Cosmic Flows: Green Bank Telescope and Parkes H i observations

Monthly Notices of the Royal Astronomical Society, Jul 2011

The neutral hydrogen properties of 1822 galaxies are being studied with the Green Bank 100-m and the Parkes 64-m telescopes as part of the ‘Cosmic Flows’ programme. Observed parameters include systemic velocities, profile linewidths and integrated fluxes. The linewidth information can be combined with the optical and infrared photometry to obtain distances. The 1822 H i observations complement an inventory of archives. All told, H i linewidth information is available for almost all of five samples: (i) luminosity–linewidth correlation calibrators; (ii) zero-point calibrators for the Type Ia supernova scale; (iii) a dense local sample of spiral galaxies with within 3000 km s−1; (iv) a sparser sample of 60-μm selected galaxies within 6000 km s−1 that provides an all-sky coverage of our extended supercluster complex; and (v) an even sparser sample of flat galaxies, extreme edge-on spirals, extending in a volume out to 12 000 km s−1. The H i information for 13 941 galaxies, whether from the archives or acquired as part of the Cosmic Flows observational programme, is uniformly re-measured and made available through the Extragalactic Distance Database website.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:

Cosmic Flows: Green Bank Telescope and Parkes H i observations

H el e`ne M. Courtois 2 3 R. Brent Tully 2 D. I. Makarov 1 3 S. Mitronova 1 B. Koribalski 5 I. D. Karachentsev 1 3 J. Richard Fisher 4 0 520 Edgemont Road, Charlottesville, VA 22903, USA 1 Special Astrophysical Observatory, Russian Academy of Sciences , N. Arkhyz, KChR, 369167, Russia 2 Institute for Astronomy, University of Hawaii , 2680 Woodlawn Drive, Honolulu, HI 96822, USA 3 Universite Lyon 1 , CNRS/IN2P3/INSU, Institut de Physique Nucleaire , Lyon, France 4 National Radio Astronomy Observatory 5 Australian Telescope National facility , CSIRO, PO Box 76, Epping NSW 1710, Australia A B S T R A C T The neutral hydrogen properties of 1822 galaxies are being studied with the Green Bank 100-m and the Parkes 64-m telescopes as part of the 'Cosmic Flows' programme. Observed parameters include systemic velocities, profile linewidths and integrated fluxes. The linewidth information can be combined with the optical and infrared photometry to obtain distances. The 1822 H I observations complement an inventory of archives. All told, H I linewidth information is available for almost all of five samples: (i) luminosity-linewidth correlation calibrators; (ii) zero-point calibrators for the Type Ia supernova scale; (iii) a dense local sample of spiral galaxies with MKs < 21 within 3000 km s1; (iv) a sparser sample of 60-m selected galaxies within 6000 km s1 that provides an all-sky coverage of our extended supercluster complex; and (v) an even sparser sample of flat galaxies, extreme edge-on spirals, extending in a volume out to 12 000 km s1. The H I information for 13 941 galaxies, whether from the archives or acquired as part of the Cosmic Flows observational programme, is uniformly re-measured and made available through the Extragalactic Distance Database website. 1 I N T R O D U C T I O N Cosmography is the study of the large-scale structure of the universe. A complete analysis involves observational and interpretive components. With spectroscopic information, a complete sample of galaxies within specified limits describes the structure in the redshift space. If distance measures are available for at least some fraction of the spectroscopic sample, then steps can be taken to transform to the physical space. Distance measures allow for the separation of redshifts into cosmic expansion and deviant (or peculiar) components. Cosmological simulations and orbit reconstructions provide tools for the recovery of information about the underlying distribution of matter from a map of peculiar velocities. The influence of the dark sector on galaxy motions can be studied from 1 Mpc, the scale of collapse, to 150 Mpc, the largest scale of useful peculiar velocity measures. After great enthusiasm for what could be learned from peculiar velocity studies in the 1990s (Willick & Strauss 1998; Courteau et al. 2000), progress slowed primarily because of the challenge presented by the need for much more and better quality data. If the goal is to have a dense grid of distance measures to a depth dominated by the Hubble expansion, then most methodologies for estimating distances have inadequacies. The Cepheid periodluminosity (Freedman et al. 2001) and Tip of the Red Giant Branch (TRGB) (Rizzi et al. 2007) methods have limited successes. The surface brightness fluctuation (SBF) (Tonry et al. 2001) and Fundamental Plane (Colless et al. 2001) methods apply to luminous early-type galaxies that are poorly represented in low-density regions. The Type Ia supernova (SNIa; Jha, Riess & Kirshner 2007) method rests on the serendipity, resulting in an accurate but sparse map of distances. The one well-established methodology that can provide decent distances with high density over an appropriately large volume is provided by the correlation between the galaxy luminosity and rotation rate, the TullyFisher (TF) relation (Tully & Fisher 1977). Two observations are needed to apply this method: a spectroscopic measure of the rotation rate, most expeditiously accomplished by observing the linewidth of a 21-cm neutral hydrogen profile, and the surface photometry at optical or infrared bands that monitor old star populations. New capabilities with the facilities for both the spectroscopy and the photometry are revolutionizing our capabilities. On the spectroscopic side, the new capabilities are both realized and promised. We will describe observations with the 100-m Robert C Byrd Green Bank Telescope at