TOWARD A COSMOLOGICAL HUBBLE DIAGRAM FOR TYPE II-P SUPERNOVAE
Peter Nugent,
1
Mark Sullivan,
2
Richard Ellis,
3
Avishay Gal-Yam,
3,4
Douglas C. Leonard,
3,5
D. Andrew Howell,
2
Pierre Astier,
6
Raymond G. Carlberg,
2
Alex Conley,
2
Sebastien Fabbro,
7
Dominique Fouchez,
8
James D. Neill,
9
Reynald Pain,
6
Kathy Perrett,
2
Chris J. Pritchet,
9
and Nicolas Regnault
6
Received 2005 April 12; accepted 2006 March 20
ABSTRACT
We present the first high-redshift Hubble diagram for Type II-P supernovae (SNe II-P) based on five events at
redshift up to z 0:3. This diagram was constructed using photometry from the Canada-France-Hawaii Telescope
Supernova Legacy Survey and absorption-line spectroscopy from the Keck Observatory. The method used to mea-
sure distances to these supernovae is based on recent work by Hamuy & Pinto and exploits a correlation between the
absolute brightness of SNe II-P and the expansion velocities deriv ed from the minimum of the Fe ii k5169 P Cygni
feature observed during the plateau phases. We present three refinements to this method that significantly improve the
practicality of measuring the distances of SNe II-P at cosmologically interesting redshifts. These are an extinction
correction measurement based on the VI colors at day 50, a cross-correlation measurement for the expansion ve-
locity, and the ability to extrapolate such velocities accurately over almost the entire plateau phase. We apply this
revised method to our data set of high-redshift SNe II-P and find that the resulting Hubble diagram has a scatter of
only 0.26 mag, thus demon strating the feasibility of measuring the expansion history, with present facilities, using a
method independent of that based on supernovae of Type Ia.
Subject headinggs: distance scale — supernovae: general
Online material: color figures
1. INTRODUCTION
The discovery of a cosmic acceleration based on the analysis
of the Hubble diagram of Type Ia supernovae (SNe Ia; Riess
et al. 1998; Perlmutter et al. 1999) has far-reaching implications
for our understanding of the universe. While indirect evidence
for the acceleration can be deduced from a combination of stud-
ies of the cosmic microwave background and large-scale structure
(Efstathiou et al. 2002; Bennett et al. 2003; Eisenstein et al. 2005),
distance measurements to SNe provide a valuable direct and
model independent tracer of the evolution of the expansion scale
factor necessary to constrain the nature of the proposed dark
energy. The mystery of dark energy lies at the crossroads of
astronomy and fundamental physics: the f ormer is tasked with
measuring its properties and the latter with explaining its origin.
Systematic uncertainties (rather than statistical errors) may
soon limit SN Ia measurements of the expansion rate at z 0:5
(see Knop et al. 2003; Astier et al. 2005, for recent analyses). A
largely unexplored source of potential bias is evolution in the
progenitor properties and/or the SN explosion. While several pro-
grams are underway to measure, test, and constrain SN Ia system-
atics (Sullivan et al. 2005; R. S. Ellis et al. 2006, in preparation), it
is highly desirable to consider independent tests of the cosmology
in which both the underlying physics and susceptibility to bias and
evolution are different.
As cosmological probes, SNe II have lagged behind their
brighter and better calibrated cousins, SNe Ia, but their potential
has improved significantly as a result of several recent studies.
Baron et al. (2003, 2004), Mitchell et al. (2002), and two doctoral
theses (Hamuy 2002; Leonard 2000) have used new samples
of SNe II and demonstrated that a subset, the plateau SNe II-P,
are particularly promising as distance indicators. From an astro-
physical standpoint, SNe II-P hold three advantages over SNe Ia
as cosmological probes: (1) their progenitor stars are well un-
derstood ( Heger et al. 2003; Li et al. 2005; Smartt et al. 2003);
(2) the physics of their atmospheres, dominated by hydrogen,
is much simpler to understand and model (Baron et al. 2004);
and (3) while fainter, they are more abundant per unit volume
(Mannucci et al. 2005; Cappellaro et al. 2005). The two main
disadvantages are that they are on average 1.2 mag fainter in the
optical than SNe Ia and that all distance measurements currently
based on SNe II-P require a reasonable quality spectrum of the
event.
