[astro-ph/0510160] Detailed WMAP/X-ray comparison of 31 randomly selected nearby clusters of galaxies - incomplete Sunyaev-Zel'dovich silhouette

Authors:  Richard Lieu, Jonathan P.D. Mittaz, Shuang-Nan Zhang
Abstract:  The WMAP Q, V, and W band radial profiles of temperature deviation of the cosmic microwave background (CMB) were constructed for a sample of 31 randomly selected nearby clusters of galaxies in directions of Galactic latitude $|b| >$ 30$^o$. The profiles were compared in detail with the expected CMB Sunyaev-Zel'dovich effect (SZE) caused by these clusters, with the hot gas properties of each cluster obtained directly from X-ray observations, and with the WMAP point spread function fully taken into consideration. While the WMAP profiles of some clusters do exhibit the SZE, the phenomenon is also noted to be weak or absent from other clusters. A reliable overall assessment can be made using the combined (co-added) datasets of all 31 clusters, because (a) any remaining systematic uncertainties are low, and (b) the data are extremely clean (i.e. free from foreground contaminants). Both (a) and (b) are facts which we established by examining hundreds of random fields. The verdict from the 31 co-added cluster fields is that the observed SZE only accounts for about 1/4 of the expected decrement. The discrepancy represents too much extra flux for optically thin intracluster thermal emission to be the cause. Radio sources (discrete or halo) are also excluded because they have negative sloping spectra which are inconsistent with the ratio of the signals in different WMAP filters. A resolution of this discrepancy between predicted and observed decrements have potentially extreme ramifications for our interpretation of the CMB. One is forced to conclude that either the CMB is non-cosmological, or there are issues with the WMAP data itself which must be taken into account when interpreting the CMB emission.
[PDF]  [PS]  [BibTex]  [Bookmark]

Discussion related to specific recent arXiv papers
Post Reply
Alessandro Melchiorri
Posts: 128
Joined: September 24 2004
Affiliation: University of Rome

[astro-ph/0510160] Detailed WMAP/X-ray comparison of 31 rand

Post by Alessandro Melchiorri » November 22 2005

Any comments on this paper ?

Ben Gold
Posts: 81
Joined: September 25 2004
Affiliation: University of Minnesota

[astro-ph/0510160] Detailed WMAP/X-ray comparison of 31 rand

Post by Ben Gold » November 22 2005

I'm not an SZ expert at all, so I'd really have two questions for the authors:

1) Why can't it be that the cluster gas is clumpier than you expect? It seems this would naturally explain high X-ray emission and low SZ.

2) What about dust contamination of the WMAP data? The W-band in particular is known to contain a fair amount of foreground that proves frustratingly difficult to remove.

I also don't quite understand how they can get such nice-looking data points all the way down to angular scales of 0.1^\circ , since that's half the beamsize even in the W-band, AND the beam is known to be non-Gaussian.

Gil Holder
Posts: 27
Joined: September 25 2004
Affiliation: McGill University

[astro-ph/0510160] Detailed WMAP/X-ray comparison of 31 rand

Post by Gil Holder » November 22 2005

Some of the most problematic clusters according to this paper (Abell 1689, Abell 2218) are very well observed at high resolution (OVRO/BIMA, Suzie, Ryle) and show strong SZ signals. A1689 has a huge decrement at low frequencies (almost 2 mK according to Reese at al) with OVRO/BIMA (although the exact value depends on the X-ray profile) and has been used to place upper limits on peculiar velocities with SuZie. If there isn't an SZ decrement in A1689 there are a lot of people with some explaining to do. They refer to "observations of high-redshift clusters..." yet don't seem to realize there is overlap with their sample. That's a bit weird.

It wouldn't be crazy to get less SZ flux than expected if they are assuming an isothermal cluster. A lot of the flux is coming from the outer bits where the temperature could be much lower than the X-ray emission-weighted temperature. For the most part, I think the clusters here are unresolved, so they are really saying "the mean flux from 33 clusters is way too low." How low is a good question. Their Monte Carlo of CMB fluctuations seems to indicate that variations on the order of 0.03 mK are typical, so this would seem like a reasonable error bar to assign. So this is at least a 3 sigma conflict between isothermal beta model fits from ROSAT/ASCA X-ray data and unresolved SZ measurements with WMAP. Before rejecting the cosmological origin of the CMB, perhaps we should repeat this with Chandra/XMM and higher resolution measurements of this "cosmic microwave background."

A less "oh my god, everything in cosmology is crazy" spin on this would be to say that the ROSAT/ASCA/WMAP SZ-derived Hubble constant found by these authors is very, very far from the HST Key Project (a factor of about 10 off). I am a bit confused by their methods, but I would not be surprised if this turned out to be a systematics-prone way to measure the Hubble constant. After all, many people don't believe the SZ Hubble constants from Chandra/XMM data with resolved cluster measurements that have been carefully cleaned of all residual radio sources.

Niayesh Afshordi
Posts: 49
Joined: December 17 2004
Affiliation: Perimeter Institute/ University of Waterloo

[astro-ph/0510160] Detailed WMAP/X-ray comparison of 31 rand

Post by Niayesh Afshordi » December 01 2005

A point of caution is that the error bars plotted in this paper only show detector noise, and leave out the CMB primary/cosmic variance as a source of uncertainty. Therefore, the discrepancies, especially those coherent on the scale of a degree or larger, are exagerrated.

Of course, I agree with Gil that the system analysis prone to systematics, including the dropping temperature in the outskirts, and the failure of the beta model.

What has always amazed me is how people convince themselves that X-ray luminosities can be used for anything! We know that matching any smooth profile to the observed data gives biased amplitudes due to inhomogeneities that you don't resolve but make <n^2> larger than <n>^2. Is there any upper real limit on this bias?

Post Reply