the National Radio Astronomy Observatory (NRAO-GBT) and with the 13-channel multibeam receiver on the 64-m Parkes telescope. Our programme also makes use of archival data from ongoing multibeam observations with the Arecibo telescope (Giovanelli et al. 2005). The promised capabilities are those that will accrue with the wide-field interferometric surveys that will cover the entire Southern (ASKAP, MeerKAT) and Northern hemispheres (Apertif, LOFAR) with unprecedented sensitivities. As for the photometry, the ground technology revolution comes from the large multiband surveys from new wide-field CCD camera systems: Pan-STARRS in the north and SKYMAPPER in the south. Also the satellites Spitzer and WISE are providing unprecedentedly accurate surface photometry of galaxies in the mid-infrared. The data must be compared with analytical or numerical models. The theoretical tools have been developed tremendously in the last 20 yr from small N-body programmes to extremely large hydrodynamical simulations that have culminated with such data products as the Millennium Simulation Project (Springel et al. 2005). The advances have focused on increasing the computational speed, the number of particles and parallelism in the codes. The field is now embracing higher levels of refinement in the constraints on initial conditions, coming from the observations. The theoretical universe being built in the new generation of simulations must implement the observational evidence with a high level of detail. In particular, on scales of 1150 Mpc, there is a tension between the observed luminous matter distribution and the complexity of galaxy flows, both in directions and in amplitudes that must be reconciled within theoretical constructs. New developments in analysis methods (Zaroubi et al. 1995) use the information in the gradients and convergence of flows to recover the mass distribution information that extends beyond the observational data. There is an encouraging synergy in the advances on both observational and theoretical fronts. This conjunction of the observational and theoretical progress creates a special opportunity for the emergence of the Cosmic Flows programme that was proposed to the NRAO as a large programme of observations of the H I 21-cm line of galaxies with the GBT. It is ongoing since 2007 and already has been scheduled for more than 1000 h on the sky. The radio observational programme is extended to the full sky with access to the most southern targets using the ATNF-CSIRO Parkes 64-m radiotelescope. An accompanying photometry programme is designed to provide the surface photometry of the targets. The theoretical programme accompanying the observational programme is using numerical action methods (Peebles et al. 2001) and numerical constrained simulations: CLUES (Gottl ober, Hoffman & Yepes 2010) . This paper will focus on (1) a presentation of the five complementary data samples of the Cosmic Flows project; (2) a description of the pipeline developed for the consistent measurement of tens of thousands of H I profiles (in anticipation of the forthcoming large surveys); and (3) the release of the currently accrued radio H I material associated with our five samples. 2 C O S M I C F L OW S P R O G R A M M E : F I V E G A L A X Y S A M P L E S The goal of the Cosmic Flows programme is to obtain an all-sky grid of galaxy distances as dense and deep as current capabilities allow and to complement the observational endeavour with collaborative theoretical studies to try to gain a more rigorous understanding of the local dark sector ( The volume of the universe that is currently targeted is bounded by the sensitivity of the GBT and the Parkes single-dish radiotelescope at a practical limit of the 6000 km s1 radius. It is expected that the next generation of the Cosmic Flows programme will be testing a volume up to 15 000 km s1 using the wide-field interferometric technology with blind H I all-sky surveys. For the moment, the competition for the telescope time makes it impractical to target even the 7000 appropriately edge-on spiral systems with measured redshifts within 6000 km s1. Our compromise has been to distinguish three discrete samples and to strive for a high level of completion with each of the three components. One sample, we refer to as V3k, strives to provide a high-density mapping of the volume within 3000 km s1. The second sample selected by the far-infrared flux, IRAS Point Source Catalog - Redshift (PSCz), extends to 6000 km s1 and gives a good coverage at low Galactic latitudes. The third sample of extreme edge-on systems, Revised Flat Galaxy Catalog (RFGC), provides a sparse coverage over a large volume. Two smaller samples containing several hundred targets were observed for the calibration purpose. One provides the slope and zero-point calibrations for the TF relation, while the other provides a scale coupling with the SNIa method. These five samples are described in more detail below. While our new observations were focused on the five samples to be described, our archival searches have been unrestricted. We have gathered all available digital H I spectra and coherently remeasured the H I profiles for 16 004 spectra of 13 941 galaxies, out of which 11 074 galaxies have an adequate measurement to derive distances. Our observations were thus targeting only the galaxies within our specified samples and without adequate H I profile measurements. These limitations have allowed us to essentially complete the coverage of our samples with about 1200 observed targets in 3 yr. As points of comparison, in recent discussions based on literature data, Tully et al. (2008) is based on 1791 distances within 3000 km s1, including 1252 TF measures, and Feldman, Watkins & Hudson (2010) is based on 4536 distances dominated by the TF measures of Springob