In this paper
AbstractIntroductionObservations and Data ReductionResultsDiffuse Radio Emission in Abell 370Multiwavelength DataSpectral IndexDiscussionX-ray-radio CorrelationSpectral Index MapScaling RelationsClassificationConclusionsACKNOWLEDGEMENTSData AvailabilityReferencesWorksPhysics and Astronomy
MeerKAT reveals an ultra-steep spectrum radio halo in Abell 370
Abstract
Hubble Frontier Field cluster Abell 370 () has been the target of a suite of legacy multi-wavelength observational campaigns and is a prime candidate to host a diffuse radio halo, given its high mass () and non-relaxed dynamical state. Previous detections of its radio halo were of low significance and imaging fidelity, leading to an uncertain characterisation of its nature. We use deep MeerKAT Ultra-High Frequency observations () to robustly map the halo's spatial extent and spectral properties. We find a radio emission size of , significantly larger than previously reported. We measure a MeerKAT UHF in-band integrated spectral index of . This result is consistent with an ultra-steep spectrum radio halo (). We find a strong, sub-linear correlation () between the radio and X-ray surface brightness. The radio halo's position on the scaling relation is underluminous, consistent with the known USSRH population. The combined evidence strongly supports that the diffuse central radio emission in Abell 370 is an Ultra-Steep Spectrum Radio Halo (USSRH), refining its previous classification and providing further support for the turbulent re-acceleration model.
Introduction
Galaxy clusters are the largest gravitationally bound and virialised structures in the Universe, with halo masses of . They are situated at the nodes of the cosmic web, where large-scale structure filaments intersect (Aragón-Calvo et al. 2010), and are assembled through the hierarchical merging of smaller galaxy groups and clusters ([55]). The primary component of these massive halos is dominated by dark matter, which makes up 70-80 per cent of the mass (Blumenthal et al. 1984). The remaining mass is baryonic matter composed of the dominant intra-cluster medium (ICM), a hot ( K), optically thin gas filling the cluster potential (Mitchell et al. 1977; Serlemitsos et al. 1977), and the negligible stellar mass in member galaxies.
Historically, the ICM has been traced by its thermal X-ray emission. However, modern radio telescopes have detected non-thermal synchrotron emission counterparts in a fraction of galaxy clusters, generated by the presence of relativistic particles in large-scale magnetic fields. This diffuse radio emission is broadly classified into four main categories: giant radio halos, mini-halos, radio relics, and radio phoenixes (see [69] (Brunetti & Jones 2014), for a review).
Giant radio halos (RHs) are spatially extended ( Mpc), centrally located, and found predominantly in massive, dynamically disturbed (merging) systems (Cassano et al. 2013; Cuciti et al. 2021b). Typical RHs have spectral indices1 in the range of to (Giovannini et al. 2009; Ferretti et al. 2012). Their radio emission is known to spatially correlate with the ICM distribution, traced by thermal X-ray emission (Govoni et al. 2001a; Giacintucci et al. 2005). In the last two decades, a new population of ultra-steep spectrum radio halos (USSRHs) has been revealed, with steeper spectra of , with their origin connected to less energetic mergers (Brunetti et al. 2008; Dallacasa et al. 2009; Macario et al. 2010; Bonafede et al. 2012; Duchesne et al. 2021; Cuciti et al. 2021a; Pasini et al. 2022; Edler et al. 2022; Venturi et al. 2022; Rajpurohit et al. 2023; Magologo et al. 2026).
Radio relics are elongated, arc-shaped sources, often highly polarised and located in cluster periphery (Ensslin et al. 1998; de Gasperin et al. 2022; Pasini et al. 2022). Their origin is linked to particle acceleration from shock fronts, supported by their spatial coincidence with X-ray surface brightness and temperature discontinuities (Akamatsu et al. 2015; Eckert et al. 2016).
A related relic classification, radio phoenixes, are smaller ( kpc) sources, thought to be "revived" from fossil lobes of past active galactic nucleus (AGN) activity. Most commonly, they have filamentary-like morphology, though this can vary (Slee et al. 2001). Adiabatic compression from passing shocks is thought to re-energise the old plasma, resulting in steep power law spectra () sources closer to the cluster centre when compared to relics ([68] (Feretti et al. 2012)).
