A&A 417, 499-514 (2004)
DOI: 10.1051/0004-6361:20034105
G. Gavazzi1 - A. Zaccardo1 - G. Sanvito1 - A. Boselli2 - C. Bonfanti1
1 - Università degli Studi di Milano - Bicocca, Piazza delle scienze 3,
20126 Milano, Italy
2 -
Laboratoire d'Astrophysique de Marseille, Traverse du Siphon, 13276 Marseille
Cedex 12, France
Received 23 July 2003 / Accepted 16 December 2003
Abstract
Drift-scan mode (3600-6800 Å) spectra with
500<R<1000 resolution are presented
for 333 galaxies members of nearby clusters, covering the whole Hubble sequence. The majority (225) were obtained
for galaxies in the Virgo cluster where a completeness of 36%, if all Hubble types are considered, and of 51%,
restricting to late-types, was reached at
16. Our data
can be therefore considered representative of the integrated spectral
properties of giant and dwarf galaxies in this cluster. Intensities and
equivalent widths (EWs) are derived for the principal lines, both in emission and
in absorption. Deblending of the underlying absorption from emission was achieved in most cases.
Key words: galaxies: clusters: individual: Virgo - techniques: spectroscopic - catalogs
Local galaxies are the relics of evolutionary processes that took place in the universe from the early collapse of primordial matter fluctuations up to the present cosmological epoch. Such processes have not yet been convincingly unveiled, in spite of an increasing observational effort involving today's major observational facilities. A satisfactory characterization of the properties of local galaxies is building up only recently, as data obtained through a variety of observational windows of the electromagnetic spectrum are being gathered (see Kennicutt 1998; Roberts & Haynes 1994; Gavazzi et al. 1996; Tuffs et al. 2002; Popescu et al. 2002). Complete imaging data sets taken in a broad frequency range are becoming available owing to extensive observational campaigns, e.g. 2MASS (Jarrett et al. 2003), SLOAN (Stoughton et al. 2002), SINGS (Kennicutt et al. 2003) just to mention few.
Table 1: The spectrograph characteristics.
As far as nearby clusters, such as the Virgo cluster and the Coma supercluster, multifrequency data for over 3000 galaxies are collected and distributed via the WEB site "GOLDMine'' (Gavazzi et al. 2003).
Spectroscopic data are equally invaluable sources of information, but are more difficult to obtain and more time consuming. The spectroscopic characterization of the stellar continua provide us with "clocks'' on stellar populations, while line indices enable us to quantify the chemical evolution of the stars and of the interstellar medium in galaxies. However much less extensive surveying was carried out in the spectroscopic than in the imaging mode, if one excludes the pioneering work of Kennicutt (1992, K92 hereafter) who first tried to assess the systematic spectral properties of nearby galaxies along the Hubble sequence and of Jansen et al. (2000) who extended the spectral survey of K92 to a large sample of isolated galaxies, later analyzed by Stasinska & Sodre (2001). These spectroscopic surveys were carried out in the drift-scan mode, i.e. with the slit sliding over the whole galaxy area. Spectra taken in this way are representative of the mean galaxies, unlike most long slit observations which are dominated by the nuclear light.
In 1998, inspired by the work of K92 and of Jansen et al. (2000) we initiated a long term project aimed at characterizing spectroscopically the galaxies in the nearest rich cluster: the Virgo cluster. In Paper I of this series (Gavazzi et al. 2002a) we analyzed the spectral continua based on preliminary 124 spectra obtained until 2001. Here we present the full set of 333 spectra obtained so far (spring 2003). They are available in JPG and FITS format at the WEB site GOLDMine (http://goldmine.mib.infn.it) (Gavazzi et al. 2003). Their analysis is postponed to Paper III (in preparation).
The present paper is organized as follows: Sect. 2 describes the surveyed sample, the observational and data reduction techniques. Section 3 gives the details of the line measurements, including the deblending of unresolved lines and the separation of emission from underlying absorption lines. The derived spectral parameters are given in Sect. 4 and briefly summarized in Sect. 5.
Table 2: The number of observed spectra.
Table 3: Completeness of spectroscopic observations for Virgo cluster members and possible members regardless of their Hubble type and for late-type galaxies.
