwa 277-8582,Japan Institute of Astronomy, Graduate School of Science, University of Tokyo, 2-21-1 Osawa, Mitaka, Tokyo181-0015, Japan National Astronomical Observatory of Japan, 2-21-1, Osawa, Mitaka, Tokyo, 181-8588, Japan Institute of Industorial Science, University of Tokyo, 153-8505, Japan Department of Physics, Stockholm University, 106 91 Stockholm, Sweden. Oskar Klein Centre for Cosmo Particle Physics, AlbaNova, 106 91 Stockholm, Sweden. Center for Particle Astrophysics, Fermi National Accelerator Laboratory, Batavia, IL 60510, USA Institut de F´ısica d’Altes Energies, Universitat Aut`onoma de Barcelona, Bellaterra, Spain Physics Department, Rochester Institute of Technology, Rochester, NY 14623, USA Department of Physics and Astronomy, University of Pennsylvania, 209 South 33rd Street, Philadelphia,PA 19104, USA Department of Astronomy and Astrophysics, Pennsylvania State University, University Park, PA 16802,USA Astronomy Department, University of Texas, Austin, TX 78712, USA
Subject headings: methods: observational - techniques: spectroscopic - super-novae: general - supernovae:individual - surveys - cosmology: observations
1. Introduction
Type Ia Supernovae (SNe Ia) have played a significant role in observational cosmologyas a probe for the expansion history of the Universe (e.g. Riess et al. 1998; Perlmutter etal. 1999). This is because the majority of the observed SNe Ia have homogeneous lightcurves and spectra.Recent nearby observations have enabled the construction of large datasets of SNe Iaobtained by a homogeneous procedure. Many spectra have been stored in the SUpernovaSPECTra (SUSPECT) database . A supernova group in the Center for Astrophysics (CfA)presented UBVRI photometry of 44 nearby SNe Ia (Jha et al. 2006) and 185 SNe Ia (Hickenet al. 2009) and 32 optical spectra (Matheson et al. 2008). The Carnegie Supernova Projectrecently published ugri BVYJHK s light curves of 35 nearby SNe Ia (Contreras et al. 2010).The Nearby Supernova Factory (Aldering et al. 2002) has been gathering spectrophotometrydata of nearby SNe Ia. High redshift (high-z) SNe Ia have been observed by several groups:the High-Z team (Riess et al. 1998), the Supernova Cosmology Project (Perlmutter etal. 1999), the SuperNova Legacy Survey (Astier et al. 2006) and the Equation of State:SupErNovae trace Cosmic Expansion survey (Miknaitis et al. 2007). However, since widefield observations are required to obtain comparable numbers of SNe Ia, few have beenobserved in the intermediate redshift range (0 . < z < . http://bruford.nhn.ou.edu/˜suspect/ ugriz , Fukugita et al. 1996). The observed cadence wastwo days minimum for the same sky region. Analysis was conducted in real-time to assignSN candidates for spectroscopic observations (Sako et al. 2008). Follow-up spectroscopicobservations were carried out at a large number of telescopes to determine at least thesupernova type and redshift; the 2.4m Hiltner, the 2.5m NOT, the 3.5m SDSS ARC, the3.6m NTT, the 4.2m WHT, the 8.2m Subaru, the 9.2m HET and the 10m Keck I telescopes.The spectroscopy of the first season (Year 2005) is presented in Zheng et al. (2008). ¨Ostmanet al. (2010) present the spectra obtained at the NTT and the NOT during Year 2006 and2007. Holtzman et al. (2008) provide high-quality SN light curves in ugriz for the firstseason. Foley et al. (2010) describes the Keck followup spectra and presents a mismatch inthe ultraviolet spectra of these SNe Ia from nearby counterparts. The final data release forall spectra obtained by these telescopes will be published in Sako et al (in preparation).In this paper, we present an observational summary of the Subaru spectroscopic fol-lowup. § § § §
2. Observations
We used the Subaru telescope for follow-up spectroscopic observations of the SN can-didates found by the SDSS-II Supernova Survey for the first two seasons (Years 2005 &2006). The Faint Object Camera and Spectrograph (FOCAS; Kashikawa et al. 2002) wasused in the long-slit mode. The atmospheric dispersion corrector (ADC) was used in ourobservations. The instrumental dispersions were 1.34 ˚A/pix. We used a 0.8” slit for SNcandidates and a 2.0” slit for spectroscopic standard stars. The pixel scale was 0.104” andthe data were read out with three pixel binning in the spatial direction. In order to obtainspectra in as wide a wavelength range as possible, we observed red and blue sides separatelywith different sets of grisms and filters. We obtained a red side spectrum first to identifythe “Si ii λ ◦ , so thatthey suffered the least from atmospheric absorption.For Year 2005, we adopted the following grating and filter combinations for the blue andthe red sides; (i) 300 line mm − grating blazed at 5500 ˚A (300B) and the L600 filter, which is 4 –sensitive