New HI scaling relations to probe the HI content of galaxies via global HI-deficiency maps
MMon. Not. R. Astron. Soc. , 1– ?? (2002) Printed 7 November 2018 (MN LaTEX style file v2.2) New H i scaling relations to probe the H i content ofgalaxies via global H i -deficiency maps H. D´enes , (cid:63) , V. A. Kilborn , B. S. Koribalski Centre for Astrophysics & Supercomputing, Swinburne University of Technology Australia Telescope National Facility, CSIRO Astronomy and Space Science, P.O. Box 76, Epping, NSW 1710, Australia
Released 2002 Xxxxx XX
ABSTRACT
We present new multi-wavelength scaling relations between the neutral hydrogencontent (H i ) and the stellar properties of nearby galaxies selected from the H i ParkesAll-Sky Survey (HIPASS). We use these new scaling relations to investigate the en-vironmental dependency of the H i content of galaxies. We find that galaxies in highdensity environments tend to have on average less H i than galaxies with the samestellar mass in the low density environment. Our new H i scaling relations allow usto identify individual galaxies, as well as group/cluster environments, that have an‘anomalous’ H i content. We map the global distribution of H i -deficient and H i -excessgalaxies on the sky and compare it to the large scale structure of galaxies. We findgalaxy clusters to be H i -deficient, and we identify that the regions surrounding clus-ters tend to be H i -excess. Finally, we demonstrate the potential of using H i scalingrelations to predict future H i surveys based on an optical redshift survey. We applyour scaling relations to 16709 galaxies in the 6dF Galaxy Survey (6dFGS) that lie inthe HIPASS volume and compare our predictions to the measurements. We find thatscaling relations are good method to estimate the outcome of H i surveys. Key words: galaxies: evolution – galaxies: general – radio lines: galaxies – surveys.
One of the main challenges of extragalactic astrophysics isto understand the role of cold gas in the formation and evo-lution of galaxies. Neutral hydrogen gas (H i ) is one of themost important building blocks of galaxies, since it is themain reservoir for future star formation along with molec-ular hydrogen. Without a significant reservoir of cold gas,star formation terminates and the galaxy becomes red andpassive (e.g. Larson et al. 1980).We know that the environment plays an important rolein the gas content of a galaxy and in galaxy evolution. Forexample the fraction of early-type galaxies increases andthe fraction of late-types decreases with increasing galaxydensity. This is called the ‘morphology density relation’(Dressler 1980; Fasano et al. 2000; Goto et al. 2003). An-other important observation is that the fraction of late-typegalaxies in galaxy clusters increases with increasing redshift,which is called the Butcher-Oemler effect (Butcher & Oem-ler 1978). These effects suggest that environmental processes (cid:63) E-mail: [email protected] play an important role in transforming late-type star form-ing galaxies into passive red early type galaxies over time.In terms of H i content, spiral galaxies in high densityenvironments tend to have, on average, less H i than galax-ies of the same type and size in the field (e.g. Davies &Lewis 1973; Giovanelli & Haynes 1985; Solanes et al. 2001).This implies that late type galaxies in high density regionsare getting stripped of, or use up, their gas and are trans-forming to gas poor early type galaxies (e.g. Gunn & Gott1972; Fasano et al. 2000; Bekki et al. 2002; Bekki & Couch2011). Possible gas stripping mechanisms, such as mergers,tidal interactions (Mihos et al. 2005), ram pressure stripping(Gunn & Gott 1972), turbulent or viscous stripping (Nulsen1982), thermal evaporation (Cowie & Songaila 1977), star-vation (Larson et al. 1980) and harassment (Moore et al.1996), are well studied in high density environments, espe-cially in the Virgo cluster (e.g. Kenney et al. 2004; Vollmeret al. 2001; Chung et al. 2009). There is more and more evi-dence that environment starts to influence galaxy evolutionat densities typical of poor groups (Chamaraux & Masnou2004; Kilborn et al. 2005; Sengupta & Balasubramanyam2006; Kilborn et al. 2009; Westmeier et al. 2011), but gas c (cid:13) a r X i v : . [ a s t r o - ph . GA ] J u l H. D´enes, V. A. Kilborn, B. S. Koribalski stripping mechanisms are not nearly as well studied in lowdensity environments than in galaxy clusters.HI-optical scaling relations are a useful tool to inves-tigate how different density environments influence the gascontent of galaxies. We can characterise the H i content ofgalaxies with scaling relations between the H i content andother intrinsic properties of galaxies. Then we can approachenvironmental effects from two points of view. Firstly wecan investigate global, statistical trends between the H i con-tent and the environment density or, secondly we can lookat individual galaxies that differ from the average. We canidentify such galaxies with anomalous H i content by usingscaling relations to calculate their expected H i mass andcomparing it to their measured H i mass. Anomalous galax-ies may have either less H i than expected (H i -deficient) orthey have more H i than expected (H i -excess). To quan-tify the relative gas content of galaxies Haynes & Giovanelli(1983) introduced the “deficiency factor”. The H i deficiencyfactor ( DEF ) is expressed as a logarithmic quantity, posi-tive for H i -deficient galaxies and negative for galaxies withH i -excess. Def HI = log[M HIexp ] − log[M HIobs ] , (1)where M HIexp is the expected H i mass, usually calculatedfrom H i scaling relations, and M HIobs is the calculated H i mass from the measurements.Previous works investigating HI-optical scaling relationsfor late type galaxies found that the optical diameter andluminosity correlate well with the H i mass (e.g. Haynes& Giovanelli 1984; Chamaraux et al. 1986; Solanes et al.1996). Initial scaling relations were determined from rela-tively small, optically selected samples, that have a naturalbias against blue, low surface brightness objects, which tendto have significant amounts of H i . More recent studies favourthe use of H i selected samples combined with optical prop-erties from the Sloan Digital Sky Survey (SDSS; York et al.2000). For example Toribio et al. (2011) investigated a sam-ple of isolated galaxies from the Arecibo Legacy Fast ALFA(ALFALFA) blind 21 cm line survey (Giovanelli et al. 2005)and found that the best indicator for H i mass is the SDSS r -band diameter, followed by the total luminosity and themaximum rotational speed of the galaxy. Another approachto estimate a galaxy’s H i content is to use recent star for-mation indicators such as UV to optical colours. Catinellaet al. (2010) showed that the linear combination of NUV-r colour and stellar surface density is a good predictor of thegas content of massive galaxies with stellar masses greaterthan 10 M (cid:12) . However H i scaling relations are sensitive tothe optical data. Scaling