GMRT Observation of H I Gas
in the COSMOS field at z~0.37
Jonghwan Rhee1,2,3!
1International
Centre for Radio Astronomy Research, University of Western Australia!
2Research School of Astronomy & Astrophysics, Australian National University!
3ARC Centre of Excellence for All-Sky Astrophysics (CAASTRO)!
In collaboration with:!
Frank H. Briggs (ANU), Philip Lah (ANU), Jayaram N. Chengalur (NCRA)
H I gas evolution study
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
2
H I in the local universe
•
H I gas quantified from 21-cm blind surveys:
-
-
HIPASS: H I Parkes All Sky Survey (Zwaan et al. 2005)
survey area 21341 deg2, 4315 detections, z < 0.042
ALFALFA: Arecibo Legacy Fast ALFA survey (Martin et al. 2010)
survey area 2799 deg2, 10119 detection, z < 0.06
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
3
H I gas evolution study
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
4
H I gas at high redshifts
•
•
Knowledge of HI gas from 21-cm observations is
not available.
Different techniques for HI measurement at high
redshift (z>2): Damped Lyman-𝜶 absorption (DLA).
DLA
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
5
H I gas between low and high-z
•
0.2 < z < 1.5 less explored regime by observations
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
6
H I gas between low and high-z
•
•
0.2 < z < 1.5 less explored regime by observations
Both techniques (21-cm and DLA) not practically
available in the redshift range:
- Existing radio observing facilities are not sensitive enough to detect
weak 21cm signal
Observa+on"+me"
#"of"Galaxy"
Zwaan"(2001)"z=0.18"
Verheijen"(2007)"z=0.20"
Ca+nella"(2008)"0.17<z<0.25"
0"
100"
200"
300"
400"
500"
600"
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
7
H I gas between low and high-z
•
•
0.2 < z < 1.5 less explored regime by observations
Both techniques (21-cm and DLA) not practically
available in the redshift range:
- Existing radio observing facilities are not sensitive enough to detect
weak 21cm signal
- Lyman-𝛼 line can be observed in UV wavelength and incidence of
DLA is low.
•
•
Next generation of radio telescopes can reach this redshift range:
strong drive to initiate SKA and SKA pathfinders projects
New techniques emerged to overcome the limitation before the future
telescopes: H I spectral stacking and H I intensity mapping technique
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
8
H I spectral stacking technique
Using known optical data (RA, Dec and Redshift),
H I spectra are extracted from a 3D radio map.
S. Fabello et al.
• Co-add the spectra to obtain average HI spectrum.
•
998
Dec
RA
Figure 4. A schematic representation of the fully processed ALFALFA 3D data cube. The data cubes are 2.4◦ × 2.4◦ in size and about 5500 km s−1 in velocity
Radio Data Cube
range (25 MHz in frequency). The raw spectral resolution is ∼5.5 km s−1 ; the angular resolution is ∼4 arcmin. For each pixel, which is a point in RA, Dec.
and velocity, a value of flux density is recorded. For each target in sample A, we extract a spectrum at a given position of the sky, over the velocity range of
the data cube which contains the source. Two examples of extracted spectra are shown on the right, illustrating an H I detection (green, bottom) and an H I
non-detection (red, top).
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
9
Target Fields for HI stacking
Target!
Field
Redshift!
(z)
Lookback!
time (Gyr)
Obs. Time!
(Radio)
Radio!
Telescope
# of stacked
galaxies
FoV!
(deg)
CNOC2!
0920+37
0.1, 0.2
1~2.7 Gyr
120 hr
WSRT
59, 96
1.0
VVDS14h
0.32
3.6 Gyr
136 hr
GMRT
165
0.98
zCOSMOS
0.37
4.0 Gyr
134 hr
GMRT
474
0.95
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
10
zCOSMOS Field at z~0.37
A large optical redshift survey on 1.7 deg2 COSMOS field.
