CMB Experiment Baseline Determination by Minimizing Pixel Observations Dispersion

This paper discusses a couple of approaches to removing instrument gain and baseline variation from observations of the CMB anisotropy. These techniques were tested on the COBE DMR data, a balloon experiment, and by monte carlo simulations. This work, though considered technical, is made available because CMB anisotropy mapping data processing is currently very active and a significant number of groups are working on this issue. Specifically, time-ordered estimates of the gain and baseline via fitting to limited sections of data or to the full or a large section of data are considered and compared. An expression is given for the correlation between pixels introduced by the residual baseline and removal technique.