relations derived with SDSS datacan only be reliably applied to galaxies in SDSS due to theunique photometric filters of SDSS. UV and SDSS photome-try is not available for a large fraction of galaxies. Upcominglarge scale H i surveys in the southern hemisphere make itnecessary to also establish scaling relations between otheravailable optical data and the H i content of galaxies.In this work we derive H i scaling relations for galaxiesusing the H i Parkes All Sky Survey (HIPASS) and a vari-ety of optical and near-infrared luminosities and diameters.We use a multi-wavelength approach to determine scalingrelations between the H i content of galaxies and their di-ameter and luminosity, in 5 and 6 optical/IR wavebandsrespectively. The different bands are not uniformly affected by extinction and they probe different stellar populations ofa galaxy. Moreover the large sky coverage of the cataloguesfrom which we derive our scaling relations, makes them suit-able to investigate how different environments influence theHI content of galaxies. In section 2 of this paper, we describethe H i and optical datasets. In section 3, we present ourscaling relation and the influence of environment on the H i content. In section 4, we present applications of our scalingrelations. We show that it is possible to identify individualgalaxies or groups/clusters that have an HI content that de-viates substantially from the expected values. InvestigatingH i scaling relations in the southern hemisphere is especiallyimportant now, in preparation for the upcoming large H i and optical surveys, such as the ASKAP H i All Sky Survey,known as WALLABY (Koribalski & Staveley-Smith (2009);Koribalski 2012) and SkyMapper (Keller et al. 2007). Wepresent a method to estimate the outcome of a large blindH i survey by predicting the H i mass of 16709 galaxies in the6dF Galaxy Survey (Jones et al. 2009). We compare these tothe current HIPASS catalogues to investigate how effectiveour predictions are. In section 5, we summarise our results.Throughout this paper we use H = 70 km s − Mpc − . The H i Parkes All-Sky Survey (HIPASS; Barnes et al. 2001)is a blind H i survey conducted with the 64 m Parkes radiotelescope covering two-thirds of the entire sky from decli-nation δ = − ◦ to +26 ◦ in the radial velocity range of − < cz < − . HIPASS has a gridded beamsize of 15 . (cid:48)
5, a velocity resolution of 18 km s − , and an r.m.s.noise of ∼
13 mJy. Details of the survey and of the datareduction are described in Barnes et al. (2001).In our work we use the following catalogues: theHIPASS Bright Galaxy Catalog (HIPASS BGC; Koribalskiet al. 2004), the southern HIPASS catalogue (HICAT; Meyeret al. 2004) with its optical counterpart (HOPCAT; Doyleet al. 2005), and the northern HIPASS catalogue (NHI-CAT; Wong et al. 2006) with its optical/infrared counterpart(NOIRCAT; Wong et al. 2009).The HIPASS BGC lists the H i properties of the 1000H i -brightest extragalactic sources in the southern sky ( δ < ◦ , H i peak flux >
116 mJy). Koribalski et al. (2004) found853 HIPASS sources associated with single optical galaxies(68 of these are marked as confused), 44 with galaxy pairsand 11 with compact groups. All but nine of the 853 singlegalaxies (with v LG <
300 km s − ) are also listed in HICAT/ HOPCAT. The positional accuracy of HIPASS sources isgiven by the gridded beam divided by the signal-to-noise ra-tio. Most HIPASS BGC sources have H i spectra with signalto noise of at least nine, ie a position uncertainty of 1 . (cid:48) δ = +2 ◦ and contains 4315 extragalactic H i sources ( v LG >
300 km s − , H i peak flux (cid:38)
40 mJy). To derive our mainscaling relations we use the optical properties of the mostreliable single galaxy identifications listed in HOPCAT. Thissample consists of 1798 galaxies, about half of all the opticalidentifications by Doyle et al. (2005) and nearly twice the c (cid:13) , 1– ?? ew H i scaling relations to probe the H i content of galaxies via global H i -deficiency maps number of HIPASS BGC single galaxy identifications. Forour study it is important to avoid using H i sources withmultiple or uncertain optical identifications.In addition to our southern galaxy samples we also se-lect a sample of northern HIPASS sources. NHICAT coversthe sky from δ = +2 ◦ to 25.5 ◦ and contains 1002 extra-galactic sources. Of these, 414 galaxies have reliable, singleoptical counterparts in NOIRCAT. This sample includes theVirgo cluster, which enables us to investigate the effect of alarge galaxy cluster like Virgo on the H i scaling relations.The NOIRCAT sample has corresponding SDSS data, whichmakes it possible to compare the difference of using SDSS r -band photometry to non SDSS R -band photometry. We obtain homogeneous sets of optical and infrared prop-erties (magnitudes and diameters) for the three HIPASSgalaxy samples. These consist of magnitudes and diametersin three optical bands ( B , R , I ) and three infrared bands ( J , H , K ). Each individual property is drawn homogeneouslyfrom one source. In the following we briefly describe thedata used for our analysis.For the southern galaxy samples we use the optical B , R and I -band magnitudes catalogued in HOPCAT. Doyleet al. (2005) obtained these measurements from SuperCos-mos plates. The catalogued B -band magnitudes are consis-tent with B -band magnitudes in other catalogues, but wefind the R and I -band magnitudes have a systematic offset.The reason for this is that Doyle et al. (2005) measured the R and I -band magnitudes inside the same elliptical aper-ture that was used to measure the B -band magnitudes. Weinvestigate this systematic offset of their magnitudes andconclude that the R and I -band magnitudes are consistentwith themselves, but need a linear scaling to be comparablewith other catalogues. We scale the HOPCAT magnitudeswith the following equations:M R scaled = − .
99 + 1 .
18 M R , (2)M I scaled = − .
94 + 1 .
06 M I . (3)The scaled magnitudes are in good agreement with Super-Cosmos magnitudes in the 6dF Galaxy Catalogue (Joneset al. 2009). Optical B and R -band luminosities and diam-eters are good tracers of the young stellar populations ingalaxies, but these bands are also very sensitive to extinc-tion which can contribute significant measurement errors forindividual galaxies.For a large fraction of the southern sample (1250 galax-ies) we also obtain J , H and K -band magnitudes from the2MASS Extended Source Catalog (Skrutskie et al. 2006).2MASS is currently the largest near-infrared catalog avail-able for galaxies that covers the whole sky. The advantages ofusing near-infrared wavelengths are that they are less sensi-tive to dust obscuration and they can be used as stellar massindicators. Furthermore, the whole sky coverage of 2MASSalso makes it possible to compare our results with galaxiesin the northern hemisphere.For the northern galaxy sample we obtain the 2MASS J , H and K magnitudes directly from NOIRCAT (Wong et al.2009). All other optical and infrared properties are obtained Table 1.