Long-slit spectra of 333
galaxies, obtained during approximately 50 nights distributed in 6 years (1998-2003) using the
telescope of the Observatoire de Haute Provence (OHP),
the ESO/3.6 m telescope, the Loiano/1.52 m telescope and the San Pedro Martir (SPM) 2.1 m telescope are presented. The observations were taken in the "drift-scan'' mode: i.e. with the slit, generally parallel to the galaxy major axis, drifting over the optical surface of the galaxy
.
At ESO we set the guide velocity of the telescope such that during the integration time the slit slides one time through the full length of the galaxy. At the remaining observatories the drifting was obtained by slewing manually several times the telescope between two extreme positions checked on one offset star or on the galaxy itself. Not unexpectedly spectra obtained in this way have lower S/N ratio than traditional long-slit spectra of similar integration time, because a large fraction of the time is spent on low surface brightness regions. Our spectra cover the wavelength range 3600-6800 Å (from [OII] to [SII]) with a resolution of 500<R<1000. The spectrograph characteristics are given in Table 1. For 6 bright emission line galaxies observed at ESO we used both a low resolution grism and a high resolution red grism.
The observations at OHP, Loiano and San Pedro Martir were carried out in approximately 1.5-3 arcsec seeing conditions, while subarcsecond conditions were often encountered at ESO. The present data, owing to the "drift-scan'' method are marginally affected by the seeing conditions. The OHP observations were sometimes taken through cirrus, otherwise in transparent or photometric conditions, while the observations obtained at ESO, SPM and Loiano were transparent or photometric. The spectrophotomectric standards Feige 34, Hz 44 and Hiltner 600 (ESO) were observed twice on each night.
Targets of the present spectrophotometric measurements were primarily selected from the Virgo Cluster Catalog (Binggeli et al. 1985: VCC). Among these we observed 225 objects.
Limiting to the 621 galaxies with mp
16 which are Virgo cluster members (V<3000 or classified as possible members by Binggeli et al. 1985, 1993; Gavazzi et al. 1999),
223 have their spectra measured. At this limiting magnitude the completeness of our spectroscopic
work is thus 36% and it increases to 46% at mp
15 and to 62% at mp
14, as listed in Table 3. Most unobserved galaxies are dE and E, therefore the completeness results significantly higher among late-type galaxies. In these morphological classes we covered more than 50% of
the galaxies with mp
16. At the adopted distance of 17 Mpc (or
)
mp=16 corresponds to Mp=-15, thus our survey can be considered as representative of the spectroscopic properties of late-type galaxies in the Virgo cluster including dwarf systems.
Observations of CGCG (Zwicky et al. 1961-68) galaxies with mp
15.7 in other nearby clusters (45 in Coma+A1367, 37 in Cancer, 8 in the A262 and Centaurus clusters and another 10 isolated objects) were taken as fillers when Virgo was not observable. These do not form a complete set.
Table 2 summarizes the number of obtained spectra in each run and cluster and the approximate seeing conditions.
General parameters derived from the literature for galaxies in the observed sample, along with the log-book of the observations, are given in Table 6, arranged as follows:
The reduction of the spectra was carried out using standard
tasks in the IRAF package.
Visual inspection of the raw images provided us with a list of bad-pixel which were masked from the science frames. Bias subtraction and flat-field normalization was applied using median of several
bias frames and exposures of quartz lamps. When at least three exposures were obtained for
an object (see "number of exposures'' in Col. 17 of Table 6), they were combined using a median filter, thus removing the cosmic rays. Otherwise the point-like ones were subtracted using COSMICRAY and the remaining extended features were removed under visual inspection of the spectra.
The
calibration was carried out using
IDENTIFY-REIDENTIFY-FITCOOR on exposures of
He/Ar lamps and the calibration was transferred to the science frames using TRANSFORM.
Typical errors on the dispersion solution are of few tenths of
,
as confirmed from the measurements of the sky lines.
The two-dimensional frames were sky subtracted using
BACKGROUND. One-dimensional spectra
were obtained integrating the signal along the slit using APSUM.
The apertures were limited to regions where the signal intensity was above
1
of the sky noise.
The flux calibration was achieved using
STANDARD-SENSFUNC-CALIBRATE on spectra of
the standard stars Feige 34, Hz 44 and Hiltner 600 (ESO) taken twice on each night.