in the wavelength window of 4000 to 6000 ˚A, and (ii) 300 line mm − grating blazedat 7500 ˚A (300R) and the SY47 filter, which is sensitive in the wavelength window of 5000to 9000 ˚A. The configuration for Year 2006 was the same as the first year except the filterfor the blue side. We replaced the L600 filter with a L550 filter which is more efficient forshorter wavelengths, widening the spectral window to 3600 to 9000 ˚A. Table 1 summarizesour configurations.Because of the Subaru’s large aperture and a superb seeing site, we observed the subsetof SDSS targets with large host galaxy contamination. We aligned the slit so as to gothrough each SN candidate and host center. The slit angles were initially determined fromthe images taken by the SDSS 2.5m telescope, but then adjusted by checking the r -bandimages taken by the Subaru telescope at the time of each observation. This slit alignmentenabled us to simultaneously determine the SN type as well as its redshift. The redshiftwas generally determined using spectral lines of the host galaxy (see § ×
300 seconds for both the blue and red sides. Candidates spectra were observed onlyonce. The only exception was a peculiar variable SN10450. We observed SN10450 with anexposure of 900 seconds and 1,800 seconds ( §
3. Spectra reduction
The spectroscopic data were reduced using the Image Reduction and Analysis Facility(IRAF) and our own programs written in C. The signals from objects were recorded in two-dimensional (2D) CCD detectors (called, “frames” hereafter). We carried out the reductionto obtain calibrated 2D spectra along the spatial and wavelength axes with IRAF as follows. Bias subtraction:
Even when the exposure time was zero, the pixel ADU countswere not zero but some positive value; a bias voltage was applied when reading the CCD toavoid negative values due to readout noise. This bias value (ADU counts when the exposuretime was zero) had to be subtracted from every frame. We used an overscan region of 2Dbias spectra to subtract the bias.
Flat fielding:
We observed three dome flat fields for each grism and filter combinationand created high S/N dome flats by taking the median of these three dome flat frames. Wecorrected for the non-uniformity by dividing the bias subtracted frame by the median domeflat field.
Wavelength calibration:
We used night glow OH emission lines recorded in objectspectra for the red side and the Th-Ar lamp emission lines taken separately from objectframes for the blue side. The overall wavelength calibration was done using a Legendrepolynomial of third order for the red side and a Legendre polynomial of fifth order for theblue side. The polynomial order was determined to minimize the rms of the wavelengthresiduals.
Distortion correction:
Our spectra are distorted in the spatial direction to someextent. We obtained a spatial distortion map for each frame by tracing the spatial locationof spectrophotometric standard stars at each wavelength. We positioned the telescope so asto obtain spectra of objects and standard stars at the same position on the CCD. We thenused the distortion map to correct for each object spectrum.
Background subtraction:
Background photons from night glow contaminated theframes. We subtracted this background contamination by masking the object region andfitting a low order polynomial function to the spatial profile along the wavelength axis.
Flux calibration:
We observed spectrophotometric standard stars: BD+28 ◦ ◦
325 and Feige34. Library spectra in IRAF were used for BD+28 ◦ IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the As-sociation of Universities for Research in Astronomy, Inc., under cooperative agreement with the NationalScience Foundation. ◦
325 and Feige34. A spectrum in the HST Calibration Database System(CALSPEC) was used for GD71 to create the sensitivity function for the flux calibration,which is expressed as a fifth order cubic spline function. The best fit polynomial function wassearched for each standard star spectrum. Table 3 is the list of spectrophotometric standardstars and the origin of the spectral energy distributions (SEDs) used in this reduction. Theextinction file for the Mauna Kea telescope was used to remove the telluric extinction. Correction for dust extinction in our Galaxy:
We corrected for dust extinctionin our Galaxy using R V = 3 . E ( B − V ) from Schlegel, Finkbeiner &Davis (1998). The extinction was assumed to follow the Cardelli et al. (1989) law.We developed C programs for the following processes: Telluric absorption removal:
Any spectra observed from the ground suffers fromtelluric absorption lines due to water and oxygen molecules in the atmosphere, especially atwavelengths longer than 5000 ˚A. The deepest is the A-band line at ∼ z ∼ .
