A curious reality check onto SNIa-based distances

A new A&A paper, arXiv:1203.7132, "The distance to NGC 1316 (Fornax
A): yet another curious case" uses "surface brightness fluctuations"
(this is shear, i.e.cosmological twinkling) to obtain the distance to
Fornax A which turns out to be 17% greater than derived from 4 SNIae
in that galaxy. The paper is state-of-the-art, they derive the
distance twice using two independent methods -- a complete review of
the state of play is given. It seems there is still a lot of
uncertainty in the whole distance modulus schtick.

They drill into the 17% difference a bit, but only incrementally and
not very satisfactorilly -- slicing bits off because some earlier
author said they could. I'd instead look for a single factor causing
most of the offset.

So if we take the new distance to Fornax A as correct, then SNIa
luminosity was 37% brighter for its 4 SNIae. As that is unlikely
(because the luminosity is constrained by the model parameters like
rise time and stretch), some other assumption needs adjusting.
Possibly this impacts the "accelerating expansion" model?

Anyway, I thought the 17% variance from the SNIa-derived distance to
Fornax A is interesting enough to have called it to the attention of
this group, should any have some thoughts on it. cheers all.

In article , Eric Flesch
writes:

A new A&A paper, arXiv:1203.7132, "The distance to NGC 1316 (Fornax
A): yet another curious case" uses "surface brightness fluctuations"
(this is shear, i.e.cosmological twinkling) to obtain the distance to
Fornax A which turns out to be 17% greater than derived from 4 SNIae
in that galaxy. The paper is state-of-the-art, they derive the
distance twice using two independent methods -- a complete review of
the state of play is given. It seems there is still a lot of
uncertainty in the whole distance modulus schtick.

Even if individual distances have a 17% error, that doesn't mean that
all conclusions based on such distances have the same size error.

So if we take the new distance to Fornax A as correct, then SNIa
luminosity was 37% brighter for its 4 SNIae. As that is unlikely
(because the luminosity is constrained by the model parameters like
rise time and stretch), some other assumption needs adjusting.
Possibly this impacts the "accelerating expansion" model?

On the face of it, if the difference is large enough to have such an
effect, then I would doubt the scintillation distance, since one can
forget the SNIa stuff and still other data point to accelerated
expansion.

Preliminary Planck results will be announced in a couple of weeks.

In article , Eric Flesch
writes:

A new A&A paper, arXiv:1203.7132,

http://arxiv.org/abs/1203.7132 says:

There is no record of an article with identifier '1203.7132'. You might
instead try to search for papers.

Correct is: http://arxiv.org/abs/1302.7132

"The distance to NGC 1316 (Fornax
A): yet another curious case" uses "surface brightness fluctuations"
(this is shear, i.e.cosmological twinkling)

Are you sure it has something to do with shear? And is the twinkling
cosmological? Is it even twinkling?

Normally, surface-brightness fluctuations in the context of determining
a distance to a galaxy refers to the variation in brightness between
pixels in a CCD (whereas "twinkling" is usually change in brightness
with time). You can read more he

http://ned.ipac.caltech.edu/level5/J...Jacoby9_1.html

Note that at low redshift, all else being equal, surface brightness is
independent of distance: stars are fainter but the area is smaller and
the two effects exactly cancel. The point is that the SCATTER in
fluctuations DOES depend on distance. (The brightness is related to the
luminosity distance and the size to the angular-size distance. Since
these are not the same at high redshift, surface brightness actually
decreases with the 4th power of the redshift, whatever the cosmological
parameters. Surface brightness in a fixed band then goes down with the
5th power of the redshift and signal-to-noise decreases with the 10th
power of the redshift, so it is clear why it is difficult to image
extended sources at high redshift. However, this is relevant only at
high redshift.)

On Thu, 07 Mar 13, Phillip Helbig wrote:
Correct is: http://arxiv.org/abs/1302.7132

Oops, sorry about that.

I wrote: (this is shear, i.e.cosmological twinkling)
Normally, surface-brightness fluctuations in the context of determining
a distance to a galaxy refers to the variation in brightness between
pixels in a CCD...

Yoicks, thanks for the kind tutorial & reference on SBF, Phil. You're
right, I was misunderstanding what was being measured.

The point is that the SCATTER in fluctuations DOES depend on distance.

Like they are resolving inhomogeneities as opposed to stars per se.
This presupposes that a galaxy twice the size of another would still
be the same internally per, say, kpc^3. If it were not so, there
could be degeneracy in having to estimate the size of the galaxy in
order to deciphre & interpret the SBF.

Thanks again,
Eric

In article ,
Phillip Helbig---undress to reply writes:
http://arxiv.org/abs/1203.7132 says:

Cantiello et al., accepted for publication in A&A.