Cubic spline sensitivity functions of 20th order or higher were fit to the calibration spectra, allowing the transformation of the measured intensities into flux densities (
), including the atmospheric extinction correction. However, because during an exposure taken in the "drift-scan'' mode
the fraction of light collected by the slit changes with time, an absolute calibration of our science spectra cannot be achieved. Thus all spectra were normalized to their intensity at
Å.
The spectrophotomectric standards were instead used to calibrate the absolute response
of the system as a function of wavelength. The uncertainty of this measurement over the whole
spectral region resulted within 15%,
as derived comparing the spectra of four galaxies taken in this work with the corresponding
ones by K92 (see Sect. 2.3).
![]() |
Figure 1: Comparison between the spectra of NGC 1357 and 1832 obtained in this work and by K92. ( top panels). The ratio of the two measurements is given in the bottom panels. |
Open with DEXTER |
![]() |
Figure 2: Same as Fig. 1 for NGC 3379 and 6181. |
Open with DEXTER |
![]() |
Figure 3: Comparison of the synthesized B-V color from our spectra and from photometry. Unless otherwise specified, in this and following figures dE-E-S0a are represented with circles; Sa-Sb with squares; Sbc-Scd with triangles; Sd-BCD with pentagons. The dashed line represents the best linear fit to the data. |
Open with DEXTER |
Four bright galaxies (NGC 1357, 1832, 3379 and 6181) were measured in common with K92 and their
comparison is useful to assess the quality of our data. Figures 1 and 2 show the spectra of these objects as obtained by us (top spectra) and as given by K92 (bottom spectra). The ratio of the two measurements is shown in the bottom panel of each figure. The two sets of data
are found in agreement within 15% in the range 3800-6800 .
Synthesized spectroscopic colors B-V and B-R were obtained deconvolving the continua
with the profiles of the B, V and R Johnson filters.
Since the width of the R filter (5500-7000 )
exceeds by 200
in the red the domain covered by our spectra, B-R is computed deconvolving the Bruzual & Charlot (1993) population synthesis models fit to the observed spectra (see Paper I) with the B and R filter profiles. Figure 3 shows the comparison between the photometric colors (B-V)T and the B-V color synthesized on the spectra for 312 galaxies.
The two differ by 0.05 mag with an rms scatter of 0.15 mag.
The comparison of the synthesized B-R color from our spectra with (B-V)T color from photometry is given in Fig. 4. Unless otherwise specified, in this and in the following figures we plot the best fit linear regression obtained using the bysector method of Feigelson & Babu (1992). The results of the linear regression analysis, including the uncertainties in their slope and zero point are summarized in Table 4.
![]() |
Figure 4: Comparison of the synthesized B-R color from our spectra with (B-V)T color from photometry. Same symbols as in Fig. 3. The dashed line represents the best linear fit to the data. |
Open with DEXTER |
Under visual inspection to the spectra we carried out a first-order measurement to all lines, both in emission and in absorption, using SPLOT. This provided a list of fluxes and EWs with respect to a user defined continuum level. This preliminary measurement was then refined as described in the following sections.
Table 4: The bysector linear regression analysis.
(
)
is bracketed by the weaker [NII] doublet ([NII1]
and [NII2]
).
The three lines are clearly resolved in the OHP (R = 1000) spectra, thus for spectra taken at OHP
the measurements of the individual lines is reliable.
In the lower resolution (R = 500) spectra taken at ESO, Loiano and SPM the three lines are not well resolved, thus the deblending obtained with SPLOT is often inaccurate.
In most of these cases we could only measure the global
flux T of the triplet
+[NII1]+[NII2].
To measure them individually we proceed as follows:
we calibrate on the OHP spectra the empirical relation between
(([NII1]+[NII2])/
)
and the V-band luminosity shown in Fig. 5,
which derives from the well known metallicity-luminosity relation (see Raimann et al. 2000). We find:
![]() |
Figure 5:
![]() |
Open with DEXTER |
![]() |
Figure 6:
Relation between [NII] and H![]() |
Open with DEXTER |
We apply the above procedure only if [NII2] emission is detected, but not resolved from
,
i.e. when the line separation is lower than the sum of the individual HWHM.
Otherwise
and [NII2] were measured individually and [NII1] was set as 0.34 [NII2].