Overview of the optical and infrared data and the num-ber of galaxies in the samples.B R r I J H K δ < +2 ◦ magnitude 1796 1796 - 1795 1249 1249 1250(Ref) (1) (1) (1) (2) (2) (2)diameter 1179 262 - 1021 632 - 1211(Ref) (3) (4,5) (6) (6) (2) δ > +2 ◦ magnitude 343 - 175 158 414 414 414(Ref) (3) (7) (8) (2) (2) (2)diameter 343 471 166 158 - - 354(Ref) (3) (6) (7) (8) (2)(1) Doyle et al. (2005), HOPCAT – SuperCosmos magnitudes(2) Skrutskie et al. (2006), external 3 σ diameter(3) Paturel et al. (2000), 25 mag arcsec − isophote(4) Nilson (1973), external visual diameter(5) Nilson (1974), external visual diameter(6) Paturel et al. (2005), 25 mag arcsec − isophote(7) Adelman-McCarthy et al. (2007), external 3 σ diameter(8) Springob et al. (2007), 23.5 mag arcsec − isophote from HyperLEDA using only the catalogues specified inTable 1. This gives us homogeneous data for each individualband.We correct all optical and infrared magnitudes forGalactic extinction based on Schlegel et al. (1998) and forinternal absorption following Driver et al. (2008). Galaxy di-ameters are also corrected for extinction following (Graham& Worley 2008) assuming they are entirely disc dominated. i scaling relations To derive scaling relations we use sub samples of our maingalaxy sample. To ensure that we are deriving scaling rela-tions from an environmentally unbiased sample, we only usegalaxies in low density environments. We calculate the envi-ronmental density around each of the galaxies in our maingalaxy sample and only use galaxies with Σ < − (see details in section 3.3). We also exclude galaxies with H i fluxes lower than the HIPASS 95% reliability limit ( S int < − , Zwaan et al. 2004) to ensure that we are usingonly the best quality data. These cuts exclude about 30 %of our main galaxy sample, but improve the reliability of ourscaling relations.We investigate the effect of using volume limited sam-ples, which result in significantly smaller sample sizes withonly a few hundred galaxies. The linear regression fits tothese samples are in good agreement with our non volumelimited sample considering the significantly larger errors onthe fitting because of the small sample size. Since the re-gression fitting for the volume limited sample is consistentwith the non volume limited sample we decide against us-ing a volume limited sample. We also investigate the effect http://leda.univ-lyon1.fr/c (cid:13) , 1–, 1–
Overview of the optical and infrared data and the num-ber of galaxies in the samples.B R r I J H K δ < +2 ◦ magnitude 1796 1796 - 1795 1249 1249 1250(Ref) (1) (1) (1) (2) (2) (2)diameter 1179 262 - 1021 632 - 1211(Ref) (3) (4,5) (6) (6) (2) δ > +2 ◦ magnitude 343 - 175 158 414 414 414(Ref) (3) (7) (8) (2) (2) (2)diameter 343 471 166 158 - - 354(Ref) (3) (6) (7) (8) (2)(1) Doyle et al. (2005), HOPCAT – SuperCosmos magnitudes(2) Skrutskie et al. (2006), external 3 σ diameter(3) Paturel et al. (2000), 25 mag arcsec − isophote(4) Nilson (1973), external visual diameter(5) Nilson (1974), external visual diameter(6) Paturel et al. (2005), 25 mag arcsec − isophote(7) Adelman-McCarthy et al. (2007), external 3 σ diameter(8) Springob et al. (2007), 23.5 mag arcsec − isophote from HyperLEDA using only the catalogues specified inTable 1. This gives us homogeneous data for each individualband.We correct all optical and infrared magnitudes forGalactic extinction based on Schlegel et al. (1998) and forinternal absorption following Driver et al. (2008). Galaxy di-ameters are also corrected for extinction following (Graham& Worley 2008) assuming they are entirely disc dominated. i scaling relations To derive scaling relations we use sub samples of our maingalaxy sample. To ensure that we are deriving scaling rela-tions from an environmentally unbiased sample, we only usegalaxies in low density environments. We calculate the envi-ronmental density around each of the galaxies in our maingalaxy sample and only use galaxies with Σ < − (see details in section 3.3). We also exclude galaxies with H i fluxes lower than the HIPASS 95% reliability limit ( S int < − , Zwaan et al. 2004) to ensure that we are usingonly the best quality data. These cuts exclude about 30 %of our main galaxy sample, but improve the reliability of ourscaling relations.We investigate the effect of using volume limited sam-ples, which result in significantly smaller sample sizes withonly a few hundred galaxies. The linear regression fits tothese samples are in good agreement with our non volumelimited sample considering the significantly larger errors onthe fitting because of the small sample size. Since the re-gression fitting for the volume limited sample is consistentwith the non volume limited sample we decide against us-ing a volume limited sample. We also investigate the effect http://leda.univ-lyon1.fr/c (cid:13) , 1–, 1– ?? H. D´enes, V. A. Kilborn, B. S. Koribalski of volume weighting on our sample. We find that using in-verse volume weighting on our data - with the same methodthat is used when deriving HI mass functions - results insignificantly underestimating the H i mass when calculatedfrom the optical data. The same can be seen in Toribio et al.(2011). Since our aim is to establish scaling relations thatcan be used to predict the H i content of late-type galax-ies based on their optical properties we decide against usingvolume weighting to derive our scaling relations. We determine scaling relations between the logarithm of theobserved H i mass of galaxies and their magnitudes in 6 dif-ferent wavebands. We tested different regression fitting algo-rithms to determine the best method to derive our scalingrelation. We find that the ordinary least-squares bisector(OLS bisector) regression line best recovers the one-to-onerelation between the predicted H i mass and the observed H i mass. An ordinary least square (OLS) fit to the data resultsin an under prediction of the H i mass for bright galaxies andover prediction of the H i mass for faint galaxies. Feigelson &Babu (1992) also recommends to use a symmetric regressionline fit, such as the OLS bisector for investigating physicalprocesses behind regressions. We use OLS bisector regres-sion line fitting through the whole of this work. The scalingrelations for the magnitudes are in the formlogM HI = α + β · M x (4)where M HI ( M (cid:12) ) is the H i mass of a galaxy, M x is theabsolute magnitude in the various bands and α and β arethe parameters of the relation (Table 2). The observed H i mass ( M (cid:12) ) is calculated with the following equation M HI = 2 . × D F HI , (5)where, F HI (Jy) is the integrated flux of the 21 cm emissionline and D is the distance of the source in Mpc using D = v LG /H . v LG is the Local Group velocity calculated from theradial velocity v LG = v +300 sin l cos b . We use these Hubbleflow distances trough the whole paper. For nearby galaxiesHubble flow distances might have large errors, but accordingto Zwaan et al. (2003) southern HIPASS galaxies shouldnot be largely effected by local galaxy over-densities. Wecompare our calculated distances to distances derived fromThe Extragalactic Distance Database (EDD). We find areasonably good agreement between our calculated distancesand the distances in the database.Figure 1 shows our relations between the H i mass ofthe galaxies and their magnitudes in the 6 multi-wavelengthbands. The grey line shows the OLS bisector regression fitto the data. The grey dashed lines show the 1 σ standarddeviation from the fitted line and the solid green lines markdeficiency factors of ± . http://edd.ifa.hawaii.edu/ for the line fitting. The fitted parameters and the bootstraperrors for all 6 bands are in Table 2.We find a tighter correlation with the optical B , R and I band magnitudes ( σ = 0 . , . , .
29) than with the near-infrared J , H and K bands ( σ = 0 . , . , . i con-tent. The 2MASS K -band diameter of a disk galaxy is onaverage 1.5-2 times smaller than it’s B -band diameter (Jar-rett et al. 2003).To investigate if there is a trend between the scalingrelations in the different wavebands, we compare the binneddata points of the H i mass - magnitudes scaling relations.We find that galaxies with a similar H i mass tend to havebrighter magnitudes at the longer wavelengths (Figure 3).This trend is most likely caused by the increasing dust ab-sorption towards shorter wavelengths or the different sam-pled stellar populations. After binning the data in log H i mass as well, we conclude that there is a small increase inthe difference between the different magnitudes from the lowto the high H i masses, which explains the slightly differentslopes of the H i scaling relations, but this difference is notsignificant.For comparison we also derive scaling relations in the B -band only using galaxies that are also in the BGC. Thesegalaxies are the brightest galaxies in our sample with thebest position accuracy and have the most certain opticalidentifications. We find that the scaling relation derived fromthis sub-sample is in good agreement with the HOPCATsample B -band scaling relation. We also determine scaling relations between the observedH i mass of galaxies and the logarithm of their diameterin 5 different wavebands using the same method as for themagnitudes. The scaling relations for the diameters arelogM HI = α + β · log d x (6)where M HI ( M (cid:12) ) is the H i mass of a galaxy and d x (kpc)is the diameter in the various bands and α and β are theparameters of the relation (Table 2).Figure 2 shows the correlation between the H i mass ofthe galaxies and their B , R , I , J , K band diameters. Mark-ings are the same as in Figure 1. We perform a bootstrapanalysis for the line fitting, the fitted parameters and thebootstrap errors for all 5 bands are in Table 2.Similar to the magnitudes, we see tighter correlationsbetween the H i mass and the optical B , R and I diameters( σ = 0 . , . , .