25, most of our sample, this absorption is superimposed around the “Si ii λ ii λ ◦ ◦ Seeing measurement:
Seeing is the Full Width at Half Maximum (FWHM) of thespatial profile recorded on the detector. We calculated the seeing using a 2D spectrum of astandard star, observed with a 2.0” slit (wide enough to pass through all the photons froma point-like object), along the spatial and wavelength axes. The spatial profile of the starwas traced in the wavelength direction with a Gaussian function to obtain the seeing size asa function of wavelength. We used BD+28 ◦ ◦ − . ± . θ ∝ ( λλ ) − . . (1)We used this relation for the SN extraction. The index for the FOcal Reducer and lowdispersion Spectrograph (FORS) attached to the Very Large Telescope (VLT) was reportedto be -0.29 (Blondin et al. 2005). The Subaru/FOCAS result is comparable to their value. SN extraction:
The observed SNe suffered from host galaxy contamination. Thebest way to eliminate the galaxy light from the SN candidate may be to observe the samegalaxy with the same telescope configuration after the candidate disappears; however, thistakes considerable time. As an alternative, we have developed algorithms to eliminate thegalaxy contribution. Two proposed methods had been discussed in the literature: one wasto extract SN spectra from 1D galaxy-contaminated spectra with the aid of galaxy templatespectra (e.g. Howell et al. 2005; Ellis et al. 2008; Zheng et al. 2008; Foley et al. 2008a;Tokita 2009; ¨Ostman et al. 2010) and the other was to extract SN spectra from 2D galaxy-contaminated spectra without the aid of galaxy template spectra (Blondin et al. 2005; Bau-mont et al. 2008). An advantage of the latter method is that no assumption of a host galaxyspectrum is necessary. For this method, one should determine the spatial profiles for the SNand host components for the extraction. We developed an algorithm of SN extraction basedon the latter concept. The basic concept is similar to the technique used by the SuperNovaLegacy Survey collaboration (Baumont et al. 2008), but developed independently.A model spectrum M ( x, λ ) is assumed to be the sum of the SN and host galaxy flux: M ( x, λ ) = g SN ( x, λ ) + g gal ( x, λ ) , (2)where x is the pixel index in the spatial direction and λ is the discretized wavelength. Thefunction g SN ( x, λ ) is the SN candidate flux density in the pixel ( x, λ ) and g gal ( x, λ ) is thehost galaxy flux density. The spatial profile of a point source g SN ( x, λ ) is assumed to be a 8 –Gaussian function: g SN ( x, λ ) = h SN ( λ ) exp − (cid:18) x − x SN ( λ ) θ SN ( λ ) (cid:19) ! , (3)where θ SN is the bestfit Gaussian width for the SN candidate. The spatial profile of theextended component g gal ( x, λ ) is assumed to be the sum of two Gaussian functions. Thesumming was introduced in order to approximate the bulge and disk components of the hostgalaxy. We should note that a convolution with the PSF of more sophisticated functionssuch as the de Vaucouleurs R / law or an exponential law are more suited to fit galaxymorphologies. Although a sum of two Gaussians is just an approximation, the χ of fittingtells us that this approximation is reasonable for the majority of our sample. For our hostswith complicated spatial profiles, we performed a χ fitting by adjusting two Gaussian peakpositions representing the hosts. A larger number of Gaussians or a more generalized function(Blondin et al. 2005) would be required for hosts with more complex spatial profiles.In the fitting process, photon statistics was assumed to be the only source of the un-certainty σ ( x, λ ). This uncertainty was assumed to be constant in the spatial direction σ ( x, λ ) = σ ( λ ) and was estimated from the blank areas adjacent to the object. The widthof the blank (no SN nor hosts) area was typically 25 arcsec on both sides, centered on theobject. We minimized: χ ( λ ) = X x (cid:18) O ( x, λ ) − M ( x, λ ) σ ( λ ) (cid:19) (4)where O ( x, λ ) was the observed flux density at each pixel.The fitting was done with a two stage process. First, the model function M ( x, λ ) wasfitted to the averaged spatial profile created by averaging the 2D flux over the wavelengthranges (3600 to 5400, 4000 to 6000 and 5000 to 9000 ˚A) to determine the central positionsand the widths of each component such as θ SN . The width θ SN should be identical to theseeing size. Then, the spatial profile at each wavelength was fitted, with the spatial positionsof the SN and its host fixed at the values obtained in the first step and the widths changingdue to the seeing-wavelength relation of Equation 1.After finding the parameters, a candidate spectrum g SN ( λ ), its uncertainty g SN,err ( λ )and the host galaxy spectrum g gal ( λ ) were derived. If we created SN spectra by integratingflux over too wide a x region, the S/N ratio became low. We restricted SN candidate fluxwithin the FWHM (2 √ ln 2 θ SN ). g SN ( λ ) = Z x sma x smi g ( x, λ ) dx, (5) 9 –where x smi ≡ x SN − θ SN √ ln ≤ x ≤ x SN + θ SN √ ln ≡ x sma , and the integral was calculatedusing the Trapezoidal formula.The SN flux uncertainty g SN,err ( λ ) was estimated from: g SN,err ( λ ) = ∆( Z x sma x smi g SN ( x, λ ) dx ) , (6)∆ h SN = σ ( λ ) q θ SN p π/ . (7)Therefore, g SN,err ( λ ) = Z x sma x smi exp − (cid:18) x − x SN θ SN (cid:19) ! dx × ∆ h SN θ SN , = Z x sma x smi exp − (cid:18) x − x SN θ SN (cid:19) ! dx × (cid:18) π (cid:19) / p θ SN σ ( λ ) . (8)Host galaxy spectra were constructed by subtracting the SN component g SN from O ( x, λ ).In Figure 4, we show an example of the SN extraction procedure.We should note that the fitting would be affected by the presence of emission lines.Our method fits spatial profiles at each wavelength with the sum of two or three Gaussianfunctions. When a host galaxy spatial profile can be approximated by a Gaussian, the totalspatial profile would be the sum of two Gaussians. For the galaxy approximated by twoGaussians, the total would be that of three Gaussians. Spatial profiles at the lines could betoo complicated for our model if there were strong lines from host galaxies. Since SNe Ia donot show emission lines in their spectra, we can regard emission lines in the extracted SNspectrum, more than 80 % of our sample, as those originated in their host galaxies. We canregard absorption lines in these spectra as galaxy origin if its width is smaller than expandingvelocities of SN ejecta ( ∼ ,
000 km sec − ).A profile fit to a Gaussian function might miss a few percent of the observed flux. Ananalysis of spatial profiles for standard stars showed a slight excess in flux at the outskirtsof the spatial profile compared to the prediction of the Gaussian function. We neglect thisexcess for our current analysis. Combining spectra:
In most cases, SN spectra were taken with two different combi-nations of grisms and filters: blue and red side spectra. We combined these two spectra by 10 –taking the average flux within the overlapping wavelength regions and scaling the red side tomatch the blue side flux and re-binning to 2 ˚A per pixel. The result is a contiguous sciencespectrum together with an uncertainty spectrum representing the statistical uncertainties inthe flux in each binned pixel.
Slitloss correction:
Since seeing sizes were comparable to the slit width, some frac-tion of the object photons could not go through the slit. We corrected for this slitloss byestimating gri magnitudes at the spectroscopy date from light curves and spectra. Magni-tudes at the spectroscopy date were determined by interpolating the bestfit templates for theSALT2 light curve fitter (see the light curve fitting in § gri filter responses. We matched the interpolated magnitudes with thesynthetic magnitudes to determine a second-order polynomial slitloss correction function; a ( λ ) = X i =0 c i λ i , (9)where c i is a coefficient. This function was determined for each spectrum.
