This paper describes the processing applied to the HFI cleaned time-ordered data to produce photometrically calibrated maps. HFI observes the sky over a broad range of frequencies, from 100 to 857 GHz. To get the best accuracy on the calibration on such a large range, two different photometric calibration schemes have to be used. The 545 and 857 GHz data are calibrated using Uranus and Neptune flux density measurements, compared with models of their atmospheric emissions to calibrate the data. The lower frequencies (below 353 GHz) are calibrated using the cosmological microwave background dipole.One of the components of this anisotropy results from the orbital motion of the satellite in the Solar System, and is therefore time-variable. Photometric calibration is thus tightly linked to mapmaking, which also addresses low frequency noise removal. The 2013 released HFI data show some evidence for apparent gain variations of the HFI bolometers detection chain. These variations were identified by comparing observations taken more than one year apart in the same configuration. We developed an effective correction to limit its effect on calibration, and assess its accuracy. We present several methods used to estimate the precision of the photometric calibration. We distinguish relative (from one detector to another, or from one frequency to another) and absolute uncertainties. In both cases, we found that these uncertainties range from a few $10^{-3}$ to several per cents from 100 to 857 GHz. We describe the pipeline producing the maps from the HFI timelines, based on the photometric calibration parameters and we detail the scheme used to a posteriori set the zero level of the maps. We also briefly discuss the cross-calibration between HFI and the SPIRE instrument on board Herschel. We finally summarize the basic characteristics of the set of the HFI maps from the 2013 Planck data release.
Download