Unlike the other members of the core-collapse supernova fam-
ily, SNe II-P maintain a massive hydrogen envelope prior to ex-
plosion. From analyses of their opti cal light curves and spectra
(e.g., Chugai 1994), they evidently suffer little subsequent inter-
action with the surrounding medium—they are the result of the
putative red supergiant exploding into a near vacuum. Recent re-
sults from spectropolarimetric studies also sugges t that, at least
during the plateau epoch, the ejecta and electron scattering pho-
tosphere are quite spherical (see Leonard & Filippenko 2005 and
references therein).
For SNe II-P, distance measurements from the spectral expand-
ing atmosphere method (SEAM; see Baron et al. 2004), the de-
scendent of the traditional expanding photosphere method (EPM;
A
1
Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA
94720.
2
University of Toronto, 60 St. George Street, Toronto, ON M5S 3H8,
Canada.
3
California Institute of Technology, 1200 East California Boulevard, Pasa-
dena CA 91125.
4
Hubble Postdoctoral Fellow.
5
NSF Astronomy and Astrophysics Postdoctoral Fellow.
6
Laboratoire de Physique Nucleaire et de Haute Energies de Paris ( LPNHE),
CNRS-IN2P3; and University of Paris VI and VII, 75005 Paris, France.
7
CENTRA, Centro Multidisciplinar de Astrofı´sica, Instituto Superior Te
´
c-
nico, Avenida Rovisco Pais, 1049 Lisbon, Portugal.
8
Centre de Physique des Particules de Marseille (CPPM ), CNRS-IN2P3 and
University Aix Marseille II, Case 907, 13288 Marseille Cedex 9, France.
9
Department of Physics and Astronomy, University of Victoria, P.O. Box
3055, Victoria, BC V8W 3P6, Canada.
841
The Astrophysical Journal, 645:841–850, 2006 July 10
# 2006. The American Astronomical Society. All rights reserved. Printed in U.S.A.
see Kirshner & Kwan 1974; Schmidt 1993), can be replaced by a
more practical empirical method that requires less input data, as
developed in Hamuy & Pinto (2002) and Hamuy (2003) (here-
after HP02 and H03, respectively). The approach advocated by
HP02 is a particularly significant development, since it is moti-
vated by sound physical principles. In more luminous supernovae,
the hydrogen recombination front is maintained at higher ve-
locities, pushing the photosphere farther out in radius. When a
SN II-P is on the plateau phase, a period that lasts f or around
100 days, a strong correlation is expected and observed between
the velocity of the weak Fe ii lines near 5000 8 (which suitably
track the electron scattering photosphere) and the luminosity.
Calibration of these standardized candles requires line-of-
sight extinction corrections, which can be determined in one of
two ways. When the supernova leaves the plateau phase, the
photospheric temperature should be close to that of recombina-
tion, and thus the observed color at this point will provide a red-
dening estimate. Alternatively, for well-studied SNe II-P, one can
employ detailed modeling of a series of high signal-to-noise ratio
(S/ N) spectra to determine the extinction. Here, deviations be-
tween the line profiles of selected absorption features and those
expected given the observed spectral energy distribution (SED)
can be used to estimate selective extinction. Application of these
techniques to 24 SNe II-P in the Hubble flow (Fig. 1 of H03)
yields an I-band Hubble diagram with a scatter of only 0.29 mag,
corresponding to a precision of 15% in distance.
The present paper is concerned with extending the HP02
method so that it can be used at cosmological distances. Although
economical compared to the SEAM /EPM methods in terms of
input data, the HP02 method is still poorly suited as a basis for
verifying the cosmic acceleration due to the difficult demands on
spectroscopy and the photometric measurements necessary for
an extinction correction at high redshift. Our discussion is timely
because of the feasibility of locating distant SNe II-P from com-
prehensive ‘‘rolling searches’’ such as the Canada-France-Hawaii
Telescope SN Legacy Survey (SNLS; Astier et al. 2005). Such
surveys can generate SNe II-P with 0:1 < z < 0:4 continuously
with tight constraints on their explosion dates and excellent pho-
tometric coverage, making prescheduled spectroscopic campaigns
to measure the expansion rate during the plateau phase a prac-
tical proposition.