Finally, mini-halos (MHs) are smaller-scale ( kpc), steep-spectrum, diffuse sources found at the centre of dynamically relaxed, cool-core clusters (Giacintucci et al. 2017; Kolokythas et al. 2025). Their origin is thought to be related to turbulent re-acceleration driven by gas "sloshing" in the cluster core, distinctly different from the highenergy turbulence associated with merging clusters (Giacintucci et al. 2014; Brunetti & Jones 2014; Giacintucci et al. 2017).
The origin of RHs is still debated, with two primary competing models: the turbulent re-acceleration model and the hadronic model. In the currently favoured turbulent re-acceleration model, a pre-existing population of electrons is accelerated by turbulent energy injected through merger events (Ensslin et al. 1998; Brunetti 2004). In the hadronic model, relativistic electrons are produced as secondary particles from collisions between ICM and cosmic-ray protons (Dennison 1980). The existence of USSRHs supports the turbulent re-acceleration model, as it is difficult to reproduce USSRHs in purely hadronic scenarios. While a combination of these mechanisms being responsible is a possibility (Brunetti & Lazarian 2011), next-generation radio telescopes will provide important constraints on the faint radio emission in relaxed galaxy clusters, helping determine the contributions of different mechanisms. It is important in this new era to build up large statistical samples of USSRHs with high fidelity mapping, broad spectral coverage, and well calibrated masses.
The galaxy cluster Abell 370 () is a massive unrelaxed system () in a Hubble Frontier Field (Abell et al. 1989; Planck Collaboration et al. 2016; Lotz et al. 2017). It is renowned for being the first cluster discovered to gravitationally lens a background galaxy (Hoag 1981; Soucail et al. 1987; Paczynski 1987), and has been extensively studied across the electromagnetic spectrum (Broadhurst et al. 2008; Richard et al. 2010; Lagattuta et al. 2017; Xie et al. 2020; Kolokythas et al. 2025). Its dynamical state is a clear merger, with a bimodal mass distribution centred on two distinct brightest cluster galaxies (BCGs) (Richard et al. 2010). Abell 370 is expected to host diffuse radio emission consistent with a typical RH, given its high mass and dynamically disturbed state. Diffuse radio emission in Abell 370 was first discovered by Xie et al. (2020), who detected a faint radio halo at 325 MHz and 1.5 GHz using Giant Metrewave Radio Telescope (GMRT) and Very Large Array (VLA) observations, and place an upper limit on the emission at 3 GHz. They report an integrated spectral index of and classify the source as a radio halo. At 1.28 GHz this faint emission was detected and classified as a candidate radio halo by Knowles et al. (2022) as part of the MeerKAT Galaxy Cluster Legacy Survey (MGCLS; see Kolokythas et al. 2025).
In this work, we begin by describing the data processing strategy in § 2. In § 3.1, we present the radio continuum images, which reveal a larger "largest linear size" (LLS) than previously reported and unknown radio extensions to the southeast and southwest not included in the literature. § 3.2 qualitatively explores the co-location of thermal X-ray emission and radio emission. In § 3.3 we derive a steep spectral index consistent with an USSRH for Abell 370. We then quantitatively confirm the X-ray-radio co-location in § 4.1, using a point-to-point region-based analysis. We investigate the parameter space in the power-mass scaling relation occupied by typical radio halos versus that of USSRHs in § 4.3. Finally, we summarise our findings in § 5.
Observations and Data Reduction
The data used in this work stems from MeerKAT open time data (project ID: SCI-20241101-GL-01), and is comprised of four observations of hours each. The observations were taken using the MeerKAT UHF-band (580 − 1015 MHz frequency range and 544 MHz bandwidth) in the wideband fine 32K channelisation mode (16.6 kHz channel width). All four observations use more than 61 of the 64 available antennas, have dump times of sec, and observe the following sources: J0408-6545 (flux and bandpass calibrator), J0240-2309 (complex gains calibrator), and A370 (target galaxy cluster). The observations were made on the 21st, 24th, 27th, and 28th of December 2024.