et al. (2005). Our goal was not merely to incrementally augment the literature information. The current procedures involve a new definition of the H I linewidth and, as a consequence, require new measurements of 100 per cent of the profiles and a complete recalibration of the TF relation. The number of calibrators has increased by more than 100 per cent since the previous extensive calibration of Tully & Pierce (2000), giving an additional good reason to reconsider the calibration. The new procedures and calibration will then be applied to the samples of field galaxies. We measure the integrated flux of the H I linewidth and then derive the profile width at 50 per cent of this cumulative flux. This method requires access to electronic profiles but then enables the recomputation of all the available profiles in all archives. Our procedures have been compared with similar procedures by the Cornell group and with an accurate non-digital sample (Courtois et al. 2009) with satisfactory results. 2.1 Calibrators sample The luminositylinewidth calibration divides into two parts: the first is related to the slope of the correlation and the second is related to the determination of the zero-point. The determination of the slope of the luminositylinewidth relation is critical for the minimization of systematic errors due to one of the two Malmquist biases: the effect on errors that can result from magnitude-limited samples (Malmquist 1924; Teerikorpi 1984; Sandage 1994; Willick 1994).1 A maximum-likelihood procedure such as that employed by Giovanelli et al. (1997) can be used that provides a basis for corrections for the bias if there is an accurate understanding of sources of errors. Instead, our procedure, discussed by Tully & Pierce (2000), does not require a detailed understanding of errors to correct biases but, rather, acts to null the bias. Individual distance measures have uncertainties but are not systematically offset from true distances. An essential ingredient to the recipe to the null bias is a slope fit to the luminositylinewidth correlation that is insensitive to magnitude limits. The desired calibration is achieved through the compilation of a cluster template created by successively combining data from individual clusters with offsets in the magnitude reflecting their different moduli. The calibration is achieved with the consideration of 370 galaxies in 13 clusters of diverse morphologies with distances ranging from 15 to 100 Mpc. Each cluster contributes 15 70 galaxies (median of 25) to the template. Each cluster sample is only magnitude limited, and not limited otherwise, in a parameter that constrains the fit for a derivation of distances.2 The absolute zero-point calibration of the template relation is provided by 39 galaxies that pass the same inclination, type and luminosity criteria as the cluster calibrators and have accurately known distances from external measurements. These calibration distances are based on Hubble Space Telescope observations of either Cepheid stars or the luminosities of stars at the TRGB, with the scale set by the HST Distance Scale Key Project (Freedman et al. 2001). In Tully et al. (2008), they showed a preliminary result demonstrating that TRGB, SBF and luminositylinewidth (TF) distances are all on a consistent zero-point scale with those established by Cepheid measurements. The previous zero-point calibration in this collaboration was based on 24 galaxies (Tully & Pierce 2000), so a new calibration is possible using 60 per cent more galaxies. To complete these calibrator samples, we observed an additional 165 galaxies with no previous satisfactory measurements with the NRAO-GBT Large Program. We now have digital data for the entire sample. The linewidth and flux analysis that was described in Courtois et al. (2009) was performed in a consistent way. The clus 1 There is confusion because Malmquist (1920, 1922, 1924) discussed pos sible biases that affect galaxy distance measurements in two distinct ways. Suppose individual galaxy distances are unbiased but have errors. Since there are an increasing number of sources at larger distances, a given shell in a measured distance will be populated by more galaxies with larger true distances scattered inwards than smaller true distances scattered outwards. Lynden-Bell et al. (1988) drew attention to this effect in the context of galaxy distances and flows and subsequently these authors have referred to it as the EddingtonMalmquist bias in recognition of a prior contribution by Eddington 1913. The other relevant effect is caused by a magnitude limit. For galaxies at the same distance, intrinsically fainter systems can be lost from a sample, while intrinsically brighter systems are retained. Strauss & Willick 1995, in their extensive discussion of the subject, referred to the latter effect as a selection bias and identified the Malmquist bias with the EddingtonMalmquist effect. In this paper, we will continue to use the term Malmquist bias in connection with the magnitude-selection problem because of the past popular usage and because for what we are doing it tends to be of greater concern. 2 Galaxies with morphological types Sa and earlier are excluded. This selec tion has a qualitative aspect that could introduce a small bias. Also, potential candidates are excluded if there is confusion in the radiotelescope full width at half-maximum (FWHM) beam or strong evidence of tidal disruption. Galaxies are excluded if they are more face-on than 45 but this property is not intrinsically correlated with the distance and tests have not suggested any distance bias with our inferred inclinations. ter slope calibration sample with accurate linewidths has reached 326 galaxies, an improvement of more than 100 per cent compared to the 155 galaxies in 12 clusters in Tully & Pierce (2000). 