A plan of the paper follows. In x 2 we introduce an enhanced
local data set suitable for exploring the potential of an improved
method and detail the difficulties we need to overcome to use the
HP02 method at intermediate redshifts. We then introduce tw o
extensions to the method and show, via our data set, that we can
retain the precision with indicators of velocity and extinction
more suited to high-redshift data sets. In x 3 we present the first
observations of high-redshift SNe II-P obtained via SNLS and
construct a Hubble diagram from these supernovae that dem-
onstrate the feasibility of locating and studying high-redshift
SNe II-P for verifying the cosmic acceleration. We summarize
our conclusions in x 4.
2. IMPROVING THE HP02 METHOD
The drawbacks of the method introduced by HP02 for studies
of cosmologically distant SNe II-P are twofold.
Foremost, the extinction correction cannot be measured us-
ing colors determined at the end of the plateau phase, since its
precise timing would require continuous monitoring, which is
impractical for faint supernovae, and at intermediate redshift,
SNe II-P discovered prior to the time they explode will typically
be too close to the Sun at this stage in their evolution. Moreover,
corrections for extinction based on this method in H03 often pro-
duced negative results. Likewise, the S/ N of plateau spectra will
generally not be adequate for detailed spectrum synthesis mod-
eling of various line profiles in order to determine the intrinsic
SED. We thus seek a more appropriate way to esti mate the line-
of-sight extinction.
Second, it is a challenge at moderate redshift to secure ac-
curate measures of the w eak diagnostic Fe ii k5169 line used to
measure the photospheric expansion during the plateau phase,
even with the largest ground-based telescopes. In particular, as
diagnostic lines in this region of the spectrum are redshifted into
the OH forest beyond z ’ 0:5, it becomes advantageous to ex-
plore alternatives should the Fe ii k5169 line profile be polluted by
a sky feature. We therefore wish to explore the practicality of al-
ternative, stronger lines to measure the photospheric expansion.
To explore the possibilities, we have constructed a sample of
nearby SNe II-P that includes all of those in H03 for which there
is both V- and I-band data. We also include SN 2004dh, a SN
discovered by the Lick Observatory Supernovae Search ( Moore
& Li 2004), and follow ed as part of the Caltech Core-Collapse
Program (CCCP; see Gal-Yam et al. 2004, 2006a [in prepara-
tion]), which leads to a total of 19 SNe II-P. We have updated the
distances to two supernovae presented in H03. In the case of SN
1999em we use the Hubble Space Telescope (HST )–determined
Cepheid distance to the host ( Leonard et al. 2003). For SN 1999gi
we use the improved distance derived via a variety of methods as
discussed in Leonard et al. (2002). These changes have very little
effect on the overall fit, as both galaxies are nearby with large un-
certainties in their relative distances. We have also removed all of
the extinction corrections applied in H03.
We now turn to two improvements necessary to enable us to
use SNe II-P as distance indicators at moderate redshift.
2.1. Incorporatin
g VI Colors
The first modification of the HP02 method we have explored
makes use of the rest-frame VI color during the plateau phase at
day 50 to perform an extinction correction rather than relying on
colors at the end of the plateau phase or on detailed modeling, as
in H02. Our goal is to explore empirically the extent to which we
can retain a tight Fe ii velocity-luminosity relation using the
observed VI color at this epoch.
If we fit for the velocity and luminosity using plateau-phase
data, interpolated to day 50, we find
M
I
¼ log
10
(V
Fe ii
=5000) 1:36½(V I ) (V I )
0
þM
I
0
:
ð1Þ
Here, as in H03, we have adopted the relative surface bright-
ness fluctuation distance scale (Tonry et al. 2000). The resulting
fit to the data yields ¼ 6:69 0:50 and M
I
0
¼17:49
0:08 (H
0
¼ 70 km s
1
Mpc
1
)fora(VI )
0
¼ 0:53. In this fit we
have employed the standard relationship between the VI col-
ors for a dust law with R
V
¼ 3:1 for a SN II-P at day 50 (A
I
¼
1:36E
V I
). As is do ne in the SN Ia studies (Guy et al. 2 005) for
the color-stretch relationship, we have adopted a ridgeline, un-
extingui sh e d (VI )
0
color for SN II-P of 0.53 mag. Under the
assumption that the extinction laws are similar from galaxy
to galaxy, the exact choice is irrelevant, since this term and the
M
I
0
term are degenerate for the purpose of using these objects
as cosmological probes. Using this technique we can produce
a Hubble diagram in rest-frame I band with a scatter of only
0.28 mag (Fig. 1) for those SNe II-P in the Hubble flow (cz >
3000 km s
1
), similar to that found in H03. The scatter for all
SN II-P is reduced from 1.11 to 0.52 mag. Crucially, we find
NUGENT ET AL.842 Vol. 645
there is no advantage in using reddening estimators based on late-
time color measures or on detailed modeling of the spectroscopic
data.