The MeerKAT UHF-band raw visibilities are reduced using the semi-automatic MeerKAT data reduction pipeline OXKAT (Heywood 2020), which uses routines that are standard practice for MeerKAT data calibration, such as flagging, reference calibration (1GC), self-calibration (2GC), and direction-dependent calibration (3GC). OXKAT uses tasks from CASA (McMullin et al. 2007), CUBICAL (Kenyon et al. 2018), and KILLMS (Tasse 2023), while using WSCLEAN (Offringa et al. 2014) and DDFACET (Tasse et al. 2023) for de-convolution and imaging at the 2GC and 3GC stage, respectively.
Together with the automated RFI excision software package, TRICOLOUR (Hugo et al. 2022), we perform manual flagging of bad antennas, ranges, and occasionally calibrator scans, to remove radio frequency interference (RFI) and bad data. Cross calibration is done through OXKAT, which uses standard CASA tasks such as GAINCAL and APPLYCAL to correct for atmospheric phase shifts among other effects. Self-calibration is then done by using CUBICAL, in conjunction with WSCLEAN, to iteratively improve the cleaning mask, resulting in higher imaging fidelity. However, significant DDEs in the field, caused by bright off-axis sources, limit the improvements gained by self-calibration.
To mitigate the effects of these DDEs, we utilise the "peeling" functionality in the 3GC section of the OXKAT pipeline. Peeling involves predicting visibilities for a problematic source, and for the full sky model, then subtracting the problematic visibilities from the full sky model with CUBICAL. We sequentially peel six bright, off-centre sources using this method. The result is a significant reduction in DDEs in the field, and a significant improvement in both the imaging fidelity and dynamical range.
To accurately characterise diffuse radio emission, it is vital to correct for the flux contribution from compact sources hosted within the diffuse emission. To achieve this, we perform point source subtraction using CRYSTALLBALL [42], CASA’s UVSUB, and WSCLEAN in a similar manner to Magolego et al. (2026). A high-resolution continuum image, produced with a robust Briggs weighting of -2, is used to identify and draw regions around point sources in and around the already visible, diffuse emission. We then use CRYSTALLBALL to generate visibilities for these identified point sources, which are subtracted from the calibrated visibilities using UVSUB. This procedure is performed twice, as re-imaging reveals additional point sources due to reduced blending. The final, point-source free, visibilities are imaged with a robust Briggs weighting of 0.5 to produce a lower-resolution image which is more sensitive to the faint, diffuse, and extended radio emission. Despite the -plane subtraction, residual emission from bright point sources can remain in the image plane. To mitigate this we implement an additional image-plane masking technique where we apply spatial masks over regions affected by point-source emission. We then extrapolate the mean surface brightness of surrounding clean diffuse emission across the masked regions by interpolating the data using a 2D Gaussian kernel scaled to the synthesized beam size. This combined approach allows for a robust extraction of the radio surface brightness and total flux density across the halo emission.
Results
Diffuse Radio Emission in Abell 370
A crucial step in characterising radio halos in galaxy clusters is to model and subtract contaminating point sources (outlined in § 2). This isolates the diffuse radio emission, allowing for a study of its full extent and morphology. Here, we showcase the enhanced image fidelity achieved through meticulous subtraction, which enables the high-fidelity mapping of the full halo extent and features discussed here.
In Figure 1 we show the MeerKAT UHF-band view of the radio halo in Abell 370. In the left panel, we show a higher-resolution (robust Briggs=) radio continuum map before point source subtraction with point spread function (PSF) dimensions of arcsec, while the right panel shows the lower-resolution (robust Briggs=) arcsec Gaussian-tapered, point source subtracted image with a PSF of arcsec, see § 2 for details. In both panels, we indicate the radius within which the mean total mass density is times the critical density of the Universe (), the PSF, a kpc scale bar, and the Atacama Cosmology Telescope (ACT) SZ detection centroid (Hilton et al. 2021). The physical radius is derived from the ACT DR5 cluster mass (Hilton et al. 2021). Contours are overlaid at where is determined using a nearby source-free region in the respective radio map, and Jy beam for the full- and lower-resolution images, respectively. We encircle the BCGs to the north and south of the SZ centre in orange. Finally, pixels are colourised by radio brightness in units of Jy beam.

