🔭 astrophysics

SPT-3G D1: Maps of the millimeter-wave sky from 2019 and 2020 observations of the SPT-3G Main field

This paper presents the production, validation, and public release of the deepest multifrequency millimeter-wave sky maps to date from the South Pole Telescope's SPT-3G camera, covering 95, 150, and 220 GHz observations of the Main field from 2019 and 2020, which are optimized for precise measurements of CMB anisotropies and gravitational lensing potential.

Original authors: W. Quan, E. Camphuis, C. Daley, N. Huang, Y. Omori, F. Guidi, E. Anderes, A. J. Anderson, B. Ansarinejad, M. Archipley, L. Balkenhol, D. R. Barron, K. Benabed, A. N. Bender, B. A. Benson, F. Bianchini

Published 2026-03-23

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: W. Quan, E. Camphuis, C. Daley, N. Huang, Y. Omori, F. Guidi, E. Anderes, A. J. Anderson, B. Ansarinejad, M. Archipley, L. Balkenhol, D. R. Barron, K. Benabed, A. N. Bender, B. A. Benson, F. Bianchini, L. E. Bleem, S. Bocquet, F. R. Bouchet, M. G. Campitiello, J. E. Carlstrom, J. Carron, C. L. Chang, P. M. Chichura, A. Chokshi, T. -L. Chou, A. Coerver, T. M. Crawford, T. de Haan, K. R. Dibert, M. A. Dobbs, M. Doohan, D. Dutcher, C. Feng, K. R. Ferguson, N. C. Ferree, K. Fichman, A. Foster, S. Galli, A. E. Gambrel, A. K. Gao, F. Ge, S. Guns, N. W. Halverson, E. Hivon, G. P. Holder, W. L. Holzapfel, J. C. Hood, A. Hryciuk, T. Jhaveri, F. Kéruzoré, A. R. Khalife, L. Knox, K. Kornoelje, C. -L. Kuo, K. Levy, Y. Li, A. E. Lowitz, C. Lu, G. P. Lynch, T. J. Maccarone, A. S. Maniyar, E. S. Martsen, F. Menanteau, M. Millea, J. Montgomery, Y. Nakato, T. Natoli, A. Ouellette, Z. Pan, P. Paschos, K. A. Phadke, A. W. Pollak, K. Prabhu, S. Raghunathan, M. Rahimi, A. Rahlin, C. L. Reichardt, M. Rouble, J. E. Ruhl, A. C. Silva Oliveira, A. Simpson, J. A. Sobrin, A. A. Stark, J. Stephen, C. Tandoi, C. Trendafilova, J. D. Vieira, A. G. Vieregg, A. Vitrier, Y. Wan, N. Whitehorn, W. L. K. Wu, M. R. Young, J. A. Zebrowski

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe is a giant, ancient ocean. Most of the time, it looks calm and uniform, but if you look closely, there are tiny ripples and waves on the surface. These ripples are the Cosmic Microwave Background (CMB)—the leftover heat from the Big Bang. By studying these ripples, scientists can learn how the universe was born, how it grew, and what it's made of.

This paper is a report card from a team of astronomers using a super-powered camera called SPT-3G, sitting at the very bottom of the world (the South Pole). They spent two years (2019 and 2020) taking a very deep, high-resolution "selfie" of a specific patch of the sky.

Here is the story of how they took that picture, cleaned it up, and made sure it was real, explained in simple terms.

1. The Camera and the Mission

Think of the SPT-3G camera as a digital camera with 16,000 tiny eyes (detectors) instead of just one. These eyes are tuned to see three different "colors" of invisible light (microwaves) that our eyes can't see: 95, 150, and 220 GHz.

They pointed this camera at a patch of sky about the size of 16 full moons. It's a small area, but they looked at it so deeply and for so long that they could see details 3 to 8 times sharper than previous space telescopes like Planck. It's like taking a photo of a grain of sand from a mile away and being able to read the text on it.

2. The "Filter-and-Bin" Recipe

When the camera takes a picture, it doesn't just snap a photo; it records a continuous stream of data (a "timestream") for every single eye. But this data is messy. It's like trying to hear a whisper in a room where the wind is howling (the atmosphere) and the lights are buzzing (instrument noise).

To get a clear picture, the team used a method called "Filter-and-Bin":

Filtering: Imagine you are listening to a song, but there's a low, rumbling bass noise drowning out the melody. They used a "high-pass filter" to cut out that low rumble (the slow, noisy changes in the atmosphere). They kept the fast, interesting parts (the actual cosmic signals).
Binning: Once the noise was cut, they took all the tiny data points and dumped them into "buckets" (pixels) on a map. If a bucket had 1,000 data points, they averaged them to get one clear value for that spot on the sky.