4. Dataset § One can determine the redshift of the supernova either from the supernova itself or fromthe host galaxy; the latter was our primary choice. The representative emission lines areBalmer lines (H α and H β ) and forbidden lines of [N ii ], [O ii ], [O iii ] and [S ii ]. Redshifts weredetermined from the wavelengths determined by fitting Gaussian functions to all the emissionlines. The absorption lines we used were Ca H&K, Mgb (SN12991) and NaD (SN12979). Weidentified absorption lines by visual inspection. The final redshift was the average value ofall measured lines with equal weight. Lines and their rest-wavelengths are listed in Table4. If there were no spectral lines in the host galaxy spectra, we adapted the technique ofspectra template fitting to the SN spectrum. This technique cross-correlates SN spectra witha series of SN library spectra (spectral templates and high quality nearby SN spectra) and
11 –search for the χ minimum to determine redshifts and types; details are explained in Tokita(2009). The redshifts for 64 objects were determined from spectral lines of their host galaxiesand seven were from the fitting of SN spectra. Figure 7 shows the redshift distribution forall our targets except 1 AGN at z = 1 . ∼ ii λ i at λλ ii and O i λ ii λ α line of SNe II. Column 9 is themean airmass of our observations. The median value is 1.2, with 23 % of the observationshaving an airmass larger than 1.3. Column 10 shows two ratios, F host /F all , of host galaxyflux to reconstructed SN and host flux for the blue (4000 to 6000 ˚A or 3600 to 5400 ˚A)and red side spectra (5000 to 9000 ˚A), respectively. Ratios are determined from averagedspatial profiles over the wavelength region of the spectrum, which was used to determine thecentral position and width of each component in the SN extraction. The spatial region wasintegrated over the FWHM centered on the SN position. When either the blue or red sideis missing, this is represented in the table by ”–”. When a SN occurs in a host core, SN andhost profiles can be degenerate. This is especially the case when the brightness of the hostis comparable to the SN. Since these candidates can be a variable active galactic nucleus, alower priority was put for these SN candidates for spectroscopic followup (Sako et al. 2008),and thus we have few such candidates. For 62 of 66 SN candidates with blue and red sidespectra, the difference of host contamination between the two sides ( | F host,blue − F host,red | ) wasless than 0.2. The difference was larger than 0.4 for the remaining four candidates (SN5751, 12 –16779, 16938, and 17081). Either side of the first three (SN5751, 16779 and 16938) showsa degenerate spatial profile and have large host contamination (around 0.6). The bestfitGaussian width θ SN for the spatial profile of these SNe was larger than the width of theirhost. Our SN extraction method did not work sufficiently. A different subtraction methodbased on the principle component analysis is adapted for these spectra (Sako et al. in prep).The Gaussian width for SN17081 was normal ( θ SN = 1 . . ′′ . From its negative value of F host,blue − F host,red (-0.40), the lowest contamination ratioof these four candidates, we speculate that SN17081 occurred in a red galaxy.For each SN, we measured the average S/N over the entire spectral range in 2 ˚A bins.Figure 5 shows the S/N distribution for the 62 SNe Ia. Most spectra of the sample show S/Nbetween 4 and 10. The very high S/N spectra are those of SNe Ia at z ∼ .
