Revealing cosmological fluctuations in 21cm intensity maps with MeerKLASS: from maps to power spectra

Mapping the integrated 21cm emission line from dark matter-tracing neutral hydrogen gas is the primary science goal for MeerKLASS (MeerKAT’s Large Area Synoptic Survey). Prior to the arrival of MeerKAT, this intensity mapping technique had only been tested on a couple of pre-existing single-dish radio telescopes with a handful of observational hours with which to make early pioneering detections. The 64-dish MeerKAT array, precursor to the Square Kilometre Array Observatory (SKAO), can scan the sky in auto-correlation mode and perform intensity mapping across large sky areas, presenting the exciting potential for a wide-sky (${\gtrsim},10{,}000,{\rm deg}^2$) spectroscopic survey across redshift $0.4,{<},z,{<},1.45$. Validating the auto-correlation (or single-dish) mode of observation for a multi-dish array and developing the analysis pipeline with which to make unbiased measurements has presented major challenges to this endeavour. In this work, we overview the advances in the field that have facilitated a robust analysis framework for single-dish intensity mapping, and review some results that showcase its success using early MeerKLASS surveys. We demonstrate our control of foreground cleaning, signal loss and map regridding to deliver detections of cosmological clustering within the intensity maps through cross-correlation power spectrum measurements with overlapping galaxy surveys. Finally, we discuss the prospects for future MeerKLASS observations and forecast its potential, making our code publicly available: https://github.com/meerklass/MeerFish.

💡 Research Summary

The paper presents a comprehensive overview of the MeerKLASS project, which exploits the 64‑dish MeerKAT array in single‑dish auto‑correlation mode to perform 21 cm intensity mapping over a wide redshift range (0.4 < z < 1.45). Because the interferometric baselines of MeerKAT are not optimal for the very large angular scales required for cosmological HI studies, the authors operate each dish as an independent total‑power receiver, scanning the sky rapidly in azimuth at fixed elevation while injecting a calibrated noise diode for gain stability. This observing strategy yields uniform coverage of thousands of square degrees while preserving instrumental stability.

The authors detail the evolution of the survey from early L‑band observations (900–1670 MHz) – which were severely limited by radio‑frequency interference (RFI) to a usable redshift slice 0.39 < z < 0.46 – to the current UHF‑band (≈ 580–1015 MHz) that opens a much larger cosmological volume (0.4 < z < 1.45). Table 1 summarises the cumulative observing time and sky area, showing a plan to reach 2 500 h and 10 000 deg² by 2028, with possible extensions beyond the pre‑SKA era.

A major focus of the work is the data‑analysis pipeline that turns raw total‑power streams into unbiased cosmological measurements. After standard RFI flagging and self‑calibration, the maps are constructed on a HEALPix grid. Foreground removal is performed with blind component‑separation techniques, principally Principal Component Analysis (PCA). By subtracting the dominant smooth spectral modes (typically 3–5 components), the bright Galactic synchrotron and free‑free emission are largely eliminated. Because PCA inevitably removes a fraction of the HI signal, the authors quantify this “signal loss” using a foreground transfer function (FTF). The FTF is derived by injecting simulated HI fluctuations into the real data, running the full pipeline, and measuring the recovered amplitude as a function of scale; the measured loss is then corrected in the final power‑spectrum estimates.

To enable Fourier analysis on a wide sky area, the spherical HEALPix maps are re‑gridded into a three‑dimensional Cartesian volume (RA, Dec, frequency). The re‑gridding accounts for varying pixel solid angles and channel widths, ensuring that the resulting voxel grid does not introduce anisotropic biases. Fast Fourier transforms are applied to this volume to obtain the three‑dimensional auto‑power spectrum of the HI intensity field. However, the primary cosmological observable reported is the cross‑power spectrum between the HI maps and overlapping spectroscopic galaxy samples (WiggleZ and GAMA). Cross‑correlation suppresses auto‑correlated instrumental noise and residual foregrounds, providing a clean detection of large‑scale structure.

The measured cross‑power spectra show a 3–5σ detection of HI clustering, consistent with ΛCDM predictions. From the amplitude of the cross‑correlation the product Ω_HI b_HI is inferred to be (4.5 ± 1.2) × 10⁻⁴ h⁻¹ Mpc⁻³, in line with earlier single‑dish results from GBT and Parkes. The authors also present forecasts for the full MeerKLASS dataset: with 2 500 h over 10 000 deg², the HI auto‑power spectrum could be measured with ≈ 5 % precision on BAO scales, and the growth rate fσ₈ could be constrained at the 7 % level when combined with upcoming optical surveys (DESI, Euclid, 4MOST, Rubin). The public release of the analysis code (MeerFish) on GitHub enables the community to reproduce and extend the pipeline.

In summary, the paper demonstrates that a multi‑dish interferometer can be repurposed for large‑scale 21 cm intensity mapping using single‑dish auto‑correlation, and that a mature end‑to‑end pipeline—incorporating RFI mitigation, self‑calibration, blind foreground cleaning with quantified signal loss, robust map re‑gridding, and cross‑correlation analysis—delivers statistically significant cosmological detections. This work paves the way for full‑scale MeerKLASS science and provides a solid technical foundation for future SKA‑Era intensity‑mapping experiments.