The Planck/ROSAT project


In a recent series of papers, we published radial profiles of gas density, temperature, entropy, and gas fraction obtained from combined X-ray (ROSAT) and Sunyaev-Zel’dovich (Planck) data, for a sample of 18 systems in the redshift range 0.04-0.2. This is the first relatively large sample of clusters for which high-quality constraints for all these quantities could be obtained over the entire cluster volume. This web page makes these profiles for each system available for download.

If you find this data useful, don’t forget to cite Eckert et al. 2012 (gas density), Eckert et al. 2013a (thermodynamic properties), Eckert et al. 2013b (gas mass fraction) and/or Planck Collaboration V 2013 (pressure).

Average profiles:

Entire sample

CC and NCC profiles

Individual clusters:

Abell 85

Abell 119

Abell 401

Abell 478

Abell 665

Abell 1651

Abell 1689

Abell 1795

Abell 2029

Abell 2163

Abell 2204

Abell 2218

Abell 2255

Abell 2256

Abell 3112

Abell 3158

Abell 3266

Abell 3558


  1. The parametric profiles were obtained by fitting a 7-parameter functional form to the projected emission-measure data. The errors were obtained through Monte-Carlo Markov Chain, with direct sampling of the posterior temperature, entropy, and gas fraction distributions. See Sect. 2 and 3 of Eckert et al. 2013a for details.

  1. The deprojected profiles were extracted through the Kriss et al. 1983 «onion peeling» technique (see Eckert et al. 2012). In addition, a correction for edge effects was applied following McLaughlin 1999, and a median smoothing regularization was performed, to alleviate the typical «roller-coaster» effect. The errors were propagated using a Monte Carlo over the data points.

  1. For the average profiles, both the profiles obtained through the median of the 18 individual systems and by combining the average Planck and ROSAT profiles are provided. In general, the former should be preferred, since the average Planck and ROSAT profiles were not extracted from the same systems, and thus are prone to selection effects.

  1. The error on the median was extracted using a randomized bootstrap technique (see Sect. 4 of Eckert et al. 2013a). Namely, we randomly select a set of 18 systems with repetition in our dataset, randomize the data points, and compute the median. This procedure is repeted 10,000 times and the errors are extracted from the resulting distribution. Note that this technique accounts for the intrinsic scatter of the data points.