25) than for the near-infrared J and K diameters ( σ = 0 . , . R -band diam-eter scaling relation has the smallest scatter of our derivedrelations, however this is not necessarily due to the strength c (cid:13) , 1– ?? ew H i scaling relations to probe the H i content of galaxies via global H i -deficiency maps −24−22−20−18−16−14−12 M B [mag]7.07.58.08.59.09.510.010.511.0 l o g ( M H I / M ⊙ ) −24−22−20−18−16−14−12 M R [mag]7.07.58.08.59.09.510.010.511.0 −24−22−20−18−16−14−12 M I [mag]7.07.58.08.59.09.510.010.511.0 −28−26−24−22−20−18−16−14−12 M J [mag]7.07.58.08.59.09.510.010.511.0 l o g ( M H I / M ⊙ ) −28−26−24−22−20−18−16−14−12 M H [mag]7.07.58.08.59.09.510.010.511.0 −28−26−24−22−20−18−16−14−12 M K [mag]7.07.58.08.59.09.510.010.511.0 Figure 1.
The logarithm of H i mass plotted against 6 different optical and infrared magnitudes. The solid grey line is the fitted bisectorline to the data and the dashed grey lines are the 1 σ standard deviation. The red points are magnitude bins with 200 galaxies each(except for the last bin). The green lines are marking DEF = ± .
6, galaxies below this threshold are considered H i -deficient and galaxiesabove this threshold are considered to have H i -excess. The numbers in the top right corner of the sub-plots show the sample size in eachband and the grey points with the error bars show the average uncertainty in the measurements. In the top left panel we marked theHOPCAT sample with grey points and the BGC sample with black points. of the scaling relation. It could also be caused by the sig-nificantly smaller sample size of this band compared to theother bands. Considering the sample sizes, the scatter of therelations and the Pearson’s correlation coefficient we find the B -band magnitude H i scaling relation the best to estimatethe H i mass of a galaxy and use it for estimating H i masseshereafter.We did not make any morphological selections for ourscaling relations, but it is important to note that our scal-ing relations are only valid for late type galaxies, since ourgalaxy sample contains a relatively insignificant amount ofearly type galaxies (ellipticals and S0s). Our sample alsocontains a small number of dwarf galaxies, which do not in-fluence the relations significantly and follow the H i scalingrelations derived for all the sample galaxies. An important property of the scaling relations is the scatter.We investigate several effects that can broaden the scatterin the relations. −26−24−22−20−18−16−14 M x [mag] l o g ( M H I / M ⊙ ) BRIJHK
Figure 3.
Comparing the binned data points of the H i mass -magnitude scaling relations. Different coloured points show thedata in the different wavebands.c (cid:13) , 1–, 1–
Comparing the binned data points of the H i mass -magnitude scaling relations. Different coloured points show thedata in the different wavebands.c (cid:13) , 1–, 1– ?? H. D´enes, V. A. Kilborn, B. S. Koribalski l o g ( M H I / M ⊙ ) l o g ( M H I / M ⊙ ) Figure 2.
The logarithm of H i mass plotted against the logarithm of the diameter in 5 different optical and infrared bands. The solidgrey line is the fitted bisector line to the data and the dashed grey lines are the 1 σ standard deviation. The red points are diameter binswith 100 galaxies each (except for the last bin). The green lines are DEF = ± .
6, galaxies below this threshold are considered H i -deficientand galaxies above this threshold are considered to have H i -excess. The numbers in the top right corner of the sub-plots show the samplesize in each band and the grey points with the error bars show the average uncertainty in the measurements. Table 2.
Parameters for the H i scaling relations ( δ < +2 ◦ ), α and β are the relation parameters, N is the number of galaxies used todetermine the relations and σ is the standard deviation from the least square fit.Type β α N σ Pearson’s cofficientB magnitude -0.34 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± (cid:13) , 1– ?? ew H i scaling relations to probe the H i content of galaxies via global H i -deficiency maps The internal extinction of galaxies may comprise onecomponent of the scatter. We correct the magnitudes for in-ternal extinction following Driver et al. (2008) and correctthe diameters for extinction effects following (Graham &Worley 2008) assuming disc dominated galaxies. The scat-ter in the scaling relations for the magnitudes improves afterthe correction but we do not see a significant improvementfor the diameters. This is most probably due to the largeuncertainties in the corrections for internal dust and incli-nation. We do not find any significant trend between theH i deficiency parameter and the inclination of our samplegalaxies. We also find that excluding galaxies with extremeinclination ( i < ◦ and i > ◦ ) does not decrease thescatter for our scaling relations.Part of the scatter arises from environmental differ-ences. Galaxies in high density environments have a slightlydifferent H i content compared to galaxies in low density en-vironments. We discuss this in more detail in section 3.3.After excluding galaxies in high density environments thescatter of our scaling relations decreased.Additional factors adding to the scatter can be mor-phology (e.g. Haynes & Giovanelli (1984)), uncertainty ofthe H i mass, different metallicities, different star formationhistories and of course measurement errors. This means thatH i scaling relations always have an uncertainty that needsto be considered for their applications, such as predictingthe H i mass of galaxies. We also derive scaling relations using galaxies in NOIRCAT(Wong et al. 2009) for the H i sample to investigate the ef-fects of a large galaxy cluster, like the Virgo cluster, on H i scaling relations and the difference between using SDSS ornon SDSS data. Our northern sample consists of the 414galaxies with high confidence optical counterparts in NOIR-CAT. The optical and near-infrared data for this sample isderived from HyperLEDA selecting homogeneous data fromsingle catalogues (Table 1).We derive the correlations the same way as presented insection 3.1 and present the parameters in Table 3. The scal-ing relations derived from NOIRCAT are similar to the onesfrom the HOPCAT sample with some variations. We findthat considering the sample size, the scatter of the relationsand the Pearson’s correlation coefficient the R -band diame-ter relation is the best to estimate the H i mass of a galaxy.We compare our derived scaling relations from the SDSS r -band diameter and the R -band diameter from Nilson (1973).We find that the average difference of the predicted log M HI from the r and R -band magnitudes is relatively small (0.02dex), but the standard deviation is considerable (0.41 dex)and the difference for individual galaxies can be up to 2.66dex. This shows that scaling relations derived from differentoptical data sets can only be used with caution to predictthe H i mass of a galaxy.Figure 4 shows our scaling relations between the H i mass of galaxies in NOIRCAT and their 2MASS J -bandmagnitude. We mark galaxies in cyan that are in high den-sity environments (Σ > − ), these galaxies are notused when calculating the scaling relations. We show the J -band magnitude, because we would like to emphasize thatit is possible to identify H i -deficient and H i -excess galaxies −28−26−24−22−20−18−16 M J [mag] l o g ( M H I / M ⊙ ) Virgo, low densityVirgo, high density
Figure 4. i mass(NOIRCAT sample). Grey and cyan points mark galaxies in lowand high density environments respectively. The solid grey lineis the fitted bisector line and the dashed grey lines are the 1 σ standard deviation. The red points show the data in magnitudebins (40 galaxies each bin, except the last bin). The green linesare DEF = ± .
6. The blue and cyan stars mark galaxies in thearea of the Virgo cluster. not just with optical but also with near-infrared scaling re-lations. The northern extension of HIPASS covers the skyup to a declination of +25 ◦ including the Virgo galaxy clus-ter. The Virgo cluster is known to have several H i -deficientspiral galaxies (e.g Giovanelli & Haynes (1983); Haynes &Giovanelli (1986); Chung et al. (2009)). In figure 4 blue andcyan stars mark galaxies in the area of the Virgo cluster(+2 ◦ < δ < +20 ◦ , 12 h < α <
13 h 20 m, v < − ).There are 63 galaxies in this region with available J -bandmagnitude of which 13 are among the most H i -deficientgalaxies in this sample, with at least 4 times less H i thanthe average galaxies ( DEF > . H i scaling relations can be used to investigate the environ-mental effects on galaxies. We know that the H i content ofa galaxy is strongly influenced by the environment. For ex-ample, galaxies in high density environments tend to be onaverage H i -deficient compared to galaxies in low density en-vironments (e.g. Giovanelli & Haynes (1985); Solanes et al.(2001); Cortese et al. (2011)). This means that galaxies indifferent density environments should have slightly differentscaling relations. We now investigate how our H i scalingrelations depend on environment.To quantify the environment our galaxies lie in, we cal-culate the galaxy surface density around each of our samplegalaxies, using a reference sample from HyperLEDA. Thereference sample includes all galaxies in HyperLEDA belowa declination of 10 ◦ that have an apparent B magnitudebrighter than 14 mag and have measured radial velocitiessmaller than 6000 km s − . HyperLEDA is complete to a lim-iting apparent magnitude of m B =14 mag within | b | < ◦ and cz (cid:54) − (Giuricin et al. 2000). We show thevelocity distribution of our reference sample and the HOP-CAT sample in Figure 11. To avoid edge effects we only cal- c (cid:13) , 1– ?? H. D´enes, V. A. Kilborn, B. S. Koribalski
Table 3.
