On the suspected timing-offset-induced calibration error in the Wilkinson microwave anisotropy probe time-ordered data

Abstract

In the time-ordered data (TOD) files of the WMAP CMB observations, there is an undocumented timing offset of -25.6 ms between the spacecraft attitude and radio flux density timestamps. If the offset induced an error during calibration of the raw TOD, then estimates of the WMAP CMB quadrupole might be substantially in error. A timing error during calibration would not only induce an artificial quadrupole-like signal in the mean sky map, it would also add variance per pixel. This variance would be present in the calibrated TOD. Low-resolution map-making as a function of timing offset should show a minimum variance for the correct timing offset. Three years of the calibrated, filtered WMAP 3-year TOD are compiled into sky maps at HEALPix resolution N_side=8, individually for each of the K, Ka, Q, V and W band differencing assemblies (DA's), as a function of timing offset. The median per map of the temperature fluctuation variance per pixel is calculated and minimised against timing offset. Minima are clearly present. The timing offsets that minimise the median variance are -38 \pm 8 ms (K, Ka), -27 \pm 3 ms (Q), -43 \pm 8 ms (V), and -47 \pm 194 ms (W), i.e. an average of -30 \pm 3 ms, where the WMAP collaboration's preferred offset is 0 \pm 1.7 ms. A non-parametric bootstrap analysis rejects the latter at a significance of 99.999%. The hypothesis of a -25.6 ms offset, suggested by Liu, Xiong & Li from the TOD file timing offset, is consistent with these minima. It is difficult to avoid the conclusion that the WMAP calibrated TOD and inferred maps are wrongly calibrated. CMB quadrupole estimates (3/pi)C_2 based on the incorrectly calibrated TOD are overestimated by roughly 64 \pm 6% (KQ85 mask) to 94 \pm 10% (KQ75 mask). Ideally, the WMAP map-making pipelines should be redone starting from the uncalibrated TOD and using the -25.6 ms timing offset correction.