To address the robustness of this simple approach, we in-
vestigated solutions in which we permit an additional parameter
based on a possible relationship between the Fe ii velocity and
the dereddened VI color. One could easily envision a corre-
lation that brighter SN II-P were faster and bluer. However, the
correlation we find is weak and lacks any statistical significance.
We note that for 6¼ 5:0( ¼ 5 implies that the luminosity
follows the square of the radius only), there presumably does exist
an additional correlation between the effective temperature and
density of the expanding photosphere and its radius, so conceiv-
ably there is some scope for further improvement in defining
equation (1). Future work, such as the ongoing research by the
CCCP and the Carnegie Supernova Project (CSP; Freedman
2005), will be helpful in clarifying the possibilities and allow us
to address the strongly correlated issues of extinction corrections
and velocity-dependent colors.
2.2. Usin
g Alternative Lines to Diagnose
the Photospheric Expansion
We now explore whether alternative absorption lines can be
used in addition to Fe ii k 5169 as diagnostics of the photos pheric
expansion velocity. Over the wavelength range 4500–5500 8
there exist several Fe ii lines (the strongest of which are at 4924,
5018, and 5169 8). Moreover, often the H k4861 absorption
line is prominent. Depending on the supernova and its phase, the
relative strengths of these lines can vary considerably (see Fig. 2).
In general, at earlier epochs H tends to dominate, while at later
epochs the Fe ii lines are the dominant absorption features over
this wavelength range. Accordingly, if at higher redshift we are
able to compare to any or all of these lines, we can minimize the
effects of the interference with OH night-sky features and any
underlying host galaxy contamination.
In order to explore possible systematic effects arising from
using different absorption lines as measures of the expansion, we
first examine trends in local data. To do this we assembled a
library of SNe II-P spectra based on data from the SUSPECT
database
10
for the following: SNe 1969L, 1988A, 1988H, 1993W,
1999gi, and 1999em (Benetti 1991; Turatto et al. 1993; Leonard
et al. 2002, 2003; Hamuy et al. 2001; Baron et al. 2003) as well as
data from the first year of the CCCP: SNe 2004A, 2004T, 2004dh,
2004du, 2004em, and 2004et (A. Gal-Yam et al. 2006b, in prep-
aration). A particularly important question is whether it is safe to
use H as a diagnostic rather than the Fe ii lines, given the former
is often quite strong. In what follows, we have divided the local
data set into two categories: those dominated by the H feature
and those dominated by the Fe ii features (defined according to
whether the equivalent width of H is respectively greater than or
less than that of the sum of the Fe ii features). The motivation here
Fig. 1.— Revised Hubble diagram (scaled to H
0
¼ 70 km s
1
Mpc
1
)for
local SNe II-P (circles and square) using an improved estimator for extinction
derived from VI colors an d Fe ii velocities observed during th e plateau phase
at day 50. Inset: Residual Hubble diagram for both the revi sed method ( filled
circles) and the data under the assumpti on that they are pure standard candles
in I band, corrected on ly for MW extinction (open circles). The scatter for
those SN II-P in the Hubble flow is reduced from 0.59 to 0.28 mag, similar to
the 0.29 mag original ly achieved in H03 and based on observables substan-
tially better suited to high-redshift ob servations. The scatter for all SN II-P is
reduced from 1.11 to 0.52 mag. [See the electronic edition of the Journal for a
color version of this figure.]
Fig. 2.— Comparison of two Fe ii–dominate d spectra to an H-dominated
spectrum. This demarcation is defined by the equivalent width of H being
greater than the sum of the Fe ii features. Depending on the SN and its phase, any
one of these features could potentially be the strongest and/or the most easily
measured (due to sky lines or host galaxy contamination) and thus could be used
at high redshift.