Parameters for H i scaling relations ( δ > +2 ◦ )T, a and b are the relation parameters, N is the number of galaxies used todetermine the relations and σ is the standard deviation from the least square fit.Type β α N σ Pearson’s cofficientB magnitude -0.23 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± culate the environment for galaxies in the HOPCAT samplewith | b | > ◦ . This limit is sufficient because the averagedistance to the 7th nearest neighbour in our sample is ∼ ◦ .We calculate the galaxy surface density with the pro-jected 7th nearest neighbour method. We use a ± − cylinder centred around each galaxy and the follow-ing equation to calculate the surface density:Σ ( Mpc − ) = 7 πr (7)where r (Mpc) is the distance to the 7th nearest neighbour.We divided the galaxies in our sample into 2 density groups:galaxies in a dense environment (Σ > − ) and galax-ies in a less dense environment (Σ < − ), correspond-ing to an average projected galaxy density of 1 galaxy/Mpc.Since environmental effects depend strongly on stellar mass(Kauffmann et al. 2004), we show the H i content in the twodifferent environments as a function of stellar mass (Fig-ure 5). We find that galaxies in the higher density environ-ment have a 0.17 dex lower average H i mass than galaxieswith the same stellar mass in the less dense environment.This is in good agreement with previous studies such asCortese et al. (2011). In figure 6 we show the distributionof B band magnitude, H i mass and stellar mass in the twodifferent environments. We calculate stellar masses follow-ing Bell & de Jong (2001) using J -band stellar mass to lightratio and B-R colors. We find that the peak of the distri-butions in the high density sample is shifted towards lowervalues compared to the low density sample. This suggeststhat late type galaxies in a low density environment are onaverage brighter and have a higher H i mass than galaxiesin high density environments. i deficiency & excess H i scaling relations are a useful tool to identify galaxies with‘anomalous’ H i content - i.e. H i -deficient galaxies or galax-ies with H i -excess. It can help us to understand where thesegalaxies are located and what is causing them to have H i -deficiency or H i -excess. These galaxies may have recentlyundergone some kind of external evolutionary processes.Their optical properties are still unchanged, but their H i l o g ( M H I / M ⊙ ) Σ < 1, log (M ⋆ /M ⊙ ) N Figure 5.
Environment dependency of H i mass as a functionof stellar mass. The blue line shows the OLS bisector regressionline for galaxies in low density environments (Σ < − )and red line shows the regression line for galaxies in high densityenvironments (Σ > − ). We show the number of galaxiesin the different environments in the legend and the distributionof the sample in the histograms on the side. content is quiet different. For example we know that H i -deficient spiral galaxies observed in the Virgo cluster haveundergone recent gas stripping. H i -excess galaxies might beaccreting H i from their surroundings.Generally a galaxy is considered to have a normal H i content if it has an H i deficiency parameter between -0.3 and0.3. This corresponds to an H i content 2 times more or lessof an average spiral galaxy. Since the scatter of our H i scal-ing relations is ∼ i -deficient and excess galaxies. We consider a galaxyto be H i -excess if DEF < − . i -deficient if DEF > .
6, which corresponds to 4 times more or less H i than the average. An important test of H i deficiency and H i -excess is if we derive similar deficiency parameters in morethan one waveband. This way we can avoid both individualmeasurement errors for a galaxy in a particular wavebandand some of the errors caused by strong extinction in theshorter wavebands. It is also useful to compare the defi-ciency parameter calculated from the magnitude with the c (cid:13) , 1– ?? ew H i scaling relations to probe the H i content of galaxies via global H i -deficiency maps −24−22−20−18−16−14 M B [mag] N log(M HI /M ⊙) log(M ⋆ /M ⊙) Figure 6.
Environment dependency of the B -band magnitude, H i mass and stellar mass of our sample galaxies. The distribution forgalaxies in the low density environments (Σ < − ) is in blue and for galaxies in higher densities (Σ > − ) in red. Thevertical lines show the means of the distributions. parameter calculated from the diameter of the galaxy. Wefind that the distribution of the H i deficiency parameterin our samples is approximately Gaussian, centred around 0with a standard deviation of 0.27. We find 1% of the galaxiesin our samples are H i -excess and 4% are H i -deficient.We investigate where these anomalous galaxies are lo-cated. H i -deficient galaxies are usually found in galaxy clus-ters (e.g Giovanelli & Haynes 1985; Solanes et al. 2001) anda few were found in lose galaxy groups (e.g. Chamaraux& Masnou 2004; Kilborn et al. 2005; Sengupta & Balasub-ramanyam 2006) and compact groups (Verdes-Montenegroet al. 2001). We find that about half of the galaxies thatwe identify as H i -deficient are classified as members ofgalaxy groups or clusters in the NASA/IPAC Extragalac-tic Database (NED). Based on our density calculations, 42% of the H i -deficient galaxies are in high density environ-ments (Σ > − ) and 58 % are in densities typical ofgalaxy groups or the field.To investigate the distribution of H i -deficient and ex-cess galaxies on the sky we made a 2D histogram of the H i deficiency parameter for the HOPCAT and NOIRCAT sam-ples. We divide our sample into 20 spatial bins and weighteach bin with the H i deficiency parameter of the galaxies init. To compare the H i content with local environment den-sity we overlay galaxy density contours from a magnitudelimited HyperLEDA galaxy sample similar to the sample insection 3.3. The panels in Figures 7 and 8 show differentvelocity cuts of our samples. The colour coding shows theaverage H i deficiency, calculated from the B -band magni-tude. Orange and red colours show H i -deficient regions anddark blue colours show H i -excess regions. These can be in-terpreted as regions in the sky, that have an over density ofH i -deficient or H i -excess galaxies compared to the average.To show this in more detail in Figure 10 we plot the dis-tribution of the H i deficiency parameter of the galaxies inthe different H i -deficient and excess regions from the firstpanel of Figure 8. The histograms are significantly skewed.They show that there are more gas poor galaxies in the H i -deficient regions and more gas rich galaxies in the H i -excessregions. Figures 7 and 8 show that the H i -deficient regionstend to correlate with dens regions and H i -excess regionsseem to be on the edges of dense regions. The lower veloc- http://ned.ipac.caltech.edu/ ity segments have more H i -deficient regions and the highervelocity segments have more H i -excess regions - this is aselection effect that is caused by the H i sensitivity limit ofHIPASS. Naturally, H i -deficient galaxies have less hydro-gen than average galaxies and are harder to detect at largerdistances (Figure 9).We can clearly see a few dominant H i -deficient (red)and H i -excess (blue) features in Figures 7 and 8. We identi-fied the main structures. Several of these groups and clusterswere previously identified as being on average H i -deficientcompared to other galaxy clusters or containing H i -deficientgalaxies (Table 4), but this is the first time that regions withan abundance of gas rich galaxies were identified, such asthe background of the Sculptor cluster and the outskirts ofthe Virgo cluster. The most prominent H i -deficient region isthe Virgo cluster with several H i -deficient galaxies (e.g. Gio-vanelli & Haynes (1983); Chung et al. (2009)). In Figure 8we can still see sub structures of the Virgo cluster with gasrich galaxies in the outskirts. Other big galaxy clusters andgroups including the Eridanus group, the Fornax, Sculptor,Centaurus, Antlia and Hydra clusters. The NGC 2559 groupin the Puppis wall was the first loose group of galaxies thatwas found to be on average H i -deficient by Chamaraux &Masnou (2004).Table 4 shows that the H i -deficient and H i -excessgalaxies are found in a range of different density environ-ments from large clusters to small groups and isolated galax-ies. High resolution follow up observations of these galaxiescan give us a valuable insight into recent galaxy evolutionin different environments. These observations can help usunderstand which gas stripping mechanism are effective ingalaxy groups and how the extremely gas rich spiral galaxiesacquire their H i . i surveys In this section we explore how H i scaling relations may beused to estimate the H i content and detectability of (late-type) galaxies from an optical redshift survey. In principle,the derived relations should help in predicting the outcomeof future blind H i surveys, like WALLABY (Koribalski &Staveley-Smith 2009), but in practise there are numerousdifficulties. While H i surveys predominantly detect nearby c (cid:13) , 1– ?? H. D´enes, V. A. Kilborn, B. S. Koribalski
Table 4.