2.2 SNIa host galaxy sample The SNIa host galaxy sample is designed to provide a confident link between the local scales probed by luminositylinewidth distances and dominated by peculiar motions and the cosmological scales probed by SNIa distances. To build this sample, we extracted from the literature spiral edge-on galaxies suitable for the TF relation and that have hosted a SNIa. The SNIa must have a well-sampled light curve and an accurate distance measurement. We draw samples from Jha et al. (2007) and Tonry et al. (2003). With the GBT observation programme, we completed the H I observations of a sample of 54 galaxies which should provide a robust intercalibration of the luminositylinewidth and SNIa distance scales. 2.3 V3k sample A special focus on the region within Vhelio = 3300 km s1 derives from three considerations. First of all, this velocity bounds the structure that, historically, has been called the local supercluster (de Vaucouleurs 1953) with the Virgo cluster at its core and the Fornax cluster as a secondary feature well within the boundary. Secondly, with velocity measures of 15 per cent accuracy, peculiar velocities can be separated from expansion velocities with uncertainties of less than a few hundred km s1 for individual objects and errors in collective motions of relatively small groups of galaxies can be brought below 100 km s1. Thirdly, essentially all galaxies that might be typed SbSd can be easily detected in H I at a high signalto-noise ratio anywhere in the sky if they lie within 3300 km s1. Our rigorous V3k sample is defined by the following criteria: (i) Vhelio < 3300 km s1 (ii) MKs < 21 (iii) Inclination from the axial ratio of greater than 45 (iv) Type later than Sa (v) No pronounced evidence of the tidal disruption (vi) H I signal not confused from a second galaxy. The clip in magnitude at MKs = 21 minimizes selection problems. The Ks (short K) magnitudes are drawn from the 2MASS Extended Galaxy Catalog and Large Galaxy Atlas (Jarrett et al. 2003), assuring that the uniform coverage of the sky remains relatively unaffected by the obscuration down to low Galactic latitudes. Absolute magnitudes are derived from velocities and a non-parametric model of galaxy motions (Shaya, Peebles & Tully 1995). Since the faintest galaxies have Ks = 12 and are gas-rich spirals, the sample is quite complete at high latitudes and all candidates are easily detectable in H I with the GBT. A velocity histogram of the 1228 galaxies in V3k is shown in Fig. 1. The original V3k sample contained 1228 galaxies. Of these, 993 galaxies have now measured H I spectra of adequate quality and, of these, 852 have consistently measured digital profiles. The Cosmic Flows observations provided 292 new accurate measurements, while 560 accurate measurements could be recovered from the re-analysis of the archive material. The 135 galaxies that were dropped from the initial sample frequently have a companion galaxy in the area included in the FWHM beam size of the radiotelescope or an inclination that appeared on inspection to be less than our limit of 45. The region within 3300 km s1 is being populated by many other galaxies which are not included in the V3k sample, yet suitable for the TF relation. The two samples that will be discussed next make additional contributions to this volume of 639 galaxies (PSCz sample) and 440 galaxies (Flat Galaxies). Including miscellaneous contributions, we presently explore the inner 3300 km s1 region with 2304 TF distance measures. 2.4 PSCz sample We were motivated by two considerations in defining a sample that extends beyond V3k. First, we now know that our local supercluster is just an appendage on a larger supercluster complex that includes the Norma, Centaurus and Hydra clusters as well as several other clusters, such as A3537, A3565 and A3574, all in the so-called Great Attractor region (Dressler et al. 1987). By extending to 6000 km s1, we encompass the main part of the overdense region that includes our Galaxy. It also includes the adjacent PiscesPerseus filament (Haynes & Giovanelli 1988). Our second consideration is the current capability of radiotelescopes. Satisfactory H I profiles can be obtained for most spiral galaxies within 6000 km s1 with integrations of less than an hour with the GBT. The Arecibo telescope is more sensitive but accesses only a modest fraction of the sky. The Parkes telescope is less sensitive and longer integrations are sometimes required but this facility is only required to cover 15 per cent of the sky below 45. In an effort to generate an independent sample that isolates normal spirals and is uniform around the sky, attention was given to the PSCz 0.6-Jy survey (Saunders et al. 2000). Extragalactic sources with cool far-infrared emission (100-m flux greater than 60-m flux) are typically normal spirals near morphological type Sc. Targets can be selected to low Galactic latitudes limited by the source crowding. We selected an all-sky sample of 1690 targets by adopting the following criteria: (i) Vcmb < 6000 km s1 (ii) IRAS S60 > 0.6 Jy and S100/S60 > 1 (iii) Inclination from the axial ratio of greater than 45 (iv) Type later than Sa (v) Not tidally disturbed (vi) An H I signal that is not confused Adequate H I profiles have been acquired for 1204 galaxies in this sample. The status of observations is shown in Fig. 1 where the white histogram describes the initial PSCz sample, the black histogram describes the totality of systems with adequate H I profile measurements and the grey histogram describes the subset of 470 systems with new H I information from the Cosmic Flows observations. 