• zCOSMOS-bright 10 k catalogue used.
• Redshift uncertainty ~110 km s-1
63
4.2 Optical Data
• 474 reliable redshifts available for our analysis
•
objects
denote
z ⇤1. RedHIpoints
sample to
Figure 4.1 Redshift cone diagram of zCOSMOS 10k-bright
2015 PHISCC
Workshop:
Surveys
Get 598
Real
March 16-18, 2015
11
Giant Metrewave Radio Telescope
• One of sensitive telescopes
to survey atomic hydrogen
gas out to z~0.4 before SKA
and its pathfinders.
• Array of 30 dishes of 45m ea.
(14 central antennas, 16 outer
antennas)
• Baseline range: 100m ~ 25km
• Frequency coverage:
L band 1000~1450 MHz
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
12
Radio observation (GMRT)
•
•
•
•
A total observing time:
134 hr for 20 days in 2008 & 2009
(115 hr on-source time)
Bandwidth of 32 MHz, split into
two 16 MHz sidebands (each
128 channels).
Frequency range:
1056 MHz to 1024 MHz
(0.345 < z < 0.387)
Channel width of 0.125 MHz
(~36.3 km s−1at z = 0.37)
Synthesized beam of 3.6″×2.4″
+3 0000000
zCOSMOS
10% Beam
4000000
Declination (J2000)
•
HPBW
2000000
+2 0000000
4000000
+1 2000000
04m00s
02m00s
10h00m00s
9h58m00s
Right Ascension (J2000)
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
13
Data Reduction
68 GMRT Observation of Neutral Hydrogen Gas in the COSMOS Field at
Data Reduction
- Automated flagging software
“flagcal” used before data
reduction. - CASA pipeline developed for
GMRT data reduction.
- Following standard radio data
reduction procedures: flagging,
calibration, and imaging.
Figure 4.5 The fraction of flagged data after all data reduction procedure.
non-coplanarity of the GMRT baseline configuration, the wide-field imaging schem
w projection (Cornwell et al., 2008) was adopted when making the continuum m
RMS noise level in the central area of the continuum image is ⇠12.3 µJy beam
synthesised beam resolution of 3.500 ⇥ 2.400 . Astrometric accuracy of the GMRT ima
examined
using this continuum
which March
source finding
2015 PHISCC
Workshop:
HI Surveysmap
GetinReal
16-18,software
2015 ,
14
Galaxy Classification
10
Type 1
Type 2
Type 3
Type 4
MN U V
Mr
8
6
4
2
0
2
0.0
0.5
1.0
1.5
Mr
3.0
2.0
MJ
2.5
MV
2.5
2.0
MU
1.5
1.0
0.5
0.0
0.0
•
•
0.5
1.0
MV
1.5
MJ
2.0
2.5
3.0
Figure 4.3 colour–colour diagrams of subsamples classified by spectrophotometric approach. The upper
and lower panel are NUV r+ vs. r+ J colour–colour diagram and u⇥ V vs. V J colour–colour diagram,
respectively. Four subsamples from type1 to type 4 are denoted by red circles, green diamonds, blue circles and
squares.
Galaxies classified by a SED template fitting code (LE PHARE
Ilbert at al. (2006))
4 galaxy types are defined:
Type 1 for early types (E/S0)
Type 2 for early spirals(Sa, Sb)
Type 3 for late spirals (Sc, Sd)
Type 4 for irregular and starburts
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
15
Flu
he lower pannel. Many galaxies in the z ∼ 0.1
lightly resolved in the east-west direction.