Previous HI deficiency and X-ray measurements for identified galaxy groups and clusters from the literature. The numbersafter the group name show the population count of the group from NED.Group name (S+Ir):(S0+E)% H i and group properties X-ray measurementsVirgo cluster [1500] 63:37 H i -deficient , X-ray bright,
SUZAKU , ASCA Fornax cluster (Abell 373) [58] - H i -deficient , X-ray bright
ROSAT Dorado group (NGC 1566) [46] 25:75 part of the Fornax wall , , EINST EIN , ROSAT H i -deficient group members Pupis wall [-] - a chain of loose groups -first H i -deficient galaxies in groups Pegasus cluster [6] 82:18 H i -deficient , weak X-ray emission ROSAT Centaurus cluster (Abell 3526) [100] - normal H i content ROSAT , Antlia cluster (Abell 636) [30] - -
ROSAT , Hydra cluster (Abell 1060) [107] - H i -deficient ROSAT , , SUZAKU HI rich dwarf galaxies XMM − Newton Eridanus group [39] 54:46 H i -deficient galaxies X-ray measurment IC 1459 group (Sculptor Cluster) [16] 71:29 normal group H i content , , ROSAT H i -deficient group membersNGC 5846 group [5] early-type H i -deficient X-ray bright
ROSAT ,dominated CHANDRA NGC 0628 group [7] late-type normal group H i content no observations dominated H i -deficient group member NGC 7716 group [5] late-type normal group H i content X-ray faint group dominated NGC 3256 [10] - several peculiar and interacting galaxies - Giovanelli & Haynes (1985) Levy et al. (2007) Solanes et al. (2001) Mahdavi et al. (2000) Shang & Scharf (2009) Reiprich & B¨ohringer (2002) Shibata et al. (2001) Ikebe et al. (2002) Waugh et al. (2002) Duc et al. (1999) Schr¨oder et al. (2001) Piffaretti et al. (2011) Eckert et al. (2011) Brough et al. (2006) Kilborn et al. (2005) Omar & Dwarakanath (2005) Kilborn et al. (2009) Sengupta & Balasubramanyam (2006) Brough et al. (2006) Osmond & Ponman (2004) Jones & Forman (1999) Machacek et al. (2011) Westmeier et al. (2011) English et al. (2010) Westmeier et al. (2013) Haynes & Giovanelli (1991) Chamaraux & Masnou (2004) star-forming galaxies, optical redshift surveys generally tar-get magnitude-limited samples of bright, compact galaxies.We use the 6dF Galaxy Survey (6dFGS, Jones et al.(2009)), a near-infrared selected redshift and peculiar ve-locity survey targeting galaxies in the southern sky. 6dFGSprovides reliable SuperCOSMOS B and R magnitudes - alsoused to characterise the optical properties of HIPASS galax-ies (Doyle et al. 2005) - for over 10 000 galaxies in theHIPASS volume (300 < v < − , | b | < ◦ ). Thevelocity distributions of 6dFGS and HIPASS galaxies, shownin Figure 11, differ significantly with median redshifts of0.053 and 0.009, respectively, suggesting that only very fewof the nearby 6dFGS galaxies can be detected in HIPASS.In order to make our H i mass predictions we use the fol-lowing properties from the 6dFGS catalogue: local group ve-locities, SuperCOSMOS B -band magnitudes. We only select objects that are classified as galaxies and have good qual-ity redshifts. We correct all the magnitudes obtained fromthe 6dF catalogue for galactic extinction and inclination ef-fects following the same methods as described in section 2.2.Subsequently, there are 16709 galaxies in our 6dF sample,for which we calculate their predicted H i mass using our B magnitude scaling relation.To compare our predictions with catalogued galaxies inHIPASS, we apply a ‘detection limit’ and a peak flux cut tothe calculated H i masses. We use the 95% reliability level ofthe HIPASS catalogue (HICAT), which is at an integratedflux of 5 Jy km s − (Zwaan et al. 2004). This gives a detec-tion limit of 1.179 × D M (cid:12) where D (Mpc) is the distanceto the galaxy. A reliable H i detection also requires the peakflux of the galaxy to be 3 σ above the noise level of HIPASS.We calculate the expected peak fluxes for the 6dF galaxies c (cid:13) , 1– ?? ew H i scaling relations to probe the H i content of galaxies via global H i -deficiency maps Figure 7.
Sky distribution of the H i deficiency parameter in 20 two dimensional bins overplayed with HyperLEDA density contours.The colours represent average H i deficiencies of different areas. Red and orange regions have on average more H i -deficient galaxies anddark blue regions have on average more H i rich galaxies than the green and light blue regions. Density contours are 10, 30, 50, 70, 90,110 galaxies. Black dots represent the individual galaxies of our HOPCAT and NOIRCAT sample. by using the B -band Tully-Fisher relation to estimate theirvelocity width. Using this and their predicted integrated fluxwe calculate the peak flux for our sample as well. After ap-plying the integrated flux detection limit and the 3 σ peakflux cut 749 galaxies remain in our sample.Further analysis showed that a significant fraction of this sample are elliptical galaxies (30 %), based on deVaucouleurs morphology classifications from HyperLEDA.Whilst a number of early-type galaxies contain significantamounts of H i (e.g. Sadler et al. 2002; Oosterloo et al. 2007;Serra et al. 2012), our scaling relations are mainly derivedfrom late type galaxies. Thus they are not suited to estimate c (cid:13) , 1–, 1–
Sky distribution of the H i deficiency parameter in 20 two dimensional bins overplayed with HyperLEDA density contours.The colours represent average H i deficiencies of different areas. Red and orange regions have on average more H i -deficient galaxies anddark blue regions have on average more H i rich galaxies than the green and light blue regions. Density contours are 10, 30, 50, 70, 90,110 galaxies. Black dots represent the individual galaxies of our HOPCAT and NOIRCAT sample. by using the B -band Tully-Fisher relation to estimate theirvelocity width. Using this and their predicted integrated fluxwe calculate the peak flux for our sample as well. After ap-plying the integrated flux detection limit and the 3 σ peakflux cut 749 galaxies remain in our sample.Further analysis showed that a significant fraction of this sample are elliptical galaxies (30 %), based on deVaucouleurs morphology classifications from HyperLEDA.Whilst a number of early-type galaxies contain significantamounts of H i (e.g. Sadler et al. 2002; Oosterloo et al. 2007;Serra et al. 2012), our scaling relations are mainly derivedfrom late type galaxies. Thus they are not suited to estimate c (cid:13) , 1–, 1– ?? H. D´enes, V. A. Kilborn, B. S. Koribalski
Figure 8.