2.5 Flat Galaxy sample A dominant source of error in the TF methodology arises from inclination measurements. A characteristic uncertainty is 5, a contribution of half the error budget through the correction of linewidths towards the face-on limit of 45. Worse, there is a tail of large errors with photometrically derived inclinations. The Flat Galaxy catalogues compiled alternatively from optical and near-infrared images [RFGC: Karachentsev et al. 1999; 2MASS-selected Flat galaxy Catalog (2MFGC): Mitronova et al. 2004] provide appealing lists that circumvent this problem. The extreme axial ratios of the Flat Galaxy sample assures that candidates are being viewed almost edge-on; so, uncertainties in the deprojection of circular velocities in a disc are negligible. With the optically selected component, the axial ratio requirement of a major-to-minor axial ratio greater than 7 assures that targets have a morphology around class Sc, since earlier and later types are intrinsically less thin. Obscuration issues can be minimized by going to the infrared for the required photometry. The restriction to extreme edge-on limits the coverage density but results in a coherent and tractable sample for the coverage of a large volume. Our GBT targets from the RFGC were restricted to north of = 40 but outside the range accessed by the Arecibo telescope. Remaining targets within the Arecibo range will either be satisfactorily observed with the ALFALFA survey or later with pointed observations with the Arecibo telescope. Remaining targets at < 40 will be observed with the Parkes telescope. In one GBT semester, we were able to complete most of the missing observations for RFGC galaxies with a known radial velocity less than 6000 km s1. Some targets were added to fill in the allocated time-windows drawn from the 2MFGC (Mitronova et al. 2004) or from a list of nearby galaxies for which no modern digital spectrum is available. A tiny fraction of potential targets were a priori discarded, since RFGC galaxies are usually isolated; thus, there were a very low 2 per cent rate of confusion in the radiotelescope beam or gross distortion of the candidate from the tidal disruption. The non-detection rate was of 14 per cent: 83 galaxies out of 577 Figure 2. Histogram of sample components as a function of the velocity. The V3k sample is illustrated by the filled histogram, the component of the PSCz sample at 30006000 km s1 is illustrated by the open hashed histogram and the fraction of the Flat Galaxy sample with satisfactory H I profile information is illustrated by the denser hashed histogram. The cumulative samples are outlined by the dotted histogram. The open solid histogram describes the current accumulation of sources with adequate H I profile information. were not detected at our flux limit. More accurately, 12 per cent of the RGFC and 23 per cent of the 2MFGC galaxies were not detected. The high detection rate is inherent to the gas-rich nature of the Flat Galaxy sample, while the 2MFGC is biased toward 2MASS earlier-type galaxies. The RFGC original sample contains 4444 galaxies, out of which 2788 have a published redshift (white histogram in the bottom left-hand panel of Fig. 1). Out of these 2788 galaxies, we now have an accurate linewidth measurement for 1229 galaxies (black histogram). The Cosmic Flows observations provided 323 (white histogram) new accurate measurements. Two figures summarize the properties of various samples. Fig. 2 gives combined histograms of the samples. There are large numbers of sources within the formal samples and large numbers of additional sources with good H I profile measurements within 3000 km s1. At 30006000 km s1, there are comparable numbers of sources to the inner 3000 km s1, although spread over a larger volume. Beyond, there is a tail of rapidly decreasing coverage that extends to 12 000 km s1. The distribution of many of these sources can be seen in the two slices of a three-dimensional cube in Fig. 3. It is seen that the nested samples provide good coverage of our immediate overdense region but falls off rapidly beyond the edges of the structure that we live in. 3 O B S E RVAT I O N S W I T H T H E G B T To strive for the completion of the V3k, PSCz and Flat Galaxy samples, we have been observing with the 100-m GBT at declinations above = 45, the southern limit with the GBT, but excluding the Arecibo range 0 < < +38 where data are currently acquired within the ALFALFA project. Our present strategy is to wait for the results of the ALFALFA multibeam survey currently in progress. Currently, ALFALFA data releases 13 have been processed through our pipeline. If profiles are inadequate, then we will request deeper observations of those targets in the Arecibo range of declinations. Access to the remaining sky, at < 45, requires observations with the Parkes telescope in Australia. Sources are not observed if digital spectra of good quality exist. The single-beam Robert C. Byrd GBT observations released in this paper are identified as ctf2009 and ctm2010 in our All Digital HI (ADHI) catalogue. Observations were carried out from 2007 February to 2010 June with project identifiers: 07A039 (55 h), 07C067 (218 h), 08A072 (47 h), 08B041 (50 h), 08C010 (340 h) and 10A059 (370 h). The total of the observing time allocated to the project at the GBT was 1080 h. The GBT has implemented a Legacy ID numbering for the public access of the data; our project data can thus be recovered under the Legacy IDs GF13, GC47, GC60, GC67, GC69 and GC102. For the observations we use the single-beam L-band (12 GHz) receiver and the spectral line spectrometer as the back-end detector. The final spectrum is stored with the 1.6 km s1 resolution. It was usually binned at least once to the 3.2 km s1 resolution for the H I linewidth measurement. 