05. Figure 6 givesexamples of SN Ia spectra obtained by the Subaru/FOCAS telescope: two high quality SNIa spectra are shown, a spectrum of a nearby SN Ia at z = 0 . z = 0 . z = 0 . χ . The following are notes on those targets. (i)SN6471. The bestfit type among our template spectra is a SN IIP, matching the SN1999emand SN1992H spectra well. We could not detect emission or absorption lines of this host.The H α emission of SN6471 shows that SN6471 is one of the farthest SNe IIP discovered bythe SDSS-II Survey, z = 0 . β and [O ii ] emission lines of this hostshow that this is one of the farthest SNe discovered by the SDSS-II SN Survey, z = 0 . r band than normal SNe Ia at this redshift. (iii) SN12844. The bestfit type amongour template spectra is SN2005gj, which shows an H α feature evolving with time and calleda Type Ia/IIn SN (Aldering et al. 2006; Prieto et al. 2007). The spectrum of SN12844shows strong emission lines of [O ii ], [O iii ], H β , H α , [N ii ] and [S ii ]. (iv) SN12979. Thebestfit type among our template spectra is an intrinsically faint SN1991bg. (v) SN14475.The bestfit spectrum is SN1998bw (Galama et al. 1998; Patat et al. 2001) at +6 daysafter the maximum date. Other matches are SN1997ef (Iwamoto et al. 2000) and SN2002ap(Gal-Yam, Ofek & Shemmer 2002; Mazzali et al. 2002; Foley et al. 2003). Although all the 13 –details of the absorption lines of the bestfit spectra are not completely identical, they are allhypernovae. Thus it can be said that SN14475 is a possible hypernova. (vi) SN15170. Thebestfit spectrum is the intrinsically bright SN1991T spectrum around the maximum date. We used the SALT2 light curve fitting code developed by the SNLS collaboration (Guyet al. 2007) to derive light curve parameters. This code employs a two dimensional surfacein time and wavelength that describes the temporal evolution of a SN Ia spectrum f ( p, λ ); f ( p, λ ) = x × [ M ( p, λ ) + x M ( p, λ )] × exp( c salt CL ( λ )) , (10)where p is the SN phase in restframe days, the elapsed time from maximum luminosity, M isthe average SED, and x is a normalization. The first-order spectral deviation M is includedin a linear fashion with a coefficient x . We used the spectral surfaces of M i ( i = 0 ,
1) thatwe also used for the cosmological analysis of Kessler et al. (2009a). The parameter x is the normalized deviation from the typical spectrum M and can be approximated with∆ m as ∼ . − . x (Guy et al. 2007). The function CL ( λ ) is the color law of a thirdorder polynomial function monotonically increasing with wavelength and c salt , the deviationfrom the mean SN Ia (B-V) color, is its coefficient which is sensitive to both intrinsic colordiversity and the host-dust reddening.We selected every photometric point of the gri band light curves unless a bad f lag wasassigned; this happened when the photometric scaling factor was low, the rms magnitudeof calibration stars was high, the variation in sky brightness was high, the frame fit qualityfor the SN photometry was poor, or no calibration stars were available in the frame (corre-sponding to the f lag <
256 as recommended in Holtzman et al. (2008)). We did not use u and z band data, since their S/N were not good enough for detailed analysis.The output parameters of this code were rest- B band maximum magnitude m B , which isderived from x , spectral deviation x , color excess c salt , and the date of maximum luminosity t max . In Figure 8, we present the distribution of SN phases at the spectroscopic epoch (upperleft), the redshift (upper right), the spectral deviation x (lower left) and the color excess c salt (lower right) for the Subaru sample. The SN phase is defined as the difference of datesbetween the spectral observation and maximum brightness divided by (1 + z ). We can seethat the range of phases varies from -7 days to +30 days and that the highest number is inthe bin p = 0 − z = 0 . − .
45 bin and it peaks at the z = 0 . − .
30 bin (top right). The lightcurve-width histogram resembles a right rectangle shape, with its oblique side decreasing with the 14 –value of x . The number shortage of SNe Ia with large x values is due to a bias that arosewhen the Subaru targets were selected (left bottom). Most of the sample fall in the colorrange from -0.2 to 0.2. The SALT2 fitting gives large positive values of this parameter forambiguous targets SN12844, SN12979 and SN14475, which are c salt = 0 . ± .
