The Atacama Cosmology Telescope: Data Characterization and Map Making

TL;DR

This paper presents the end-to-end data reduction and mapmaking pipeline for ACT's 2008 148 GHz observations, detailing how atmospheric and instrumental noise, along with systematic errors, are characterized and mitigated. The authors implement a PCG-based mapmaking framework that explicitly models correlated noise, applies dark-detector mode removal, and subtracts bright point-source flux in the time streams to produce unbiased sky maps. The resulting maps cover 845.6 deg^2 with a total NET of about 32 μK√s, demonstrate consistency with WMAP on overlapping scales, and include a public data release, thereby enabling robust cosmological analyses and future improvements. The work advances CMB data processing by integrating detailed noise modeling, calibration, pointing, and source handling to achieve high-fidelity maps at sub-degree angular scales.

Abstract

We present a description of the data reduction and mapmaking pipeline used for the 2008 observing season of the Atacama Cosmology Telescope (ACT). The data presented here at 148 GHz represent 12% of the 90 TB collected by ACT from 2007 to 2010. In 2008 we observed for 136 days, producing a total of 1423 hours of data (11 TB for the 148 GHz band only), with a daily average of 10.5 hours of observation. From these, 1085 hours were devoted to a 850 deg^2 stripe (11.2 hours by 9.1 deg) centered on a declination of -52.7 deg, while 175 hours were devoted to a 280 deg^2 stripe (4.5 hours by 4.8 deg) centered at the celestial equator. We discuss sources of statistical and systematic noise, calibration, telescope pointing, and data selection. Out of 1260 survey hours and 1024 detectors per array, 816 hours and 593 effective detectors remain after data selection for this frequency band, yielding a 38% survey efficiency. The total sensitivity in 2008, determined from the noise level between 5 Hz and 20 Hz in the time-ordered data stream (TOD), is 32 micro-Kelvin sqrt{s} in CMB units. Atmospheric brightness fluctuations constitute the main contaminant in the data and dominate the detector noise covariance at low frequencies in the TOD. The maps were made by solving the least-squares problem using the Preconditioned Conjugate Gradient method, incorporating the details of the detector and noise correlations. Cross-correlation with WMAP sky maps, as well as analysis from simulations, reveal that our maps are unbiased at multipoles ell > 300. This paper accompanies the public release of the 148 GHz southern stripe maps from 2008. The techniques described here will be applied to future maps and data releases.

Figures (13)

• Figure 1: Example TOD from one detector at 148 GHz from October 21, 2008. This was a good observing night with a PWV of 0.22 mm. The slow drift is dominated by changes in the atmosphere brightness. The high-frequency noise is dominated by detector noise. Units are $\milli\kelvin$ in CMB equivalent units at 148 GHz. The plot displays $3.6\times10^5$ samples at an interval of $2.5\times10^{-3}$ seconds. The telescope was scanning while these data were taken.
• Figure 2: Histogram showing the PWV during the 2008 season, measured at the zenith of the Atacama Pathfinder Experiment (APEX) facility. The median value is 0.49 .
• Figure 3: Power spectra of signals in the 2008 season compared to the data. The solid curve is the average power spectrum from 737 live detector TODs from one 15-minute 148 GHz data file, during which the PWV was 0.5 mm. The rise at low frequencies is the noise contribution from the atmosphere. The dashed curve shows the simulated CMB signal. The oscillations in the simulated CMB signal are due to enhanced power at harmonics of the scan frequency. The dot-dashed curve estimates the point source contribution (considering only one hit at the beam center) assuming a Gaussian beam. For comparison, the thin and thick dotted curves show the expected response for 218 GHz and 277 GHz, with beam sizes of 100 and 091 respectively. The amplitude of the point source power spectra corresponds to approximately the signal of Saturn.
• Figure 4: Atmospheric signature in TOD power spectrum. (a) Average power spectra for four groups of non-scanning TODs selected by PWV level. The average power spectrum from the dark detectors was subtracted from each group power spectrum and is shown for comparison. The dashed lines show a power law fit to each spectrum. The legend indicates the mean PWV and the power-law index from the fit for each group. Logarithmic binning was used to reduce the noise in the plot. (b) Power law index as a function of frequency for the average power spectra from three groups (blue ($\bullet$), green ($\blacktriangledown$) and red ($\blacktriangle$)) of observations selected by their power law index at frequencies close to the knee, all belonging to the third PWV selected group in (a). Each value was obtained by fitting a power law to the corresponding power spectrum in a small frequency range (in logarithmic space); the listed frequencies are the mean values of the associated frequency range. The error bars show the dispersion of the indexes from the members of each group. The same is shown in brown (${\blacksquare}$) for the dark detector average power spectrum for comparison.
• Figure 5: Total array NET during the 2008 season in the mid-frequency range (5-20 ) for 148 GHz. The values are grouped in days and the error bars are the standard deviations within each day. The PWV is also plotted for reference (dashed line). Values are calibrated to CMB equivalent units. Before computing the sensitivity, eight multi-common modes, four row-correlated modes, four column-correlated modes, and the residual twelve modes with highest singular values were removed (see Appendix \ref{['mode_removal']} for details). The noise improvement after September 24 came from turning off some oscillating detectors, which were contaminating neighbors.