10
See http:// bruford.nhn.ou.edu/~ suspect /index1.html.
Fig. 3.— Ratio of the velocities of H to Fe ii k5169 vs. the H velocity in the
H-dominated spectra. We fit this ratio in two pieces: a constant for velocities
below 6000 km s
1
(ratio ¼ 1:395) and linearly for higher velocities (ratio ¼
1:395
6:489ðÞ; 10
5
V (H ) 6000½). The linear portion is the only one re l-
evant to our analysis, as our high-redshift data lie within these velocities. The
dispersion about this relationship (0.054) translates to an uncertainty in mea-
suring the Fe ii velocity from H of 300 km s
1
. See text for a full discussion
and explanation of this trend.
COSMOLOGY WIT H SNe II-P 843No. 2, 2006
is twofold. First, unlike H, in the case of the Fe ii-dominated
spectra, the weak lines form at the same location in the atmosphere
near the electron scattering photosphere. Thus we might expect
any (or all) of them to be acceptable for measuring the velocity.
Second, as we later test a novel cross-correlation method across
this wavelength range, it is very helpful to understand possible
systematic velocity differences between H and the Fe ii lines. For
both Fe ii k5169 and H, we measured absorption velocities using
a routine in which we first subtract the continuum, take the wave-
length derivative and then fitted for the wavelength where this
derivative changes sign.
Figure 3 shows that, in the H-dominated spectra, the velocity
of H is significantly higher than that of Fe ii k5169 with the
trend that the ratio of H to Fe ii drops toward unit y only at very
high velocities. Clearly, simply replacing Fe ii absorption veloc-
ities with measures of H would lead to erroneous results. How-
ever, if we fit this ratio in two sections, a constant for velocities
below 6000 km s
1
(ratio ¼ 1:395) and a linear decline for higher
velocities (ratio ¼ 1:395
6:489ðÞ; 10
5
V (H ) 6000½), the
Fig. 4.— Histogram of the difference in measured velocity vs. the velocity
derived from the cross-correlation technique for Fe ii k5169 and H.Herewe
have compared each SN II-P in our library to all the others, splittin g the com-
parison library for SN II-P to those that ar e dominated by H and those by Fe ii.
The measured dispersion is 161 km s
1
for the H subset and 108 km s
1
for the
Fe ii subset. [See th e electronic e dition of the Journal for a color version of this
figure.]
Fig. 5.—Left: Plot of the velocity of Fe ii k5169 vs. time for the published SNe II-P (each symbol represents an individual SN ). Right: Plot of the individual
SNe II-P with their velocities normalized to day 50, along with the best fit for the evo lution of the decline covering the epochs fo r our hig h-redshift SNe (day 9–75).
The fit is a power law of the form V (50) ¼ V (t)(t /50)
0:4640:017
.[See the electronic edition of the Journal for a color version of this figure.]
Fig. 6.—SNLS g
0
r
0
i
0
z
0
light curves (in f
k
)forSNLS-03D3ceatz ¼ 0:2881.
Unfortunately, this SN was not observed in z
0
on the plateau. At this redshift z
0
overlaps nicely with the rest-frame I b and; thus, the VI color had to be ex-
trapolated from bluer colors than desired for this SN II-P. [See the electronic
edition of the Journal for a color version of th is figure.]
NUGENT ET AL.844 Vol. 645
dispersion about this relationship (0.054) translates to an
uncertainty in estimating the Fe ii velocity from H measures of
only 300 km s
1
.
Of course, before accepting such an empirical correction, it is
important to understand the trend physically. To accomplish this,
we examined spectrum synthesis fits to model SNe II-P discussed
by Baron et al. (2004). The predicted trend is qualitatively similar
in form to that observed and, in particular, reproduces the decline
in the ratio observed at higher velocities. The model spectra with
higher effective temperatures were in general from earlier epochs
with higher velocities, while the cooler ones were later with lower
velocities. The optical depth of H is strongly tied to the level of
ionization of hydrogen. Thus as the temperature drops, the veloc-
ities fall, less hydrogen is ionized, and the optical depth of H
increases significantly compared to the weak Fe ii lines. This in
turn increases the ratio.