The above procedure was checked a posteriori using 6 emission-line galaxies observed at ESO with both the low and the high resolution grisms. Figure 6 illustrates
the consistency of the EW of [NII2] and
obtained on the low resolution spectra
applying the deblending procedure with those directly measured on the HD spectra. The deblended measurements are found overestimated by 20% on average.
Several (223) galaxies in our sample have their
+[NII] measured
from imaging (Boselli & Gavazzi 2002; Boselli et al. 2002; Gavazzi et al. 2002b).
In spite of the different measuring techniques,
the equivalent width derived from imaging and from our spectra are found within 0.31 dex rms,
as shown in Fig. 7. This confirms that the "drift-mode'' spectroscopy
is indeed representative of the entire galaxy, as claimed in the introduction.
![]() |
Figure 7:
Comparison of
![]() |
Open with DEXTER |
Due to low sensitivity in the blue of CARELEC, only 22% of the emission line objects
(
0) observed at OHP have [OII](
3727) detected, as opposed to 79%
in ESO spectra whose rms noise at 4000 Å is half that of OHP spectra.
For the remaining OHP spectra we estimate
upper limits to the strength of [OII]
as 3
7, where 7 Å is the mean FWHM of the [OII] lines detected at OHP.
![]() |
Figure 8:
Enlargement of the spectrum of VCC 25 to illustrate the deblending of
![]() ![]() |
Open with DEXTER |
Most emission line galaxies show evidence for underlying absorption in correspondence to emission lines. In particular, out of 174 spectra where we could measure
in emission, in 151 cases we detect significant
in absorption, and in fewer cases
and
as well. We deblended the underlying absorption from the emission lines using a multiple component fitting procedure written by us in the IRAF environment. To do so we measure the emission line and subtract it from the spectra. The resulting absorption line is also measured with respect to a reference continuum. These two measurements are used as first guess in a
fitting algorithm which fits jointly the emission and absorption lines to the reference continuum (see Fig. 8). The value of the emission lines is given in Table 7, that of the underlying absorption in Table 8.
![]() |
Figure 9:
Histogram of the underlying EW in absorption at
![]() |
Open with DEXTER |
Table 5:
Dereddening law relative to
.
As shown in Fig. 9 the distribution of the underlying
is peaked at 5.7
1.9 Å, consistently with K92 who reported a mean underlying
of 5 Å.
Table 6: Parameters of the observed galaxies (one page sample).
Table 7: Emission lines (one page sample).
Table 8: Balmer absorption lines (one page sample).
For objects whose
was detected in emission but the deblending procedure was not applied (no absorption feature was evident)
a mean additive correction for underlying absorption equal to -1.8 in flux and -1.4 Å in EW was used. These values correspond to the fraction of the (broader) absorption feature that lies under the emission feature.
No mean correction was applied to other lines except
.
![]() |
Figure 10:
C1 versus
![]() |
Open with DEXTER |
From
corrected for underlying absorption (Sect. 3.4) and
corrected for
deblending from [NII] (Sect. 3.1) we evaluate the Balmer decrement:
The ratio
depends on the electron density and on the gas temperature. Assuming T= 10 000 K and n= 100
,
as in Osterbrock (1989) case B,
holds.
The corrected line fluxes are derived, relative to
,
using C1 and the reddening
function
of Lequeux et al. (1979) (see Table 3.5) based on the extinction law of Whitford (1958)
.
When EW
> 1 Å but
is undetected we derive a
lower limit to C1 using (Buat et al. 2002):
Figure 10 shows the obtained C1 as a function of the synthetic B-R color index
for our objects (coded according to the morphological type) and for
galaxies observed by Jansen et al. (2000) and analyzed by Stasinska & Sodre (2001).
Our data confirm the positive correlation between the two quantities: i.e. increasing
C1 with increasing B-R. However the dispersion appears higher and the relation steeper than in
Stasinska & Sodre (2001). Notice that
these authors did not measure the underlying absorption at
.
The 333 (rest-framed and normalized) spectra obtained in this work are illustrated in Fig. 19.
The ESO spectra are given in the range 3600-6800 .
The OHP spectra, noisier
in the blue, were resampled with a step of 5
and are given from 3850 to 6800
,
unless the strength of the [OII] line was higher than 3 times
the noise determined locally near the line (see Sect. 3.3).