04, 0 . ± . . ± .
03, respectively. These might be cool or dusty SNe.
5. Summary
REFERENCES
Aldering, G., et al. 2002, SPIE, 4836, 61Aldering, G., et al. 2006, ApJ, 650Astier, P., et al. 2006, A&A, 447, 31Baumont, S., et al. 2008, A&A, 491, 567Blondin, S., Walsh, J.R., Leibundgut, B., & Sainton, G. 2005, A&A, 431, 757Bohlin, R.C., Dickinson, M.E. & Calzetti, D. 2001, AJ, 122, 2118Cardelli, J.A., Clayton, G.C., & Mathis, J.S. 1989, ApJ, 345, 245Contreras, et al. 2010, AJ, 139, 519Ellis, R.S., et al. 2008, ApJ, 674, 51Fillipenko, A.V, 1997, ARA&A, 1997, 35Foley, R. J., et al. 2003, PASP, 115, 1220Foley, R.J., et al., 2008, ApJ, 684, 68Foley, R.J., et al. 2010, arXiv: 1010.2749Frieman, J.A., et al. 2008, AJ, 135, 338Fukugita, M., Ichikawa, T., Gunn, J. E., Doi, M., Shimasaku, K., & Schneider, D. P. 1996,AJ, 111, 1748Galama, T.J., et al. 1998, Nature, 395, 670Gal-Yam, A., Ofek, E. O., & Shemmer, O. 2002, MNRAS, 332, L73Gunn, J. E., et al. 1998, AJ, 116, 3040Gunn, J. E., et al. 2006, AJ, 131, 2332 16 –Guy, J., et al. 2007, A&A, 466, 11Hicken, M., et al. 2009, ApJ, 700, 331Holtzman, J.A., et al. 2008, AJ, 136, 2306H¨oflich, P., Wheeler, J. C., & Thielemann, F. K. 1998, ApJ, 495, 617Howell, D.A., et al. 2005, ApJ, 634, 1190Iwamoto, K., et al. 2000, ApJ, 534, 660Jha, S., et al. 2006, ApJ, 131, 527Kashikawa, N., et al. 2002, PASJ, 54, 819Kessler, R., et al. 2009, ApJS, 185, 32Konishi, K., et al. 2011b in prep.Massey, P., Strobel, K., Barnes, J.V. & Anderson, E. 1988, ApJ, 328, 315Matheson, T., et al. 2008, ApJ, 135, 1598Mazzali, P. A., et al. 2002, ApJ, 572, L61Miknaitis, G., et al. 2007, ApJ, 666, 674Oke, J.B. 1990, AJ, 99, 1621¨Ostman, L. et al., 2011, A&A, 526, A28Patat, F., et al., 2001, ApJ, 555, 900Perlmutter, S., 1999, ApJ, 517, 565Prieto, J.L. et al. 2007, arXiv: 0706.408Riess, A. G., et al. 1998, AJ, 116, 1009Sako, M., et al. 2008, AJ, 135, 348Schlegel, D.J., Finkbeiner, D.P. & Davis, M 1998, ApJ, 500, 525Tokita, K. 2009, Ph-D thesis, Univ. of TokyoZheng, C., et al. 2008, AJ, 135, 1766 17 –
This preprint was prepared with the AAS L A TEX macros v5.2.
18 – f l [ e r g s s e c − c m − Å − ] F l u x den s i t y r a t i o f H S T /f I R A F Wavelength [Å]
Fig. 1.—: Comparison of two BD+28 ◦ ◦ C o rr e c t i on f a c t o r Wavelength [Å]
Fig. 2.—: Correction factors for the telluric removal for year 2005 (red) and 2006 (green).There are calibration uncertainties of < S ee i ng [ a r cs e c ] Wavelength [Å]
Fig. 3.—: An example of the seeing-wavelength relation measurement for the Sub-aru/FOCAS instrument. Plotted are the seeings for BD+28 ◦ θ ∝ λ − . ) is shown in red. 21 – (a) r -band image for slit alignment (b) 2D spectrum in space (horizontal) andwavelength (vertical) -5e-19 0 5e-19 1e-18 1.5e-18 2e-18 2.5e-18 70 80 90 100 110 120 130 f l spatial position [pix]SNgal1gal2SN+galdata (c) Integrated spatial profile from a spectrum -1e-18 0 1e-18 2e-18 3e-18 4e-18 5e-18 6e-18 7e-18 8e-18 9e-18 5000 5500 6000 6500 7000 7500 8000 8500 9000 4000 4500 5000 5500 6000 6500 f λ [ e r g s s e c - c m - Å - ] Observed Wavelength [Å] SN16953 Redshift: 0.3387Rest Wavelength [Å] (d) Extracted SN spectrum
Fig. 4.—: An example of supernova spectrum extraction: (a) An r -band image; the SN Ia(SN16953) occurs on the right side of the host galaxy. (b) A part of the 2D spectrum centeredon an emission line: before (left) and after (right) SN extraction. (c) Spatial profile of thespectra: the observed profile (red) is fitted with Gaussians for the SN (black) and galaxy(green) components. The sum is shown with the pink curve. (d) Extracted SN spectrum(black) and uncertainty flux (red). 22 – N u m be r o f S N I a s pe c t r a S/N