3. Cleaning the Lens (Calibration)

Even after filtering, the picture might be slightly off. Maybe the camera is a little too sensitive to blue light, or maybe the "north" on the map is tilted. The team had to perform a series of "cleaning" steps:

Gain Calibration: They checked if the camera was seeing things as bright as they really are. They compared their map to a trusted map from the Planck satellite (like checking your watch against a master clock).
Leakage Removal: Sometimes, a very bright signal (like the total heat of the universe) accidentally "leaks" into the polarization signal (the direction of the light waves). It's like a loud shout drowning out a whisper. They mathematically subtracted this leakage to make sure the whisper was pure.
Angle Correction: They made sure the "compass" on their map was pointing true North. If the compass is off by even a tiny bit, the patterns of the universe look twisted. They rotated the map slightly to fix this.

4. The "Null Test" (The Lie Detector)

How do you know you didn't just imagine the patterns? You have to prove there are no hidden tricks. The team performed "Null Tests."

Imagine you split your data into two piles: Pile A and Pile B.

If you subtract Pile A from Pile B, the real universe (which is in both piles) should disappear, leaving you with nothing but random static (noise).
If you see a pattern in the difference, it means something is wrong—maybe the telescope is vibrating, or the Sun is interfering.

They found a problem! In some of their tests, they saw a weird, narrow spike in the data (like a radio station interfering with a song). It turned out to be a vibration in the telescope's drive system, happening at a specific frequency. They didn't throw the data away; instead, they put a "mask" over that specific frequency in their math, effectively muting the interference.

5. The Result: The Deepest Map Yet

After all this cleaning, filtering, and testing, they produced the deepest, most detailed map of the cosmic microwave background polarization ever made.

Why does this matter? The "polarization" is like the direction the light waves are vibrating. It holds secrets about how the universe expanded in its first fraction of a second (inflation) and how massive galaxies bent the light (gravitational lensing).
The Noise Level: The "static" in their map is incredibly low. For the 150 GHz band, the noise is only 4.4 micro-Kelvin-arcminutes. To put that in perspective, if the temperature of the CMB were the height of a mountain, the noise in this map would be smaller than a single grain of sand on that mountain.

Summary

This paper is the technical manual for the "cleanest" picture of the early universe we have ever taken. The team at the South Pole built a massive camera, developed a recipe to filter out the noise of the atmosphere and the telescope, fixed their own mistakes, and proved that their map is real.

Now, other scientists can use these maps to solve the biggest mysteries of cosmology, like: What is dark matter? and How did the universe begin? They have handed the world a new, high-definition lens to look at the beginning of time.

1. Problem and Context

The Cosmic Microwave Background (CMB) is a cornerstone of modern cosmology, providing constraints on the standard $\Lambda$ CDM model. While satellite missions like Planck have provided high-precision measurements of large-scale CMB anisotropies, ground-based experiments are required to probe small angular scales (high multipoles, $\ell$ ) with higher sensitivity.
The South Pole Telescope (SPT) with its third-generation camera, SPT-3G, features $\sim$ 16,000 detectors across three frequency bands (95, 150, and 220 GHz). Previous analyses (e.g., the 2018 dataset) were limited by detector count and observing time. This paper addresses the need to produce, calibrate, and validate the primary data products (sky maps) from the SPT-3G D1 dataset, which comprises observations from the 2019 and 2020 austral winter seasons. These maps are the foundation for subsequent measurements of CMB temperature/polarization power spectra and gravitational lensing.

2. Methodology

The paper details a rigorous pipeline for converting raw detector time-streams into calibrated Stokes parameter maps ( $I/T$ , $Q$ , $U$ ).

A. Observations and Data Selection

Field: The SPT-3G Main field covers $\sim$ 1550 deg $^2$ (4% of the sky) with low galactic dust emission. It is divided into four subfields (el0–el3) to allow detector retuning for stable operation.
Dataset: SPT-3G D1 includes 3,334 subfield observations from 2019–2020.
Detectors: An average of 11,316 detectors were operated, with 10,697 passing quality cuts (evenly distributed across 95, 150, and 220 GHz bands).
Calibration: Weekly observations of HII regions (RCW38 and Mat5a) and daily "calibrator stares" (thermal sources) and "elevation nods" (sky dips) were used to estimate detector gains and time constants.

