We apply state-of-the art data analysis methods to a number of fictitious cosmic microwave background (CMB) mapping experiments, including $1/f$ noise, distilling the cosmological information from time-ordered data to maps to power spectrum estimates, and find that in all cases the resulting error bars can be well approximated by simple and intuitive analytic expressions. Using these approximations, we discuss how to maximize the scientific return of CMB mapping experiments given the practical constraints at hand, and our main conclusions are as follows. (1) For a given resolution and sensitivity, it is best to cover a sky area such that the signal-to-noise ratio per resolution element (pixel) is of order unity. (2) It is best to avoid excessively skinny observing regions, narrower than a few degrees. (3) The minimum-variance map-making method can reduce the effects of $1/f$ noise by a substantial factor, but only if the scan pattern is thoroughly interconnected. (4) $1/f$ noise produces a $1/\mathcal{l}$ contribution to the angular power spectrum for well-connected single-beam scanning, as compared to virtually white noise for a two-beam scan pattern such as that of the MAP satellite.

• Received 27 May 1997

DOI:https://doi.org/10.1103/PhysRevD.56.4514