Normally, surface-brightness fluctuations in the context of determining
a distance to a galaxy refers to the variation in brightness between
pixels in a CCD (whereas "twinkling" is usually change in brightness
with time).

This is correct. In a nearby galaxy, one pixel will contain on
average only a few stars, and random fluctuations in the actual
number in different pixels will be large. In a distant galaxy, each
pixel will contain many stars, and random fluctuations will be
relatively much smaller. Thus the observed size of the fluctuations
gives a measure of distance. I think it was John Tonry who first
used this technique to measure distances.

I had a quick glance at the preprint and notice the following:
1. 21 Mpc is pretty far away for a SBF distance to work. Indeed the
quoted systematic uncertainty is 0.14 mag.

2. NGC 1316 (=Fornax A) has lots of patchy dust. If not corrected
for, this will make the distance too small. (Dust extinction
fluctuations will be mistaken for variations in star numbers.) If
over-corrected, the derived distance will be too big. The authors
observed a wide range of wavelengths, which should help but isn't a
guarantee.

3. The infrared SBF distances, which ought to depend least on dust,
give a distance modulus about 0.2 mag smaller than the average, in
better agreement with the SN distance.

4. The SBF calibration depends on the stellar population, and as a
radio galaxy probably undergoing a collision, NGC 1316 might have a
peculiar stellar population. (My first thought, though, is that a
peculiar population would likely make the SBF distance come out too
small, but maybe I'm missing something.)

5. Only two of the SNe have modern measurements, and one of the two
is an unusual fast-declining type that the SN authors (Stritzinger et
al. 2010) consider unsuitable for a distance derivation and omit. So
the SN distance is mostly based on a single SN, which was located in
the dusty inner part of the galaxy.

6. The SN distances themselves disagree, depending on how the
analysis is done.

Presumably the authors take all the above into account in their
uncertainty estimates, but it's too soon to get excited. At most,
the distance discrepancy is about 2.3 sigma. That's enough to
justify more work but not enough to say all of cosmology is wrong.

--
Help keep our newsgroup healthy; please don't feed the trolls.
Steve Willner Phone 617-495-7123
Cambridge, MA 02138 USA

In article , Eric Flesch
writes:

The point is that the SCATTER in fluctuations DOES depend on distance.

Like they are resolving inhomogeneities as opposed to stars per se.
This presupposes that a galaxy twice the size of another would still
be the same internally per, say, kpc^3. If it were not so, there
could be degeneracy in having to estimate the size of the galaxy in
order to deciphre & interpret the SBF.

Yes. Basically, it's just Poisson noise. If a pixel has 10,000 stars
in it, then the scatter from pixel to pixel (the fluctuations) is 100
stars, or 1% in relative terms. But a galaxy 10 times closer will have
100 times less area per pixel and thus only 100 stars, for a variation
of 10, or 10% in relative terms. Thus, the lower the relative
difference in brightness from pixel to pixel, the farther away the
galaxy.

Of course, this assumes that the galaxies are comparable, but I think
that is OK since all such distances are at cosmologically low redshift.
As to the size effect, keep in mind that a pixel contains many stars but
is still a small part of the whole galaxy. It's an interesting idea,
but not completely straightforward.

On 3/7/2013 12:52 PM, Phillip Helbig---undress to reply wrote:

.. Surface brightness in a fixed band then goes down with the
5th power of the redshift and signal-to-noise decreases with the 10th
power of the redshift,

Doesn't the first part of this statement mean that the
energy received by one CCD pixel goes down with the
5th power of the redshift? And if so, why isn't that
also the decrease in SNR? (And if not, how do the CCD
pixels know that they need to make more noise?!)

--
Jos

In article ,
Phillip Helbig---undress to reply writes:
[referring to surface brightness fluctuation distances]
Of course, this assumes that the galaxies are comparable, but I think
that is OK since all such distances are at cosmologically low redshift.

I mentioned population differences as one thing that can throw off
SBF distances. The check on that is to derive distances at different
wavelengths and see whether they are consistent. (They aren't in the
galaxy that started this thread.)

Differences in stellar densities, on the other hand, cancel out.
Essentially there are two parameters to be determined: the stellar
density in the galaxy (stars per cubic parsec) averaged over each
line of sight and the distance. There are also two measured
parameters: the surface brightness itself and the fluctuations from
pixel to pixel. These combine to determine both distance and stellar
density. The problem is that signal to noise becomes insufficient at
distances not far beyond the Virgo cluster.

--
Help keep our newsgroup healthy; please don't feed the trolls.
Steve Willner Phone 617-495-7123
Cambridge, MA 02138 USA