B. Mapmaking (Filter-and-Bin)

The team employed the filter-and-bin method:

High-Pass Filtering: To remove correlated low-frequency noise (atmospheric $1/f$ $1/ f$ noise and readout noise), a high-pass filter was applied to time-streams. This involved deprojecting Legendre polynomials (up to 30th order) and sinusoids.
- Cutoff: A scan-direction multipole cutoff of $\ell_{x,c} = 300$ was chosen to balance noise reduction with signal preservation.
- Masking: Bright sources (AGN and galaxy clusters) were masked in the time-streams during filtering to prevent "filtering wings" (artifacts extending across the map).
Low-Pass Filtering: An anti-aliasing filter ( $\ell_{x,c} = 13,000$ ) was applied to prevent aliasing before binning.
Binning: Filtered time-stream samples were binned into pixels using inverse-variance weighting. Two pixelization schemes were used: HEALPix ( $N_{side}=8192$ ) and a flat-sky Lambert azimuthal equal-area projection (ZEA).
Coadding: Individual observation maps were combined into various coadds:
- Full-depth: For final science analysis.
- Half-depth: For cross-correlation to remove noise bias.
- One-thirtieth: For power spectrum estimation.
- Signflip-noise: Created by flipping signs of individual maps to generate pure noise realizations.

C. Calibration and Cleaning

Four sequential steps were applied to correct systematic biases in the coadded maps:

Gain Calibration: Corrected for multiplicative biases in total intensity ( $T$ ) by cross-correlating SPT maps with Planck (143 GHz) and internal frequency bands.
Monopole T-to-P Leakage Removal: Corrected for leakage of temperature signal into polarization ( $Q/U$ ) caused by differential gain miscalibrations correlated with polarization angles. This was estimated by minimizing the $TQ$ and $TU$ cross-spectra.
Polarization Angle Calibration: Corrected for average miscalibration of detector polarization angles (which mixes $E$ and $B$ modes) by minimizing the $EB$ and $TB$ cross-spectra.
Polarization Amplitude Calibration: Corrected for the suppression of polarization amplitude due to imperfect polarization efficiency and angle variations, calibrated against Planck and internal bands.

D. Null Tests and Systematic Error Mitigation

To validate the maps, the dataset was split in six ways (Sun, Moon, Azimuth, Year, Scan direction, Detector Wafer) to create null maps (difference maps where astrophysical signals should cancel).

Detection: Initial null tests revealed significant narrowband contamination at $\ell \approx 600$ in the Scan and Wafer splits, localized to the edges of the field and specific declinations.
Mitigation: The contamination was traced to a $\sim$ 1.1 Hz oscillation likely related to the telescope drive system. The team mitigated this by masking the contaminated modes in spherical harmonic space ( $500 \le \ell \le 680$ and $350 \le m \le 425$ ) during power spectrum calculation.
Result: After masking, all 54 null spectra (TT, TE, EE for 3 bands $\times$ 6 splits) passed statistical tests (p-values $> 0.05$ after Bonferroni correction), confirming the absence of significant systematic errors.

3. Key Contributions

Data Release: Public release of the deepest wide-area, high-resolution CMB maps to date from the SPT-3G Main field (2019–2020).
Pipeline Refinement: Implementation of an improved mapmaking pipeline that removes the common-mode filter step (used in 2018) to avoid T-to-P leakage, while introducing robust masking of bright sources during filtering.
Systematic Error Characterization: Detailed identification and mitigation of a narrowband contamination source in the null spectra, demonstrating a rigorous approach to validating high-precision CMB data.
Comprehensive Calibration: A full suite of calibration steps addressing gain, leakage, angle, and amplitude, with uncertainties quantified and propagated.

4. Key Results

Map Depth (White Noise Levels):
- Temperature (I/T): 5.4, 4.4, and 16.2 $\mu$ K-arcmin at 95, 150, and 220 GHz, respectively.
- Polarization (Q/U): 8.4, 6.6, and 25.8 $\mu$ K-arcmin at 95, 150, and 220 GHz, respectively.
Comparison: The SPT-3G D1 maps are 3 times deeper than ACT DR6 and 8 times deeper than Planck PR3, despite covering only 4% of the sky.
Validation: The null tests confirmed that systematic errors are negligible compared to the sample variance of the CMB measurements. The narrowband contamination was successfully removed without compromising the scientific utility of the maps.
Transfer Functions: The beam and filter transfer functions were characterized and made publicly available to allow unbiased power spectrum estimation.

5. Significance

Cosmological Constraints: These maps serve as the primary input for companion papers (G25, C25, O26) that present state-of-the-art measurements of CMB temperature and E-mode polarization power spectra, as well as CMB gravitational lensing potential. These measurements provide tight constraints on the $\Lambda$ CDM model and its extensions.
Future Probes: The depth and resolution of these maps enable new applications, including multi-wavelength cross-correlations, stacking studies of galaxy clusters and point sources, and tests of new cosmological probes requiring deep, high-resolution maps.
Open Science: The authors have made the maps, weight matrices, beam/filter transfer functions, and simulation data publicly accessible, facilitating further research by the broader astrophysics community.

In summary, this paper establishes the data foundation for the next generation of CMB cosmology, demonstrating that SPT-3G can produce maps with unprecedented sensitivity and control over systematic errors, enabling precise tests of the early universe and fundamental physics.