Sky distribution of the H i deficiency parameter in 20 two dimensional bins overplayed with HyperLEDA density contours.The colours represent average H i deficiencies of different areas. Red and orange regions have on average more H i -deficient galaxies anddark blue regions have on average more H i rich galaxies than the green and light blue regions. Density contours are 10, 30, 50, 70, 90,110 galaxies. Black dots represent the individual galaxies of our HOPCAT and NOIRCAT sample. the H i mass of early type galaxies. We investigated how toexclude early type galaxies from our sample, where morpho-logical classifications are not available. The available 6dFGScolours ( B-R , B-J etc.) are not suitable for a colour cut, asthere is only a small offset between red and blue galaxies inthe optical colour-magnitude diagrams (Proctor et al. 2008). UV-optical colours would be needed for an efficient separa-tion of the blue and red sequence (Strateva et al. 2001). Theconcentration index is an indicator of galaxy morphologyand can be used to separate galaxies in the blue cloud andthe red sequence (Driver et al. 2006; Baldry et al. 2006).Thus we use a 2MASS K-band concentration index cut at c (cid:13) , 1– ?? ew H i scaling relations to probe the H i content of galaxies via global H i -deficiency maps −1.0 −0.5 0.0 0.5 1.0 DEF B N Hydra,Antlia −1.0 −0.5 0.0 0.5 1.0
DEF B N Centaurus −1.0 −0.5 0.0 0.5 1.0
DEF B N NGC 3256 −1.0 −0.5 0.0 0.5 1.0
DEF B N Virgo M −1.0 −0.5 0.0 0.5 1.0
DEF B N Virgo Y −1.0 −0.5 0.0 0.5 1.0
DEF B N Sculptor bg.
Figure 10.
Histograms of the H i deficiency parameter for the H i -deficient and H i -excess regions between 2000 km s − < v < − (first panel of Figure 8). distance [Mpc] −1.5−1.0−0.50.00.51.01.52.0 D E F ( f r o m B m a g n i t u d e ) Figure 9.
The H i deficiency factor plotted against the distanceof the galaxies. The shape of this distribution is very similar tothe shape of the H i mass - distance plot. We can only detect H i -deficient galaxies that are relatively close by because the surveysensitivity limit. Dashed lines mark deficiency factor of ± . K conc < . i mass, integrated H i flux and the velocitydistribution of these galaxies. Properties of 6dFGS galaxiesare in blue and we mark the known early type galaxies inred. We still have a small fraction of early type galaxies (11%) in our sample. These are galaxies that passed the 2MASSconcentration index cut, but for which morphological clas-sifications from HyperLEDA indicates that they are ellipti-cals. We also show the same properties of HICAT galaxies in the same volume ( | b | < ◦ ) in grey. The predicted H i massand integrated H i flux distributions have a similar shapeto the distributions of HIPASS. Both H i mass distributionpeak around H i mass (10 . M (cid:12) ). The lack of predicted lowmass galaxies is due to the lack of low velocity galaxies in6dF (Figure 12).We compare the galaxies predicted to be detectable inHIPASS to the HIPASS source catalogue HICAT and find246 (60 %) galaxies matching. (We match coordinates within15’ and ±
300 km s − ). We compare the predicted to themeasured H i masses of the matched galaxies and find a goodagreement, considering the scatter of our scaling relations(Figure 13). The standard deviation between the logarithmof the observed and the predicted H i mass is 0.44 dex. Thisis slightly larger than the standard deviation for our scalingrelations, but considering the difference in sample size it isin good agreement with our scaling relations.There are 164 galaxies that we predict to be detectablein HIPASS, but are not in HICAT. We examine the HIPASSdata cubes at the positions of these galaxies and measuretheir H i fluxes using the Miriad (Sault et al. 1995) routine mbspec . We estimate the profile width for the flux integra-tion using the Tully-Fisher relation for the B -band absolutemagnitude. We are able to measure the integrated flux for101 galaxies (61%) and find an average H i mass of 1 . × M (cid:12) . After visual examination of the derived spectra we con-clude that some of these galaxies show a typical H i profilein the data cubes, but they did not qualify for the sourcecatalogue because of low signal to noise, strong baseline rip-ples or RFI. A detailed investigation of these galaxies mayyield additional HIPASS detections, but is beyond the scopeof this work. We are unable to measure an H i flux for 63galaxies. Possible reasons for this can be poor quality H i data; poor quality photometric data used for the predic- c (cid:13) , 1– ?? H. D´enes, V. A. Kilborn, B. S. Koribalski velocity [km s −1 ] N HOPCAT6dFHyperLEDA
Figure 11.
Velocity distribution of the optically identified galax-ies in the southern HIPASS sample compared to galaxies in6dFGS and galaxies from HyperLEDA ( δ < ◦ and M B < log (M HI /M ⊙ ) HICAT l o g ( M H I / M ⊙ ) d F p r e d i c t e d ( B m a g ) Figure 13.
Comparison of the predicted H i mass from 6dF B -band magnitude against the observed H i mass from HICAT. Thesolid grey line is the 1-1 line, the dashed lines mark ± ± tions; over prediction of the H i mass for a class of galaxies;early-type galaxies in the sample or H i -deficient galaxies.We conclude that it is possible to get a good estimateof the outcome of future H i surveys using H i scaling rela-tions and an optical redshift survey. Using data from the6dFGS redshift survey and 2MASS we predicted 410 galax-ies to be detectable in HIPASS, which is about 10% of thegalaxies in the HIPASS catalogue. This is a reasonable out-come considering the significantly smaller number of nearbygalaxies in 6dFGS and the different galaxy populations ofthe two surveys. 6dFGS is more sensitive to early-type gaspoor galaxies whereas HIPASS is sensitive to late-type gasrich galaxies. To predict a whole blind H i survey, reliableoptical photometry, redshift and basic morphology data isneeded for all galaxies in the volume. H i scaling relationscan also aid the fast identification of unusual sources andthe investigation of detection limits. We derive new multi-wavelength H i scaling relations forgalaxies using the H i Parkes All Sky Survey (HIPASS) anda variety of optical and near-infrared luminosities and diam-eters. We find the B -band scaling relations have the lowestscatter and therefore are the best to predict the H i mass of agalaxy. The scaling relations in all wavebands are sufficientto give a good estimation of H i mass for late-type galaxy.We investigate the environmental dependency of the H i content of galaxies in two different environment densities.We find that galaxies in the high density environment tendto have on average less H i than galaxies with the samestellar mass in the low density environment. We also findthat galaxies in the high density environment tend to havea smaller stellar mass and lower luminosity than galaxies inthe low density environment.H i scaling relations are useful tools to identify galaxieswith anomalous H i content, i.e. H i -deficient and H i -excessgalaxies. These galaxies may have been affected by recentgas stripping or gas accretion. We find 4% of the galaxies inour samples to be H i -deficient and 1% to have H i -excess.We find that about half of the galaxies that we identify asH i -deficient are classified as members of galaxy groups orclusters in NED.We map the global distribution of H i -deficient and H i -excess galaxies on the sky and compare it to the large scalestructure of galaxies. We find several regions on the sky thathave an over density of H i -deficient and H i -excess galaxiescompared to the average. The H i -deficient regions correlatewith dense regions on the sky and the H i -excess regionsalign with the edges of galaxy clusters and groups. This isthe first time regions with an abundance of H i rich galaxieswere identified. We also identify the main galaxy groups andclusters that are aligned with the H i -deficient and H i -excessregions.We show the potential of using H i scaling relations topredict future H i surveys based on an optical redshift sur-vey. We predict the H i mas of 16709 galaxies from the 6dFredshift survey and and find that 410 galaxies should be de-tectable in HIPASS. 60% of these galaxies are in the HIPASScatalogue. We conclude that it is possible to get an estimateof the outcome of future H i surveys using H i scaling re-lations and an optical redshift survey. To predict a wholeblind H i survey, reliable optical photometry, redshift andbasic morphology data is needed for all galaxies in the vol-ume. H i scaling relations can aid the fast identification ofH i -deficient and H i -excess sources and the investigation ofdetection limits.Investigating H i scaling relations in the southern hemi-sphere is especially important now, in preparation for theupcoming large H i and optical surveys, such as the ASKAPH i All Sky Survey, known as WALLABY (Koribalski &Staveley-Smith (2009); Koribalski 2012), SkyMapper (Kelleret al. 2007) and future surveys with the Square KilometreArray (SKA).