Fig. 5.—: Distribution of the S/N for the SN Ia spectra. The S/N is calculated per 2 ˚A. 23 – f λ [ e r g s s e c - c m - Å - ] Observed Wavelength [Å] SN6057 Redshift: 0.0669Rest Wavelength [Å] dseab errorSN photometry (a) r -band image for slit alignment f λ [ e r g s s e c - c m - Å - ] Observed Wavelength [Å] SN12977 Redshift: 0.2474Rest Wavelength [Å] dseab errorSN photometry (b) 2D spectrum in space (horizontal) andwavelength (vertical) -3e-17-2e-17-1e-17 0 1e-17 2e-17 3e-17 4000 5000 6000 7000 8000 9000 3000 3500 4000 4500 5000 5500 6000 f λ [ e r g s s e c - c m - Å - ] Observed Wavelength [Å] SN16779 Redshift: 0.3983Rest Wavelength [Å] deab error (c) Integrated spatial profile from a spectrum (d) Extracted SN spectrum
Fig. 6.—: Examples of SN Ia spectra and their uncertainties obtained by the Subaru/FOCASinstrument: two high quality SN Ia spectra: late-time (one month after the maximum date)spectrum of a nearby SN Ia z = 0 . z = 0 . z = 0 . gri magnitudes can be interpolated at the spectral date (labeled as “dseab”in the figures; g , r and i magnitudes are shown in blue, green and red colors) or not (labeledas “deab” in the figure). 24 – N u m be r o f ob j e c t s Redshift
Fig. 7.—: Redshift distribution of the 71 observed objects with the Subaru telescope. 25 – N u m be r s Phase [rest-frame days] 051015200 0.1 0.2 0.3 0.4 N u m be r o f S N e I a Redshift05101520 -4 -2 0 2 4 N u m be r o f S N e I a X -1 -0.5 0 0.5 1 05101520 N u m be r o f S N e I a c salt Fig. 8.—: SN phase (top left), redshift (top right), spectral deviation x (bottom left) andcolor c salt (bottom right) distributions for the Subaru sample. 26 –Table 1. Instrument Configurations
27 –Table 2. The spectrophotometric standard starsName (grism & filter) Date ZD (deg) Exposure (s)2005BD+28 ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ a (300R & L600) Oct26 55.3 4.0BD+28 ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ a Massey et al. (1988)Feige 110 spec50cal a Massey et al. (1988)GD 71 calspec b Bohlin, Dickinson & Calzetti (2001)BD+75 ◦
325 oke1990 c Oke (1990)Feige 34 spec50cal a Massey et al. (1988) a This directory is under $onedstd/ in IRAF b c This directory is under $onedstd/ in IRAF. We used this spectrumafter a magnitude correction by +0.004mag. (Oke 1990) 30 –Table 4. Emission and absorption lines
Lines Wavelength (˚A)Balmer linesH β α ii ] 3727.092, 3729.875[O iii ] 4364.436, 4932.60, 4960.295, 5008.240[O i ] 6302.046[N ii ] 6549.86, 6585.27[S ii ] 6718.29, 6732.67Ca H&K 3934.777, 3969.588Mgb 5176.7NaD 5895.6
31 –Table 5. Supernova Candidates observed by Subaru/FOCAS
SDSS ID IAU name a Type b RA Dec E ( B − V ) MW MJD c Redshift d Airmass e F host /F all f
32 –Table 5—Continued
SDSS ID IAU name a Type b RA Dec E ( B − V ) MW MJD c Redshift d Airmass e F host /F all f a IAU names were not attached to five SNe due to on-site analysis. b “05gj?”, “91bg?”, “hyp?”, “91T?” indicates that the object resembles 2005gj-like SN Ia, 1991bg-like SN Ia, hypernova, 1991T-likeSN Ia. c The observational mid-date. The red part of a spectrum was first observed and the blue part followed for most of the cases. d Heliocentric redshift. The redshift of the AGN (SN13346) was measured from CIII] 1908.73 ˚A, Mg ii ii α emission line of the SN IIP. e Mid airmass. The red part of a spectrum was first observed and the blue part followed for most of the cases. f Host contaminations for blue and red side spectrum, respectively.
33 –Table 6. Observational Summary of FOCAS Followup Campaign
Classification Number CommentsSupernova- normal Ia 59 (= 31 + 28) 2005/2006yr incl. non-IAU named Ia (3 a )- peculiar Ia 3 SN12844 b , SN12979 c , SN15170 d - type IIn 7 SN10450 (peculiar IIn)Probable Hypernova 1 SN14475 e AGN 1 SN13346 a SN1166, SN1686, SN1688 b SN 2005gj-like Spectrum. Large c salt value. c SN 1991bg-like Spectrum. The ri -band lightcurves are well-fitted by the SALT2 code,however, the g -band brightness seems almost constant during 40 days after the B -bandmaximum date. d SN 1991T-like spectrum. The lightcurve is normal. e Peculiar spectrum. Large c saltsalt