4 O B S E RVAT I O N S W I T H T H E PA R K E S T E L E S C O P E Southern targets < 45 for the two samples V3k and PSCz are currently being observed with the Parkes telescope. The Parkes 64-m radiotelescope observations released in this paper were carried out in 2009 February 110: P660: 60 h on the sky. We were able to observe 58 galaxies with known radial velocity less than 4000 km s1. We obtained 33 spectra leading to accurate linewidth measurements and 24 are inadequate for distance measurements. A typical exposure time for a target at 3000 km s1 was 1.5 h or more. The non-detection rate was 16 per cent: nine galaxies out of 58 were not detected. Fig. 4 illustrates the general properties of the H I profiles that were adequately detected in the course of the Cosmic Flows Large Program with the GBT and the supplement with the Parkes telescope. 5 A N A LY S I S O F H I L I N E P R O F I L E S With the award of our project as an NRAO GBT Large Program, we reconsidered the traditional ways of measuring the H I linewidth. We were looking for a robust method that could be applied to both low- and high-signal-to-noise-ratio spectra. We also wanted to be able to remeasure a large quantity of archive H I data from various telescopes, so the method would be as automated as possible. Historically, Tully and Fisher measured linewidths W20 at 20 per cent of the peak intensity (Tully & Fisher 1977). However, the 20 per cent level is frequently close to the noise level leading to spurious measurements. Also the 20 per cent level is difficult to secure when profiles are strongly asymmetric. We now prefer to measure the width enclosing 50 per cent of the cumulative H I line flux, Wm50. Specifically, Wm50 is the linewidth at a flux level that is 50 per cent of the mean flux averaged in channels within the wavelength range enclosing 90 per cent of the total integrated flux. The exclusion of 5 per cent of the flux at the high- and low-wavelength edges minimizes problems in the profile wings. The Wm50 linewidths are measured at slightly higher flux levels than W20. The random scatter between these two measures is the lowest found among alternatives that were examined. The parameter Wm50 is an empirical measure of the width of an H I profile. We correct for the redshift and instrumental broadening with the formula Figure 4. Histograms of the galaxy properties of the 1423 suitable profiles obtained in the Cosmic Flows programme in H I at the GBT and Parkes telescope: (a) heliocentric recession velocity in km s1; (b) H I linewidth at half power (Wm50); and (c) logarithm integrated flux; and (d) logarithm of the signal-to-noise ratio. where z is the redshift, v is the smoothed spectral resolution and = 0.25 is an empirically determined constant. The observed linewidth can also be adjusted by separating out the broadening from turbulent motions and offsetting to produce an approximation to 2 Vmax, where Vmax characterizes the rotation rate over the main body of a galaxy. We have defined the parameter Wmix, where 2Wm50Wt,m50 1 e(Wm50/Wc,m50)2 with Wc,m50 = 100 km s1 and Wt,m50 = 9 km s1. Then Wmix = Wmx/sin (i), where i is the galaxy inclination from face-on. Details regarding the Wm50 and Wmix linewidth parameters and comparisons with alternatives are given by Courtois et al. (2009). All modern spectra are available in a digital form. Consequently, it is straightforward for us to pass archival data through the same analysis procedure as we apply to our newly observed material. It, of course, means that we do not need to observe a galaxy if we can access adequate observations from any archive. The primary sources of data in our current catalogue originate, in decreasing order, from Arecibo (7898), Nancay (3439), GBT (1444), Parkes (1052), the old NRAO 300 (1059), the 140 (696) and Effelsberg (235), see Tables 2 and 3. The assignment of errors for Wm50 or quality flags is a particularly challenging problem. Formally, we link errors to the profile signal-to-rms noise ratio as described by Courtois et al. (2009). From comparisons between alternative observations, our cited errors appear to be conservative, about 50 per cent larger than a 1 value. Also on the side of being conservative, we assume that the linewidth uncertainty is 8 km s1 in the best of cases. Estimates are difficult to quantify precisely because the error on Wm50 is linked to three characteristics: the signal-to-noise ratio, observational flux limit and shape of the H I line with the flux in the wings that may or may not be real. Our interest is to use the profile widths as a parameter in the measurement of distances. Even if our errors do not have a precise absolute sense, they do have a relative sense. Within our system, we consider that there is a threshold of acceptability: a profile may be of sufficient quality to be used in the determination of a distance or it may not be. We link our error estimate to this threshold. Specifically, an adequate profile is assigned an error of less than or equal to 20 km s1. Inadequate profiles are identified by errors greater than 20 km s1 (see examples in Fig. 5). Confused profiles are identified by the error flag of 100 km s1. Non-detections are identified by the error flag of 500 km s1. Since for distance measurement purposes we will not use any profile with an error larger than 20 km s1, measurements with such large errors are not taken into account in the averaged Wmx_av column of our ADHI catalogue. Thus, if a galaxy has only one measurement and it is considered inadequate, then there will be a value given for Wm50 among the parameters measured from the specific source profile but no value will be registered for Wmx_av (see e.g. line 2 of Table 1). In our compilation of digital spectra, some galaxies have been observed by two to three different telescopes. Intercomparisons between different H I observations of the same galaxies suggest that the characteristic accuracy of an individual acceptable profile width is 7 km s1. Our H I observations at the GBT and