We would like to thank Barbara Catinella and Ivy Wong fortheir very helpful comments and conversations. c (cid:13) , 1– ?? ew H i scaling relations to probe the H i content of galaxies via global H i -deficiency maps log (M HI /M ⊙ ) N integrated flux [Jy km s −1 ] N velocity [km s −1 ] N predicted from 6dFearly-typesHIPASS Figure 12.
Distribution predicted H i properties of 6dF galaxies from B -band magnitudes (blue hatched) compared to HIPASS (grey, | b | > ◦ ). Red steps represent early-type galaxies (9%) remaining in the sample after the concentration index cut. REFERENCES
Adelman-McCarthy J. K., Ag¨ueros M. A., Allam S. S., etal. 2007, ApJS, 172, 634Baldry I. K., Balogh M. L., Bower R. G., Glazebrook K.,Nichol R. C., Bamford S. P., Budavari T., 2006, MNRAS,373, 469Barnes D. G., Staveley-Smith L., de Blok W. J. G., et al.2001, MNRAS, 322, 486Bekki K., Couch W. J., 2011, MNRAS, 415, 1783Bekki K., Couch W. J., Shioya Y., 2002, ApJ, 577, 651Bell E. F., de Jong R. S., 2001, ApJ, 550, 212Brough S., Forbes D. A., Kilborn V. A., Couch W., 2006,MNRAS, 370, 1223Butcher H., Oemler Jr. A., 1978, ApJ, 219, 18Catinella B., Schiminovich D., Kauffmann G., et al. 2010,MNRAS, 403, 683Chamaraux P., Balkowski C., Fontanelli P., 1986, A&A,165, 15Chamaraux P., Masnou J.-L., 2004, MNRAS, 347, 541Chung A., van Gorkom J. H., Kenney J. D. P., Crowl H.,Vollmer B., 2009, AJ, 138, 1741Cortese L., Catinella B., Boissier S., Boselli A., Heinis S.,2011, MNRAS, 415, 1797Cowie L. L., Songaila A., 1977, Nat, 266, 501Davies R. D., Lewis B. M., 1973, MNRAS, 165, 231Doyle M. T., Drinkwater M. J., Rohde D. J., et al. 2005,MNRAS, 361, 34Dressler A., 1980, ApJ, 236, 351Driver S. P., Allen P. D., Graham A. W., Cameron E., LiskeJ., Ellis S. C., Cross N. J. G., De Propris R., Phillipps S.,Couch W. J., 2006, MNRAS, 368, 414Driver S. P., Popescu C. C., Tuffs R. J., Graham A. W.,Liske J., Baldry I., 2008, ApJ, 678, L101Duc P.-A., Papaderos P., Balkowski C., Cayatte V., ThuanT. X., van Driel W., 1999, A&AS, 136, 539Eckert D., Molendi S., Paltani S., 2011, A&A, 526, A79English J., Koribalski B., Bland-Hawthorn J., FreemanK. C., McCain C. F., 2010, AJ, 139, 102Fasano G., Poggianti B. M., Couch W. J., Bettoni D., Kjær-gaard P., Moles M., 2000, ApJ, 542, 673Feigelson E. D., Babu G. J., 1992, ApJ, 397, 55Giovanelli R., Haynes M. P., 1983, AJ, 88, 881Giovanelli R., Haynes M. P., 1985, ApJ, 292, 404Giovanelli R., Haynes M. P., Kent B. R., et al. 2005, AJ, c (cid:13) , 1–, 1–
Adelman-McCarthy J. K., Ag¨ueros M. A., Allam S. S., etal. 2007, ApJS, 172, 634Baldry I. K., Balogh M. L., Bower R. G., Glazebrook K.,Nichol R. C., Bamford S. P., Budavari T., 2006, MNRAS,373, 469Barnes D. G., Staveley-Smith L., de Blok W. J. G., et al.2001, MNRAS, 322, 486Bekki K., Couch W. J., 2011, MNRAS, 415, 1783Bekki K., Couch W. J., Shioya Y., 2002, ApJ, 577, 651Bell E. F., de Jong R. S., 2001, ApJ, 550, 212Brough S., Forbes D. A., Kilborn V. A., Couch W., 2006,MNRAS, 370, 1223Butcher H., Oemler Jr. A., 1978, ApJ, 219, 18Catinella B., Schiminovich D., Kauffmann G., et al. 2010,MNRAS, 403, 683Chamaraux P., Balkowski C., Fontanelli P., 1986, A&A,165, 15Chamaraux P., Masnou J.-L., 2004, MNRAS, 347, 541Chung A., van Gorkom J. H., Kenney J. D. P., Crowl H.,Vollmer B., 2009, AJ, 138, 1741Cortese L., Catinella B., Boissier S., Boselli A., Heinis S.,2011, MNRAS, 415, 1797Cowie L. L., Songaila A., 1977, Nat, 266, 501Davies R. D., Lewis B. M., 1973, MNRAS, 165, 231Doyle M. T., Drinkwater M. J., Rohde D. J., et al. 2005,MNRAS, 361, 34Dressler A., 1980, ApJ, 236, 351Driver S. P., Allen P. D., Graham A. W., Cameron E., LiskeJ., Ellis S. C., Cross N. J. G., De Propris R., Phillipps S.,Couch W. J., 2006, MNRAS, 368, 414Driver S. P., Popescu C. C., Tuffs R. J., Graham A. W.,Liske J., Baldry I., 2008, ApJ, 678, L101Duc P.-A., Papaderos P., Balkowski C., Cayatte V., ThuanT. X., van Driel W., 1999, A&AS, 136, 539Eckert D., Molendi S., Paltani S., 2011, A&A, 526, A79English J., Koribalski B., Bland-Hawthorn J., FreemanK. C., McCain C. F., 2010, AJ, 139, 102Fasano G., Poggianti B. M., Couch W. J., Bettoni D., Kjær-gaard P., Moles M., 2000, ApJ, 542, 673Feigelson E. D., Babu G. J., 1992, ApJ, 397, 55Giovanelli R., Haynes M. P., 1983, AJ, 88, 881Giovanelli R., Haynes M. P., 1985, ApJ, 292, 404Giovanelli R., Haynes M. P., Kent B. R., et al. 2005, AJ, c (cid:13) , 1–, 1– ?? H. D´enes, V. A. Kilborn, B. S. Koribalski c (cid:13) , 1– ?? ew H i scaling relations to probe the H i content of galaxies via global H i -deficiency maps c (cid:13) , 1–, 1–