ALMACAL. XV. Band 3 ALMA Survey and Number Counts

TL;DR

This paper addresses the challenge of deriving deep extragalactic number counts from ALMACAL band 3 maps, where calibrator-driven biases can contaminate faint detections. It introduces UMLAUT, a KNN-based approach that assigns contamination probabilities to detections and integrates with simulations and physically motivated models to correct for completeness and flux boosting. The result is a robust band 3 $100\,\mathrm{GHz}$ counts measurement extending by ~1.5 dex in flux density, with sub-mJy sensitivities and improved sampling around the DSFG-to-radio-AGN transition, consistent with existing models and surveys. The work demonstrates a powerful, statistically principled path to extracting reliable counts from calibrator-centered ALMA fields, with implications for constraining galaxy evolution models at mm wavelengths.

Abstract

The ALMACAL project leverages ALMA maps of calibrator-centered fields to conduct deep mm/sub-mm surveys, enabling the detection of extragalactic sources with flux densities orders of magnitude fainter than achievable with other instruments. These faint sources are critical for refining evolutionary models, as their number counts provide key constraints. In this study, we analyzed band-3 ALMACAL maps from 606 calibrator fields, employing a novel machine learning approach to mitigate the often-overlooked bias introduced by the calibrator itself. Supported by extensive simulations, we extended 100 GHz radio AGN counts by approximately 1.5 orders of magnitude in flux density and refined constraints on dusty star-forming galaxies, reaching sensitivities as low as $\sim$180 $μ$Jy. We have improved the sampling, compared to previous results, in the region of the dominant population transition (between dusty star-forming galaxies and radio AGN). Our results are in good agreement with model predictions.

Figures (5)

• Figure 1: Effective area as a function of the $4.25\,\sigma$ detection limit. The maximum value, corresponding to the brightest flux limits, amounts to $681\,\hbox{arcmin}^2$.
• Figure 2: Examples of high SNR detections found by UMLAUT to be contaminants for our counts (real sources but associated to the calibrators). In each image, the calibrator is at the center. In the bottom right of each panel, we show the size and orientation of the beam (FWHM). Left panel: the source on the right (RA=36.11429 deg, DEC=6.99068 deg) has a flux density of 3.09 mJy, a $\hbox{SNR}=12.4$, and an angular distance from the calibrator of 15.4 arcsec. Central panel: the source on the left (RA=316.25964 deg, DEC=-48.81379 deg) has a flux density of 5.94 mJy, a $\hbox{SNR}=16.7$, and an angular distance from the calibrator of 11.7 arcsec. Right panel: the source on the right (RA=344.52264 deg, DEC=-27.97191 deg) has a flux density of 1.55 mJy, a $\hbox{SNR}=8.5$, and an angular distance from the calibrator of 7.7 arcsec.
• Figure 3: Observational estimates of Euclidean normalized differential number counts compared with theoretical predictions. Error bars for our counts include Poisson errors and uncertainties on contamination, completeness and flux boosting corrections. Our best estimates ($4.25\,\sigma$ detection limit) are compared with the Chen2023 and the Adscheid2024 observational number counts in band 3. We also show the band 4 counts by Chen2023, extrapolated to 100 GHz using a spectral index of 3.4 ($S_\nu \propto \nu^{3.4}$), corresponding to the peak in the distribution of spectral indices between 150 and 220 GHz of dusty galaxies reported by Everett2020. The same spectral index was used to extrapolate to 100 GHz the 150 GHz counts of dusty galaxies by Vargas2024 and by Everett2020. For the latter, we have adopted the "no-cuts" values extracted from Fig. 13 of Vargas2024. Above $\simeq 0.3\,$mJy, the counts are dominated by radio sources, in excellent agreement with predictions by DeZotti2005, Tucci2011, and Bonato2019. Our counts smoothly connect to the SPT counts by Everett2020 and the Atacama Cosmology Telescope (ACT) counts by Gralla2020 and Vargas2024. No correction was applied to the 95 GHz SPT counts, while the 150 GHz ACT counts were extrapolated to 100 GHz using the mean spectral index between 95 and 150 GHz (-0.6) found by Everett2020 for synchrotron sources.
• Figure 4: Our integral number counts of sources detected at $\ge 4.25\,\sigma$ compared with observational estimates by Zavala2018, GL2019, Chen2023, Long2024, and Adscheid2024; the latter were extracted from their Fig. 8. The 2 mm counts by Long2024 were extrapolated to 100 GHz adopting a spectral index of 3.4 (see caption to Fig. \ref{['fig:END_nc']}). The uncertainties on our counts include Poisson errors and uncertainties on contamination, completeness and flux boosting corrections. Also shown are predictions by Tucci2011 for radio sources and by Cai2013 and Popping2020 for DSFGs; the latter counts were taken from Fig. 10 of Chen2023. The solid grey line is the sum of radio counts with those by Cai2013.
• Figure 5: Aperture flux as a function of the aperture size (i.e., curves of growth) for the five sources detected in both Chen2023 and this work (dashed red lines). The same function is measured also for the 5 calibrators in the same images (dashed green lines). The average curves are shown with solid lines. Each curve is normalized to the maximum flux reached among the apertures considered. The size of the apertures is normalized to the nominal beam size in each image.