Late
Stacked H I spectra
0.1
0.0
taining 90 per cent of the Petrosian r-band
S. We also use the SDSS data to determine
he galaxies in the Broeils & Rhee (1997)
ed a relation between the H I and the Petnewly derived relation, we estimate the H I
ies. As seen in Fig. 6, most galaxies have
a few are in the range of 50 to 57 kpc. The
at some of the galaxies in the low redshift
the WSRT synthesised beam is consistent
tional velocity gradients revealed in Fig. 5.
ta cubes, the galaxy H I signal is expected
l resolution beam size.
e success of the WSRT beam at includfor all galaxies, additional sets of image
with larger beams with half-power widths
45′′ , and 60′′ × 60′′ . Along with the default
ower resolution image cubes were carried
cedure as described below.
axy spectra, a weighted average is made.
ded using a weighted average scheme have
primary beam attenuation function as in
annel weight is computed from the image
y the primary beam correction factor at the
he formal average specification is given by:
wi′
1
= ′2 ,
σi
(2)
beam corrected spectrum. The RMS noise,
uency channel in each data cube with apcorrection and the inverse of the square of
−0.1
−1500−1000 −500
0
500 1000 1500
−1500−1000 −500
Rest Frame Velocity (km s 1 )
0
500 1000 1500
Early type
Figure 7. The averaged H I emission spectra of galaxies in each subgroup
after stacking. The top panels are the co-added H I spectra at z ∼ 0.1, for the
different galaxy types, and the bottom panels are for galaxies at z ∼ 0.2. The
vertical dashed line in each panel shows the velocity width within which all
H I emission from galaxies is included. The 1σ error is shown as a horizontal dashed line in each panel. To the right of each panel are the same spectra
Early spirals
but re-binned to a velocity width of 500 km s−1 .
range 1321 MHz to 1263 MHz. The remaining 96 galaxies lie in
the last five sub-bands i.e. in the frequency range 1222 MHz to
1160 MHz. Depending on their classification, (see Fig. 2), eachLate
of spirals /
irregular&starburst
these sub-samples is divided into 3 further groups. Fig. 7 shows the
results of the H I spectra stacking for each sub-sample separately.
The stacked signal shows a clear signal particularly for the intermediate and late-type galaxies.
From the stacked spectrum, one can determine the co-added
flux. This co-added H I flux can be converted to H I mass using the
equation (Wieringa et al. 1992):
236
MH I
=
M⊙
(1 + z)
"
DL
Mpc
#2 " $
SV dV
mJy km s−1
#
(3)
$
where SV dV is the integrated H I emission flux in units of
mJy km s−1 and DL $is the luminosity distance in unit of Mpc.
For the calculation of SV dV it is assumed that a velocity range
∆V (specified in the rest frame of the H I emitter) of 500 km s−1
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
contains all the
H I emission. This assumption takes into account
16
quires an accurate volume normalisation, which in our approach is
dependent on good measurements of the luminosities and the total
luminosity density of the CNOC2 galaxies that are used for the H I
stacking.
We make use of the optical luminosity density of the
CNOC2 0920+37 field as a function of galaxy type from Lin et al.
Table
4. TheThey
H I mass
to luminosity
ratio MH of
as a function
(1999).
investigated
the evolution
the
luminosity
func- of
I /L
B ratio
galaxy
CNOC2
sample
the redshift
of z ∼ densities
0.1 and 0.2,
tiontypes
with of
redshift
and0920+37
galaxy type
andatprovide
luminosity
Luminosity
Function
respectively
in the
M⊙Their
/L⊙ .galaxy classification
ρLB (z ) up
to zunit
∼ of
0.55.
scheme
is the
same as the one we used. The B-band luminosity and luminosity
density of each galaxy type
z ∼are
0.1used in the conversion
z ∼ 0.2 of H I gas
densitySample
as follows:
⟨MH I ⟩/⟨L ⟩
⟨MH I ⟩/⟨L ⟩
and ΩH I =
tively, for al
At z ∼
intermediate
early-type g
early-type g
alaxies were detected at z ∼ 0.1,
higher H I m
pe and 1 late-type) were detected
sensitivity e
Stacked
HI spectra
seen in Table 3, after
removing
that higher
list, 24 per cent and 44 per cent
(Roberts &
diate-type galaxies and late-type
type galaxie
At z ∼ 0.2, while H I mass of
have higher
B
B
!
y 29 per cent, there is no change
and low lum
"
M
H
I
± 0.05 MH I = ⟨M
0.01H ±
0.03
!