Parkes 64-m telescope are made available in electronic form with the online version of this article (see Supporting Information). A few lines are given in Table 1 as a sample. The columns (from the left-hand to right-hand side) are PGC name, source of observations, telescope, heliocentric velocity, W m50, error on W m50, signal-to-noise ratio and integrated flux in the line. 6 L I N K I N G A N A L O G A N D D I G I TA L L I N E W I D T H M E A S U R E M E N T S Even though there are now digital spectra for most galaxies, the collection remains incomplete. Indeed, it will be difficult to achieve completion. Some very nearby galaxies are much larger than the FWHM beam sizes of modern radiotelescopes. They could be mapped. Better, integrated profiles can be reconstructed from observations with the Westerbork, Very Large Array or Australia Telescope interferometers. Alternatively, profiles obtained in earlier days with smaller telescopes like the NRAO 140 or the Dwingeloo facilities might be the best available. A considerable effort was made in earlier years to accumulate a consistent compilation of analog linewidths. The contributing observations were made with many telescopes and reported by many PGC0000094 PGC0000129 PGC0000218 PGC0000929 PGC0003743 Signal-to-noise ratio Integrated flux (Jy km s1) 3 4 22 9 10 14.5 21 9 Literature source Koribalski et al. (2004) Springob et al. (2005) Huchtmeier et al. (2005) Theureau et al. (2006) Giovanelli et al. (2007) Saintonge et al. (2008) Kent et al. (2008) Courtois et al. (2009) Courtois et al. (this paper) sources, but all W20 measures in our compilation were coherently measured by only three people: J. R. Fisher, R. B. Tully and C. Hall. A comparison was made between analog W20 measures and our new digital Wm50 measures by Courtois et al. (2009). In applications related to distance determinations, we will be more interested in the linewidth parameters related to the physical property, Vmax, the characteristic rotation velocity across the disc of a galaxy. To this end, we now compare the parameters that approximate 2 Vmax sin (i) for the separate analog and digital samples. This comparison is shown in Fig. 6. Here, the analog parameter WR is calculated using a constant value Wt,20 = 22 km s1 in the equivalent to equation (2) for the case of the analog transformation using W20. This value for the thermal broadening constant is different from the value Wto,l2d0 = 38 km s1 originally advocated (Tully & Fouque 1985) for the transformation to WR. For a discussion of this change, see Courtois et al. (2009). Based on 1755 galaxies with good analog and digital profiles, we find a relation that allows us to transform from analog to digital parameters: Wmx = 1.015WR 11.25. The rms scatter is 10 km s1, suggesting uncertainties of roughly 7 km s1 in each of the measured parameters. It is seen that there is a slight slope to the difference between the alternative linewidth measures WR and Wmx which, in principle, are both approximating 2 Vmax sin (i). The two measures are essentially 2 Vmax alternatively with the old analog measurements and the new digital equal at large linewidths, but there is a mean difference that grows to 10 km s1 for the smallest linewidths. Because of this systematic difference between WR and Wmx, it is essential that the relation between the galaxy luminosity and linewidth (the TF relation) be calibrated for the specific linewidth measure that is used. Whatever be the linewidth measure and to whatever degree it may approximate a physical parameter, it remains a subject to observational vagaries. 7 H I C ATA L O G U E I N T H E E X T R AG A L AC T I C D I S TA N C E D ATA B A S E Our new observations from the Cosmic Flows programme and archival information are combined. Tabular information and lineprofile plots are provided for 13 941 galaxies in the ADHI catalogue in the Extragalactic Distance Database (EDD). Intensity velocity ASCII tables are provided in the cases of the material from the Cosmic Flows observational programme. Currently, the catalogue contains 16 004 H I profiles, including 1859 new profiles added through this programme. There are 11 074 profiles that are deemed acceptable, that is, with uncertainties 20 km s1. For 1339 galaxies, there are at least two acceptable profiles and in 82 cases there are three acceptable profiles. Additions are continually being made to the catalogue. Fig. 7 provides a graphic summary of aspects of the ADHI catalogue. The redshift distribution in the top left-hand panel contains peaks at 1800 and 5000 km s1. The two peaks are partially due to the development of the sample with the increasing capabilities of telescopes, with an early emphasis on the region within 3000 km s1, and partially due to a reflection of the distribution of nearby structure. In the top right-hand panel, it is seen that there is a tail to the linewidth distribution that extends to 1000 km s1. The galaxies that contribute to this tail are of sufficient interest and we will discuss them in a later publication. It is shown in the lower left-hand panel that galaxies with a full range of linewidths are seen over a wide redshift range. Within 3000 km s1 there is greater representation of narrow linewidth systems as a consequence of the sample selection. Since our full catalogue is a compendium of many sources, there are artefacts of the collection. In Fig. 8, we plot the logarithm of the integrated flux versus the logarithm of the linewidth. The top panel shows that with our Cosmic Flows observations, large linewidth targets are as well represented as small linewidth targets at the faint limit. This situation arises because our integration times are established to be long enough to acquire adequate profiles case by case. By contrast, in the middle panel, it is seen that there is a