Figure 1. The full-resolution map before point source subtraction (left) and the lower-resolution map after point source subtraction (right) of the MeerKAT UHF (816 MHz central frequency) radio emission detected in galaxy cluster Abell 370. The full-resolution map is imaged with a robust weighting of -0.5 and PSF dimensions of 10 10 arcsec, while the lower-resolution map is imaged with a robust weighting of 0 and tapered with a 14 arcsec Gaussian kernel, with a PSF of 20 20 arcsec, see § 2. The , PSF, and scale bar are shown in the top-left, bottom-left, and bottom-right of the images, respectively. Additionally, we encircle the two BCGs with solid orange lines, the with a dashed red line, 10 bright sources targeted for image-plane masking with dashed black circles, and indicate the ACT SZ detection centroid with a red cross (Hilton et al. 2021). Radio contours are overlaid at [5,10,15] = 7 Jy beam (left) and [5,10,15] = 10 Jy beam (right). While visible in both images, the lower-resolution image clearly shows a diffuse radio halo with an LLS of 1.1 Mpc at 5 significance.
In the left panel of Figure 1, we clearly see diffuse emission associated with Abell 370, traced by the contours. This is expected, given that Abell 370’s merging state results in a dynamically unrelaxed cluster (Buote 2001; Brunetti et al. 2001). As stated in previous studies, the diffuse emission in the left panel appears preferentially centred on the southernmost of the two merging clusters (Xie et al. 2020). After point source subtraction and lower-resolution imaging, the radio halo becomes prominent at both the and the contours throughout the system, revealing significantly more extended and higher signal-to-noise-ratio (SNR) diffuse emission. The LLS, previously measured to be kpc (Xie et al. 2020) or kpc (Kolokythas et al. 2025), is measured in both the full- and lower-resolution images to be Mpc. To mitigate the impact of remaining low-level residuals from the -plane subtraction, we mask point source regions (dashed black circles) as described previously. The point source subtraction appears successful, leaving prominent diffuse emission within the boundary.
Our point source subtraction and lower-resolution imaging have therefore revealed a larger LLS than previously measured. However, a multiwavelength approach, including complementary X-ray observations, is critical to fully understand the ongoing physical processes and to characterise this halo.
Multiwavelength Data
The multiwavelength data available for the Abell 370 field is rich due to it being a Hubble Frontier Field (HFF) galaxy cluster (Lotz et al. 2017). Although this cluster has extensive coverage across various instruments (including the Hubble Space Telescope (HST), Chandra, VLA, and GMRT), this work focuses on the radio continuum and thermal X-ray components (Feretti et al. 2012). We use MeerKAT UHF band observations for our primary analysis and spectral index derivation, comparing our findings to historical VLA and GMRT measurements.
Figure 2 shows an HST RGB image using the F814W, F606W, and F435W filters, on the left with contours from the lower-resolution MeerKAT UHF-band map, as well as the Chandra X-ray map with the same contours overlaid and pixels colourised by count, on the right. The Chandra data comes from OBSID 515 (PI-Garmire) with the ACIS detector and was reduced with the Chandra Data Archive processing pipeline. The data was corrected for time-dependent gain, readout streaks, and background flares and has a final cleaned exposure time was 88 ks. The X-ray map was regridded to a 1.7 arcsec pixel scale and smoothed to match the resolution of the radio map. In both images, the ACT SZ centroid is indicated with a cross. The oval region in the right panel, shown by the teal dashed line, is constructed by approximately tracing the radio emission at . The integrated flux extracted from this region is mJy ( MHz central frequency), where the uncertainties are calculated as the quadrature sum of a per cent absolute flux calibration error and the statistical image noise (), with representing the number of synthesized beams within the extraction aperture.

Figure 2. HST RGB image using the F814W, F606W, and F435W filters (left), and Chandra X-ray map (0.5 to 7.0 keV, right) where the X-ray map is regridded to match the MeerKAT radio map and smoothed by a 12 arcsec Gaussian kernel. The X-ray map is colourised by counts, and we show the spectral index extraction region with a teal dashed ellipse. ACT SZ centroid is indicated with a cross, and radio contours from Figure 1 (right) are overlaid on both images. The 10 and 15 contours trace the X-ray emission closely, while the 5 contour deviates in the southern part of the cluster.
