Harmonic analysis of cosmic microwave background data — I. Ring reductions and point-source catalogue

We present a harmonic model for the data analysis of an all-sky cosmic microwave background survey, such as Planck, where the survey is obtained through ring-scans of the sky. In this model, resampling and pixelization of the data are avoided. The spherical transforms of the sky at each frequency, in total intensity and polarization, as well as the bright-point-source catalogue, are derived directly from the data reduced on to the rings. Formal errors and the most significant correlation coefficients for the spherical transforms of the frequency maps are preserved. A clean and transparent path from the original samplings in the time domain to the final scientific products is thus obtained. The data analysis is largely based on Fourier analysis of rings; the positional stability of the instrument's spin axis during these scans is a requirement for the data model and is investigated here for the Planck satellite. Brighter point sources are recognized and extracted as part of the ring reductions and, on the basis of accumulated data, used to build the bright-point-source catalogue. The analysis of the rings is performed in an iterative loop, involving a range of geometric and detector response calibrations. The geometric calibrations are used to reconstruct the paths of the detectors over the sky during a scan and the phase offsets between scans of different detectors; the response calibrations eliminate short- and long-term variations in detector response. Point-source information may allow the reconstruction of the beam profile. The reconstructed spherical transforms of the sky in each frequency channel form the input to the subsequent analysis stages. Although the methods in this paper were developed with the data processing for the Planck satellite in mind, there are many aspects which have wider implementation possibilities, including the construction of real-space pixelized maps.