ρH I = Early
× ρLB (z0.15
(8)
I⟩ × N
),
redshift bin,
. In spite of the relatively small
L
B
Intermediate
0.22 ± 0.03
0.15 ± 0.04
higher lumi
cts individually detected at high± 0.04 galaxies,0.29
where Late
N is the number0.32
of co-added
LB±is0.04
luminosity in
Fig. 9
ibute to the co-added H I mass.
the B-band taken from the CNOC2 0920+37 catalogue, and ⟨MH I ⟩
and look-ba
denotes the averaged H I mass we measured using the stacking
by Zwaan et
The cosmic
I gas density
is then
H Irespect
gas density
technique.
We H
separately
calculate
H I calculated
gas densityas
with
to
Lah et al. (2
divided
the critical
density
(ρcrit ): and late-type at each redeach by
galaxy
type of early,
intermediate
daeme et al
74 GMRT Observation of Neutral Hydrogen Gas in the COSMOS Field at z ⇠ 0.37
shift bin (see Table 2). In this
(2013). Two
NSITY (ΩH I )
2 equation, we apply a correction to the
ρ
3H
HI
0to prevent H I mass-to-light ratio from
I
mass-to-light
ratio
term
ΩH H
=
,
ρ
=
,
(9) cm emission
I
crit
ρbiased
8πG
galaxies has varied with the coscrit
Table
H masstoward
(MH I ), B-band
luminosity
(⇢L ) and
H massThe
density
(⇢H I ) of each galaxy
type at
being4.2
luminous
anddensity
gas-poor
sample.
detailed
measuremen
the
redshift of of
z ⇠ the
0.37 correction factor is described in Appendix A. The
cosmological H I mass density,
derivation
we de-corre
where
H
is
the
Hubble
and Gstacked
is the gravitational
0
f galaxies in each subgroup (left panels). The right
panels
showconstant
the re-binned
spectra with cona velocity width o
totalFrom
H I gas
density (9)
in the
low
and high
redshift
regimes
in Table
2−3
sity measure
ical density. Calculating ΩH I restant.
equation
we
obtain
Ω
=
(0.33
±
0.05)
×
10
H
I
. The horizontal line in each left panel is the
1obtained
error by
of the
stacked
spectrum.
are
adding
the
H
I −3
density
values i/hL
for the three⇢ galaxy ⇢
Atf z ∼
isation, which in our approach is
Ngal ± hM
i z ∼hM
andSample
ΩH I = (0.34
0.09)
0.1
H I i × 10 hLBat
H I andB iz ∼ 0.2,
LB respec- H I
Luminosity
types.
ALFALFA
of the luminosities
and the total
tively,
for
all
galaxy
types
summed.
Type 1Note that
95 in
2.20
± 2.60
17.76 ±luminosity
0.04 0.12 ±
0.15 5.11
± 0.65
0.89 p
order
to derive
densities,
Lin
et al.0.56 ± 0.67
very good
galaxies that are used for the H I
Size
(arcsec)
At2 z applied
∼ 58
0.2, 1.50
it is± 2.74
found
that± most
of
H I (Ω
gas
resides
in ± 0.38
Type
21.33
0.05 Beam
0.07the
± 0.13
2.69
±=0.65
1.10
(1999)
different
cosmological
parameters
0.0,0.21
measuremen
Λ
ncertainty of the zCOSMOS survey,
3.5
1
−1
−1
Type
3+4
321
3.83 =
± 1.20
10.07
± 0.01
0.38
±60.12
8.12
±82.15
4.86
±10
1.99
1.57 res
intermediate
and
late-type
galaxies
and
the
fraction
of
H
I
gas
in
Ω
=
1.0
and
H
100
km
s
Mpc
)
from
the
current
conliminary
M
0
width
of 500 km
s 1 is ofused
cal luminosity
density
the to calcu3
early-type
is less than
per
In⇥ contrast,
z 0.7,
∼ 0.1 z ∼ 0.16 w
cordancegalaxies
ΛCDM cosmology
adopted
in this
paper
(ΩΛat =
All
⌦as
(0.42cent.