The right panel in Figure 2 shows the diffuse radio emission closely matching the thermal X-ray emission within the teal dashed line region. This morphology provides evidence that Abell 370 conforms to the well-established radio-X-ray spatial correlation (Govoni et al. 2001b). The X-ray and radio morphologies both exhibit a clear elongation in the north-south extent of the cluster. This spatial co-location, combined with the cluster’s merging dynamical state, strongly suggests that the merger-induced turbulence is responsible for the re-acceleration of electrons resulting in synchrotron emission (Brunetti et al. 2001). Notable X-ray point sources are visible in the field, with the most prominent being in the northeast.
The spatial co-location of diffuse radio and thermal X-ray emission provides evidence for the classification of the source as a radio halo. Additionally, this motivates the use of the constructed teal dashed region for flux extraction, to measure a spectral index using our MeerKAT UHF-band data.
Spectral Index
A key property of a radio halo is its spectral index, , which describes the power-law slope of its integrated flux density with frequency (). In this section, we use deep MeerKAT UHF-band observations to derive an in-band spectral index for the diffuse emission, and subsequently compare our measurement with values from the literature.
The integrated spectral index of the radio halo in Abell 370 is shown in Figure 3. Using EMCEE (Foreman-Mackey et al. 2013) we determine spectral indices from the medians of posterior probability distributions from Markov Chain Monte Carlo (MCMC) runs with a model of , where is the flux at GHz, . We run MCMC on in-band flux density from MeerKAT UHF-band images using the teal oval region in Figure 2 to extract flux values. Using seven sub-band images smoothed to a common PSF of arcsec, we obtain a median of . We indicate per cent confidence intervals shaded in red and show the 2D and 1D posterior probability distribution of the spectral index on flux densities are calculated as , where is the number of beams in the teal region and is determined from a source-free region in each sub-band image near the radio halo. These estimates are consistent with the RMS measured within the teal region of the corresponding residual images. We list the sub-band flux densities in Table 1.

Figure 3. The integrated MeerKAT UHF in-band radio spectrum is shown in dark red. Flux density values are extracted within the specified oval region from seven in-band images, where uncertainties are derived as outlined in Sec. 3.3. The red dashed line shows a power law model derived using MCMC Foreman-Mackey et al. (2013), and returns a median posterior of . MCMC confidence intervals () are indicated in shaded regions. In the bottom left, we show an inset of the 2D posterior probability distribution functions. The spectral index value indicates that the radio halo has a steeper spectrum than previously thought.
| Sub-band | Frequency (GHz) | Flux Density (mJy) |
| 1 | 0.578 | 12.30 ± 0.24 |
| 2 | 0.646 | 9.12 ± 0.24 |
| 3 | 0.714 | 8.81 ± 0.27 |
| 4 | 0.782 | 7.63 ± 0.26 |
| 5 | 0.850 | 6.57 ± 0.22 |
| 6 | 0.918 | 5.81 ± 0.12 |
| 7 | 0.986 | 5.22 ± 0.17 |
Table 1. Measured MeerKAT UHF in-band flux densities for the Abell 370 radio halo.
The in-band MCMC probability distribution shows an ultra-steep median spectral index, consistent with within uncertainties (Cassano et al. 2006; Brunetti et al. 2008). Given this steep spectrum, the non-detection at 3 GHz (Xie et al. 2020) is unsurprising due to the faint flux density predicted by the extrapolation. Xie et al. (2020) reports and an upper bound of using the VLA flux upper limit at 3 GHz. These discrepancies could be due to our larger flux extraction region (motivated by a larger measured LLS) and/or the lower fidelity of the GMRT 325 MHz image.
Our discovery of the ultra-steep nature of the radio halo in Abell 370 significantly advances our understanding of this complex merging dynamical system and adds to the growing sample of USSRHs discovered to date (e.g. Magolego et al. 2026), providing evidence in support of the turbulent re-acceleration model. We further plore and investigate this finding in the following sections, where we present the point-to-point radio-X-ray correlation analysis and spatial spectral index map for Abell 370.
Discussion
X-ray-radio Correlation
While the radio-X-ray spatial correspondence is qualitatively apparent in Figure 2, a quantitative correlation can be established by performing a point-to-point analysis of the two maps. Typically, radio halos show positive correlation between radio and X-ray surface brightness, , where is the radio surface brightness at 1.4 GHz and is the X-ray surface brightness, generally with a sub-linear slope, (Govoni et al. 2001a; Feretti et al. 2001; Giacintucci et al. 2005; Vacca et al. 2010; Hoang et al. 2019; Magolego et al. 2026). To explore this correlation, we regrid the Chandra X-ray map to that of the radio image, then smooth it with a 12 arcsec Gaussian kernel.