In order to maximize the information content in either the
series of weak Fe ii lines or in the H-dominated spectra, we then
examined a cross-correlation analysis across the wavelength
range 4500–5500 8. In the Fe ii-dominated spectra we simply
compared the measured Fe ii k5169 velocities to those obtained
in the cross-correlation analysis. For the H-dominated spectra
we compare the measured H veloci ties and then reduced these
to equivalent Fe ii k5169 velocities using the fit to the ratio de-
fined above. In performing cross-correlations, all spectra are con-
tinuum subtracted and analyzed in log wavelength space in the
manner described by Tonry & Davis (1979).
A measure of the potential bias and systematic uncertainty
arising in the cross-correlation method can be obtained by con-
sidering the distribution of velocity differences obtained by
multiple template comparisons within our local data set (Fig. 4).
Each local supernova, when compared to the average of the cross-
correlation calculated velocities, shows differences of typically
<150 km s
1
. This is a measure of template mismatch for the local
sample in the wavelength region sample: i.e., in individual SNe II-P,
the strongest features present can form at slightly different ve-
locities than Fe ii k5169, in addition to measurement uncertainties
for the minimum of that line in our spectral library. This bias is
recorded for each of our template spectra and subtracted off when
used in the cross-correlation measurement. The resulting measured
dispersion in comparing one of our template spectra to the rest in
their respective sets is 160 km s
1
for H-dominated spectra and
110 km s
1
for the Fe ii–dominated spectra.
Of course, these tests calibrate systematic uncertainties that
are dominant only for low-redshift SNe IIP. At high redshift we
can expect an additional component from the lower S/ N implicit
in the fainter sources. To estimate how effective this method will
be at intermed iate redshift, where our S/N is typically between
8 and 20 per 2 8 over the relevant wavelength range, we per-
formed a Monte Carlo equivalent of the above test by degrading
the S/N of the nearby spectra to 7 per 2 8. The resulting disper-
sion for individual measurements increased by only 15 km s
1
,
Fig. 7.—SNLS g
0
r
0
i
0
z
0
light curves (in f
k
)forSNLS-03D4cwatz ¼ 0:1543.
[See the electronic edition of the Journal for a color version of this figure.]
Fig. 8.—SNLS g
0
r
0
i
0
z
0
light curves (in f
k
) for SNLS-04D1ln at z ¼ 0:2078.
[See the electronic edition of the Journal for a color version of this figure.]
Fig. 9.—SNLS g
0
r
0
i
0
z
0
light curves (in f
k
) for SNLS-04D1pj at z ¼ 0:1556.
[See the electronic edition of the Journal for a color version of this figure.]
Fig. 10.—SNLS g
0
r
0
i
0
z
0
light curves (in f
k
)forSNLS-04D4fuatz ¼ 0:1330.
This SN was caught just after shock breakout as the g
0
r
0
i
0
z
0
colors (uncorrected for
extinction) lead to a temperature >20,000 K. [See the electronic edition of the
Journal for a color version of this figure.]
COSMOLOGY WIT H SNe II-P 845No. 2, 2006
![Fig. 12.—Left: Plot of the best-fit template spectrum (thick lines) and the SNLS SNe II-P (thin lines) used in the cross-correlation technique. Right: Resulting correlation. Top to bottom: SNLS-03D3ce, 03D4cw, 04D1ln, 04D1pj, and 04D4fu. [See the electronic edition of the Journal for a color version of this figure.]](/figures/fig-12-left-plot-of-the-best-fit-template-spectrum-thick-17cci401.webp)
![Fig. 1.—Revised Hubble diagram (scaled to H0 ¼ 70 km s 1 Mpc 1) for local SNe II-P (circles and square) using an improved estimator for extinction derived from V I colors and Fe ii velocities observed during the plateau phase at day 50. Inset: Residual Hubble diagram for both the revised method ( filled circles) and the data under the assumption that they are pure standard candles in I band, corrected only for MW extinction (open circles). The scatter for those SN II-P in the Hubble flow is reduced from 0.59 to 0.28 mag, similar to the 0.29 mag originally achieved in H03 and based on observables substantially better suited to high-redshift observations. The scatter for all SN II-P is reduced from 1.11 to 0.52 mag. [See the electronic edition of the Journal for a color version of this figure.]](/figures/fig-1-revised-hubble-diagram-scaled-to-h0-1-4-70-km-s-1-mpc-1jrrsc3v.webp)