The flux scale of Fig. 19 is given in three intervals: 0.2-2; 0.2-5, 0.2-15 according to the intensity of the brightest lines. The additional 6 high dispersion spectra obtained at ESO are given in the last page of Fig. 19.
All spectra presented in Fig. 19 are available in the FITS format at the WEB site GOLDMine (http://goldmine.mib.infn.it/).
The (corrected) emission lines parameters are listed in Table 7:
![]() |
Figure 11:
Distribution of H![]() |
Open with DEXTER |
![]() |
Figure 12:
Same as in Fig. 11 for H![]() |
Open with DEXTER |
![]() |
Figure 13: Same as in Fig. 11 for [OIII2] EW. |
Open with DEXTER |
Table 9: Absorption indices.
![]() |
Figure 14:
The ![]() |
Open with DEXTER |
Whenever an absorption feature is detected in correspondence of Balmer lines (either alone or deblended from emission, as discussed in Sect. 3.4) its EW is listed in Table 8 as follows:
Frequency distributions of the EWs of the principal (emission and absorption) lines
(
,
and [OIII] (
)) in 4 intervals of Hubble type are given in Figs. 11-13 respectively.
For each Hubble type interval, the total number of objects and the number of objects with a given
measured line (either in emission or in absorption) is labeled in each panel.
The continuum lines represent similar frequency distributions obtained with the
Jansen et al. (2000) data.
If one excludes the absorption lines (negative EW) that Jansen et al. (2000) did not measure,
whereas they are included in our analysis, the two distributions appear consistent one another. In fact the probability that the two distributions are derived from the same parent populations is:
>62% for S0-Sab, >97% for Sb-Sd and >57% for Sdm-BCD, as derived from the Kolmogorov-Smirnov test (dS0-E are excluded from this analysis due to the poor statistics).
The implications of this finding is that, to the first order, galaxies in rich clusters do not have
emission line properties dramatically different from isolated galaxies. Small differences, if any,
require a more subtle analysis to be identified, which is postponed to Paper III.
The absorption lines indices are derived for early-type objects (dE-dS0-E-S0) and are listed in Table 9. They are derived according to the Lick system (Worthey et al. 1994) as the result of the integration in defined bands. Continua are determined on both sides of the line under measurement, and averaged. The line strength (or EW)
is given as the difference between the integral in the interval containing the line
and in the adjacent continuum.
Table 9 is organized as follows:
![]() |
Figure 15:
The ![]() |
Open with DEXTER |
![]() |
Figure 16:
The ![]() ![]() |
Open with DEXTER |
![]() |
Figure 17:
The ![]() |
Open with DEXTER |
![]() |
Figure 18:
The ![]() ![]() |
Open with DEXTER |
Using 5 middle-size telescopes for 50 nights distributed over 6 years we obtained drift-scan spectra (3600-6800 Å) with 500<R<1000 for 333 galaxies in nearby clusters. The majority (225) were secured for galaxies in the Virgo cluster. The observations can be considered representative of the spectral properties of giant and dwarf galaxies in this cluster, as the completeness achieved at Mp=-15 is 36% for all types and 51% for late-type galaxies.
![]() |
Figure 19: The observed spectra. The galaxy identification, morphological type, membership, photographic magnitude and observing run are labeled on each panel (one page sample). |
Open with DEXTER |
Here we present the individual spectra reduced to their rest-frame wavelength and normalized to their intensity at 5500 Å.
Intensities (corrected for dereddening) and EWs are derived for the principal lines both in emission and in absorption. Special care is devoted to deblending of
from the [NII] doublet and of emission lines from underlying absorption. In the case of
we measured underlying absorption 87% of the times we detected emission, with a mean EW of 5.7 Å.
For early-type galaxies we derive the Lick absorption line indices.
The complete line analysis is postponed to a forthcoming Paper III where the metallicity indicators will be derived and analyzed.
The comparison of the line properties of our cluster sample with those of 200 isolated galaxies observed by Jansen et al. (2000) with a similar experimental setup will allow to study the influence of the cluster environment on the interstellar medium and on the stellar population of galaxies. This analysis will help shed light on the elusive processes that made cluster galaxies evolve separately from their isolated counterparts.
Acknowledgements
This research has made use of the NASA/IPAC Extragalactic Database (NED) and of the GOLDMine database. NED is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. GOLDMine is operated by the Universita' degli Studi di Milano-Bicocca.