We investigate the X-ray radio correlation in Figure 4 by plotting radio surface brightness as a function of X-ray surface brightness for a set of regions covering the MeerKAT UHF-band radio halo. X-ray point sources are masked out and excluded from this analysis. Uncertainties on radio surface brightness markers are calculated as the standard deviation of the pixel values within each extraction region. We use a power law of the form as our model. We filter the X-ray data to only include the soft energy band (0.5 − 2 keV) before proceeding with the analysis. X-ray surface brightness values are calculated as the sum of the pixel values normalised by the exposure time and physical area of the extraction region. Uncertainties on these values are derived using the square root of the variance (including the background counts) which is then normalised. The data points are further restricted to regions with a radio SNR . We fit the aforementioned model to the data using Bayesian MCMC linear regression, finding . The quoted uncertainties represent the standard deviation of the posterior, while the shaded regions indicate the intrinsic scatter.

Figure 4. Radio to X-ray surface brightness correlation between MeerKAT UHF and Chandra maps. Teal markers indicate the surface brightness measurements extracted across the cluster. The corresponding power-law fit, derived using Bayesian MCMC linear regression, is overlaid as a red dashed line. The lightly shaded red region represents the intrinsic scatter of the model. A radio inset is shown with the extraction regions overlaid in teal. The relation demonstrates a clear, statistically significant positive correlation.
The power-law slope of is consistent with literature region-based and pixel-by-pixel correlation analyses of radio halos (Govoni et al. 2001b; Feretti et al. 2001; Giacintucci et al. 2005; Brown & Rudnick 2011; Rajpurohit et al. 2018; Magolego et al. 2026). We find a Pearson correlation coefficient of and a Spearman correlation coefficient of , both with , indicating a statistically significant, moderate-to-strong positive correlation. We have therefore qualitatively confirmed the visual correlation from Figure 2 with this region-based point-to-point correlation, confirming the physical connection between radio emission and thermal X-ray emission. However, the synchrotron emission itself is better explored with a spatial spectral index map.
Spectral Index Map
Previous studies of Abell 370 have not had sufficiently high SNR and image fidelity to create spectral index maps of the radio halo (Xie et al. 2020). Here we present the spectral index map of the radio halo halo in Abell 370, demonstrating the high image fidelity of these MeerKAT UHF-band data.
We generate the spectral index map using Broadband Radio Astronomy ToolS (BRATS Harwood et al. 2013) in Figure 5. We use seven sub-band images smoothed to a common PSF of arcsec and an SNR level in BRATS of sigma=3 within the teal dashed region. The default "weighted least squares" fitting method is selected with the specindexcalctype command, together with a 15 per cent flux calibration error on all seven sub-band images. In Figure 5, pixel values are colourised by spectral index, and contours are from the Chandra X-ray map.

Figure 5. MeerKAT UHF in-band spectral index map of Abell 370 with X-ray contours (, , and ) and a teal dashed line indicating the spectral index extraction region. Image plane subtracted radio point sources are indicated with dashed black circles. We generate the map using BRATS (Harwood et al. 2013) with parameters outlined in § 4.2 and seven in-band radio maps smoothed to a PSF of arcsec. The pixels are colourised by the spectral index, saturated at and . The majority of the radio halo shows a spectral index between -1 and -2, with a mean of -1.40.
The BRATS spectral index map has a mean pixel value of , which is broadly consistent with our integrated in-band spectral index of derived in § 3.3 using the same smoothed sub-band images. While we lack sufficient SNR to fully populate the radio contours from Figure 1, the core of the halo is reliably mapped. The map is relatively free of discrete point sources, demonstrating the success of our subtraction procedure. However, given how faint the diffuse radio emission is, deeper integration times are required to fully populate the spectral index map. A more complete map would enable a detailed, spatially resolved study of the radio–X-ray connection (Botteon et al. 2018).