± 0.16)
10
H2
I=
−1
on of
typeHfrom
Lin et al.
100 H
ith
thegalaxy
stacked
I spectra.
ΩM = galaxies
0.3 and
s−1cent
Mpcto
correction
forthe
Survey (AU
early-type
contribute
25 per
Ω).HThe
reason for
0 = 70 km
I . The
of the
luminosity
func-calculated
2015
PHISCC
Workshop:
HI
Surveys
Get
March
17is
Note:
is the
number
galaxies
that
aremade
co-added
hM
isbe
average
H I i16-18,
the different
cosmological
parameters
is
using
the
correc0.13
solution
for each
galaxy
type were
higher
HNI gal
mass
density
ofofearly-type
galaxies
at
z Real
∼and
0.1
might
a2015zH∼ mass
Cosmic H I density (𝛀HI)
E. Zucca et al.: The zCOSMOS luminosity function
1223
B
Fig. 3. Luminosity functions of the diﬀerent spectrophotometric types in redshift bins (indicated in each panel): type 1 in red, type 2 in orange,
type 3+4 in blue, total sample in black. The squares represent the results from the C + and the solid lines are the results from the S T Y method. The
vertical dashed line represents the faint absolute limit considered in the S T Y estimate. The shaded regions represent the 68% uncertainties on the
parameters α and M ∗ .
dominate the faint end. Irregular galaxies increase their contribution at the lowest luminosities. At intermediate redshift
([0.35−0.75]), spiral galaxies increase their luminosities and
their contribution to the bright end of the luminosity function
is similar to that of the early types. At high redshift (z > 0.75),
irregular galaxies evolve strongly and the three morphological
types contribute almost equally to the total luminosity function.
Irregular galaxies show an evolution of a factor ∼3.3 in φ∗ from
low to high redshift. This evolution occurs mainly in the last redshift bin, while for z < 0.75, the contribution of these galaxies to
the global luminosity function is significantly lower than that of
spirals and early types.
Scarlata et al. (2007) derived luminosity functions for
COSMOS galaxies, using photometric redshifts and the ZEST
morphological classification. The shape parameters we find for
early types and spirals are consistent with those found by
Scarlata et al. (2007) for their early-type and disk galaxies, but
are significantly diﬀerent for irregular galaxies, which have a
much flatter slope in Scarlata et al. (2007). These diﬀerences are
likely due to the diﬀerent morphological classification applied.
Although the general trend in the luminosity functions of the
diﬀerent morphological types is similar to the results obtained
in the previous section for spectrophotometric types, some differences are present. In particular, there are more morphological
Cosmic H I gas evolution
•
The highest-redshift measurement of 𝛀HI ever made with the HI
spectral stacking technique.
•
All 21-cm measurements of 𝛀 H I are consistent.
• The
H I 21-cm emission measurements to date show no evidence
for significant evolution of H I gas abundance over the last 4 Gyr.
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
18
Summary
•
•
•
The global H I gas abundance does not change
significantly over the last 4 Gyr (z < 0.4)
This research successfully demonstrates the viability of
H I stacking technique at the intermediate redshift range
(z < 0.4) using current radio telescopes.
H I stacking is promising to give more scientific
outcomes to future HI deep surveys such as CHILES,
DINGO and LADUMA, pushing their limit still further.
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
19
Thank you for your attention !
2015 PHISCC Workshop: HI Surveys Get Real March 16-18, 2015
20