On the suspected timing error in Wilkinson microwave anisotropy probe map-making

Toruń Centre for Astronomy, Nicolaus Copernicus University, ul. Gagarina 11, 87-100 Toruń, Poland

Received: 26 April 2010
Accepted: 27 May 2010

Abstract

Context. It has recently been suggested that the compilation of the calibrated time-ordered-data (TOD) of the Wilkinson Microwave Anisotropy Probe (WMAP) observations of the cosmic microwave background (CMB) into full-year or multi-year maps may have been carried out with a small timing interpolation error. A large fraction of the previously estimated WMAP CMB quadrupole signal would be an artefact of incorrect Doppler dipole subtraction if this hypothesis were correct.

Aims. Since observations of bright foreground objects constitute part of the TOD, these can be used to test the hypothesis.

Methods. Scans of an object in different directions should be shifted by the would-be timing error, causing a blurring effect. For each of several different timing offsets, three half-years of the calibrated, filtered WMAP TOD are compiled individually for the four W band differencing assemblies (DA's), with no masking of bright objects, giving 12 maps for each timing offset. Percentiles of the temperature-fluctuation distribution in each map at HEALPix resolution N_side = 2048 are used to determine the dependence of all-sky image sharpness on the timing offset. The Q and V bands are also considered.

Results. In the W band, which is the band with the shortest exposure times, the 99.999% percentile, i.e. the temperature fluctuation in the ≈503-rd brightest pixel, is the least noisy percentile as a function of timing offset. Using this statistic, the hypothesis that a -25.6 ms offset relative to the timing adopted by the WMAP collaboration gives a focus at least as sharp as the uncorrected timing is rejected at 4.6σ significance, assuming Gaussian errors and statistical independence between the maps of the 12 DA/observing period combinations. The Q and V band maps also reject the -25.6 ms offset hypothesis at high statistical significance.

Conclusions. The requirement that the correct choice of timing offset must maximise image sharpness implies that the hypothesis of a timing error in the WMAP collaboration's compilation of the WMAP calibrated, filtered TOD is rejected at high statistical significance in each of the Q, V and W wavebands. However, the hypothesis that a timing error was applied during calibration of the raw TOD, leading to a dipole-induced difference signal, is not excluded by this method.