Scaling Relations
The radio power-mass () scaling relation is a tight correlation, with low scatter, linking the mass of the cluster to the
diffuse radio emission it hosts (Cassano & Brunetti 2005). While typical radio halos in galaxy clusters seldom deviate from this relation, USSRHs are known to be underluminous and lie below it (Cassano et al. 2013; Cuciti et al. 2021b). In this section, we contextualise the USSRH we find in Abell 370 by plotting its position on this relation. This allows us to compare it with known USSRHs and typical radio halos.
To standardize our measurement for literature comparison, we scale the observed UHF flux to a rest-frame frequency of 1.4 GHz using our derived spectral index of . We then calculate a radio power using:
where is the luminosity distance to the halo, and the rest are as previously defined. This yields . The position of Abell 370 in the context of the radio power at 1.4 GHz () - mass () scaling relation is investigated in Figure 6. Abell 370 measurements are plotted for this work in red, from Kolokythas et al. (2025) in light green, and from Xie et al. (2020) in orange. We use the weak lensing corrected mass from the ACT catalogue (Hilton et al. 2021), , while the latter literature measurements use the mass from Planck Collaboration et al. (2016), . We differentiate the masses with marker shape: diamond for ACT and circular for Planck. Note that the uncorrected mass, from ACT is more consistent with the mass from Planck Collaboration et al. (2016), however, motivated by the well constrained mass modeling of Abell 370. We plot a literature compilation of USSRHs and candidate USSRHs (cUSSRH) in dark and light purple, respectively (Cuciti et al. 2021a; Edler et al. 2022; Duchesne et al. 2021; Magolego et al. 2026; Rajpurohit et al. 2023; Venturi et al. 2022; Pasini et al. 2022), as well as a non-exhaustive sample of radio halos in black and grey (Kolokythas et al. 2025; Cuciti et al. 2021a). Over-plotted, we show the BCES Y|X fit by Cuciti et al. (2021b) for radio halos only in their extended sample, with coefficients of and . The 95 per cent confidence interval is calculated as outlined in Cuciti et al. (2021a) and shown with grey shading. For literature sources where 1.4 GHz measurements are unavailable, we first extrapolate the nearest quoted flux density to 1.4 GHz using the standard relation . We then calculate the rest-frame radio power using equation 1.

Figure 6. Radio power at 1.4 GHz () as a function of cluster mass () for radio halos and USSRHs. The red marker shows the MeerKAT UHF-band Abell 370 radio halo studied in this work, while the light green and orange markers indicate the MeerKAT L-band (Kolokythas et al. 2025) and VLA + GMRT (Xie et al. 2020) Abell 370 halo measurements, respectively. We denote the use of an ACT mass (Hilton et al. 2021) with a diamond marker, while Planck masses (Planck Collaboration et al. 2016) are shown with circular markers. Dark and light purple markers indicate USSRHs and candidate USSRHs, respectively, compiled from various works (Cuciti et al. 2021a; Duchesne et al. 2021; Edler et al. 2022; Pasini et al. 2022; Venturi et al. 2022; Rajpurohit et al. 2023; Magolego et al. 2026). Grey and black markers indicate radio halos from Kolokythas et al. (2025) and Cuciti et al. (2021a), respectively. The black dashed line represents the "RH only" BCES Y|X fit from the extended sample reported in Cuciti et al. (2021b) (=2.67 and =0.07), with 95 per cent confidence intervals shown with shading. The radio halo in Abell 370 falls below this line and within the scatter of USSRHs and candidate USSRHs.
Figure 6 shows that USSRHs and cUSSRHs are preferentially found below the scaling relation fit shown here (extended sample and RHs only), and indeed below other fits presented in Cuciti et al. (2021b). All but two cUSSRHs (Cuciti et al. 2021a) and one USSRH (Edler et al. 2022) lie below the 95 per cent confidence region of this relation. Our own measurement, and two literature measurements (Xie et al. 2020; Kolokythas et al. 2025), all place Abell 370 below this fit and outside the 95 per cent confidence region if uncertainties are taken into account. If a Planck-derived mass is adopted, our measurement of Abell 370 aligns closely with the findings of Xie et al. (2020). While the halo is then within uncertainties of the 95 per cent confidence interval of the BCES Y|X fit (Cuciti et al. 2021b), it remains relatively underluminous. Crucially, these results stem from three independent flux measurements and two independent mass measurements.