strong dependence of the faint limit in the sample of the 1000 brightest galaxies in the HIPASS (Koribalski et al. 2004) where integration times are fixed. A similar trend though at a lower flux level is found in the ALFALFA survey after 40 per cent completion (Martin et al. 2010). Blind surveys with fixed integrations tend to miss lowflux/high-rotational-velocity galaxies. The bottom panel shows that this trend is not severe for the ensemble of our ADHI catalogue, since it is comprised at 85 per cent from targeted observations. For comparison, we overplot the detection limits for the HIPASS (green upper line) and ALFALFA after 40 per cent completion (red lower line). Fig. 9 shows the distribution of the detected masses with distances. The lower green solid line is the ALFALFA 5 sensitivity at 5.6 Jy km s1 and the upper red line is the HIPASS 5 detection limit at 0.372 Jy km s1. The H I mass is computed using the integrated flux and the redshift of the galaxies: MH I = 2.36 105DM2pc FI, where DMpc is the distance and FI is the integrated flux in Jy km s1. For consistency, in a comparison with ALFALFA, velocities are converted to distance assuming H0 = 70 km s1 Mpc1. Fig. 10 shows the distribution of the H I mass in our all-sky catalogue. It is comparable with the ALFALFA 40 per cent H I mass function from Martin et al. (2010) shown with the red line. The cut-off at high H I mass is abrupt above 3 1010 M . The interesting objects in the high-H I-mass tail will be discussed in a separate publication along with the objects in the high-linewidth tail. Figure 8. The distribution of sources detected with an adequate profile and an adequate signal-to-noise ratio: (a) our GBT and Parkes observations; (b) 1000 brightest HIPASS profiles; (c) the ADHI catalogue with overplotted the detection limitation of the HIPASS (green upper line) and ALFALFA 40 per cent (red lower line). Our all-sky catalogue, which is mainly composed of targeted observations, does not show the detection bias trend of blind surveys. Blind surveys tend not to detect the low-flux/high-rotationalvelocity galaxies. 8 S U M M A R Y Over the years, many observers at the worlds largest telescopes have acquired useful information about the neutral hydrogen properties of spiral galaxies. Our primary concern is to measure distances by exploiting the correlation between galaxy H I profile linewidths and luminosities, the TF relation. Five samples have been defined: one to calibrate the TF relation with newly defined linewidth parameters and photometry, another to assure uniformity with the SNIa distance scale and three more that provide the all-sky coverage to different depths, with different densities and distinct selection criteria. Observations with the GBT and the Parkes telescope within the Cosmic Flows programme have built upon the body of the archival material to the degree that adequate H I profile information now exists for almost all the galaxies in our five samples. Figure 10. Histogram of the distribution of H I masses of 11 051 galaxies with good profiles, plotted as the logarithm of the H I mass in solar units. The results from the 40 per cent ALFALFA are shown in red (Martin et al. 2010). Tabular and graphical information on the H I properties of galaxies, whether from new observations or from the archives, is gathered and made available at the EDD website: (select the catalogue All Digital HI). AC K N OW L E D G M E N T S New observations across the entire sky have been made possible by access to three fine radiotelescopes. We made early observations with the refurbished Arecibo telescope and expect to add fresh material coming from the wide-field multibeam survey. At the GBT, our ongoing programme Cosmic Flows has been awarded the status of a Large Program. Observations of the deep southern sky began in 2009 with the Parkes telescope in Australia. The authors acknowledge the valuable support provided by the GBT friends: Franck Ghigo, Ronald J. Maddalena and Toney Minter, GBT scheduling and direction team: Karen ONeil, Jules Harnett and Carl Bignell, and all the operators who helped us to conduct our 1000 h of observations: Dave Curry, Kevin Gum, Greg Monk, Dave Rose, Barry Sharp and Donna Stricklin. The authors also acknowledge the valuable support provided by the CSIRO staff Stacy Mader and Mark Calabretta in retrieving the Parkes Archive material and for the data flux calibration. Equally important to us has been the access to the archival material from the Cornell Digital H I Archive, the Nancay Radio Telescope H I profiles of Galaxies database and the Australia Telescope online archive. Although electronic archives are a great innovation, the low-tech information gathered in the Pre Digital HI catalogue retains great value and we thank Cyrus Hall for his role in assembling that material. We have made extensive use of the NASA/IPAC Extragalactic Database operated by the Jet Propulsion Labratory, California Institute of Technology, and the HyperLeda data base hosted at the Universite Lyon 1. RBT acknowledges support from the US National Science Foundation award AST-0908846. DIM and IDK were supported by the Russian Foundation for Basic Research grants 080200627, RUS-UKR 090290414. R E F E R E N C E S S U P P O R T I N G I N F O R M AT I O N Additional Supporting Information may be found in the online version of this article: Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article. This paper has been typeset from a TEX/LATEX file prepared by the author.

This is a preview of a remote PDF:

Hélène M. Courtois, R. Brent Tully, D. I. Makarov, S. Mitronova, B. Koribalski, I. D. Karachentsev, J. Richard Fisher. Cosmic Flows: Green Bank Telescope and Parkes H i observations, Monthly Notices of the Royal Astronomical Society, 2011, 2005-2016, DOI: 10.1111/j.1365-2966.2011.18515.x