We confirm that in the radio power-mass scaling relation, USSRHs occupy a different parameter space than typical RHs. The fact that Abell 370 is underluminous and lies well within the scatter of this USSRH population provides strong supporting evidence for its ultra-steep spectrum nature discussed in § 3.
Classification
With evidence from the integrated spectral index (Figure 3), the spectral index map (Figure 5), and the location of Abell 370 on the scaling relation (Figure 6) we can now confidently classify the radio halo in Abell 370 as an USSRH. Before presenting this final classification, we perform a qualitative cross-check between MeerKAT observations to validate our in-band spectral indices and absolute flux calibration.
The purpose of this exercise is to qualitatively demonstrate spectral index consistency across MeerKAT UHF-band and L-band. Figure 7 plots the integrated spectral index of eight radio point sources that are confirmed manually to be free from source confusion in the Abell 370 field of view. We use seven in-band MeerKAT UHF-band images as well as the 12 sub-band L-band images from MGCLS archival data2 (Knowles et al. 2022) and perform the same MCMC sampling on these data. The resulting mean spectral index for these sources, differentiated by colour, is for the UHF-band and for the L-band. This agreement was achieved after correcting for an initial systematic offset by applying the updated flux calibrator model from the MeerKAT documentation3.

Figure 7. MeerKAT UHF and L-band spectra of six isolated radio point sources in the Abell 370 field, with each colour representing a source. We use MCMC (Foreman-Mackey et al. 2013) to fit a power law and derive a spectral index using seven smoothed in-band UHF-band images and 12 in-band L-band images separately. A fractional bias is apparent between the L-and UHF-band absolute calibration, however, this lies within the observatory-derived uncertainties.
As shown in Figure 7, the derived spectral indices exhibit consistency () and successfully recover the canonical point-source spectral index of (Condon 1992). This, coupled with the low scatter in the derived spectral indices, reinforces the reliability of both the UHF- and L-band MeerKAT measurements. While we note a slight amplitude offset, likely arising from differences in visibility weighting and tapering, this merely shifts the normalisation without affecting the measured slopes or the scientific conclusions of this work.
Taken together, our analysis provides multiple lines of evidence to classify the diffuse emission in Abell 370. First, its LLS of 1.1 Mpc and strong spatial correlation with the thermal X-ray emission (see § 3.1 and § 4.1) confirm its nature as a radio halo (Govoni et al. 2001b). We further refine this classification based on our in-band spectral index () and the source’s position within the USSRH population on the - scaling relation (Fig. 4.3), as we classify this source as an USSRH. This refines its previous classifications from a radio halo (Xie et al. 2020) or a candidate radio halo (Knowles et al. 2022) to a confirmed USSRH.
Conclusions
In this work, we use MeerKAT UHF-band data, from a project aimed at high- lensed HI emission, to deliver commensal diffuse radio halo science. We uncover a significantly larger spatial extent of diffuse radio emission in Abell 370 with deep MeerKAT observations and meticulous point source subtraction. The derived spectral index is consistent with an USSRH. We find a strong qualitative and quantitative radio-X-ray correlation within the halo, and confirm the ultra-steep spectrum with a spectral index map and the power-mass scaling relation. Our main findings are as follows:
(i) Our tapered imaging reveals a radio halo with an LLS of Mpc, larger than previously measured. Additionally, we find radio extensions to the southeast and southwest. (ii) We find an in-band integrated spectral index of , consistent with an USSRH.
(iii) The dynamically unrelaxed radio halo exhibits a strong radio-X-ray correlation of , supporting the turbulent re-acceleration model.
(iv) We create a spectral index map of the radio halo in Abell 370, confirming its ultra-steep-spectrum.
(v) In the context of the power-mass scaling relation for radio halos, Abell 370 is positioned within the scatter of known USSRHs, providing further evidence for its classification.
This work showcases the compelling commensal radio continuum science possible with deep MeerKAT programs focused on targeting massive clusters in order to detect gravitationally lensed H I sources in the distant universe.
ACKNOWLEDGEMENTS
We thank the anonymous referee for a careful review and insightful feedback that helped improve this work. We would like to thank the staff of the South African Radio Astronomy Observatory (SARAO4) who made these observations possible. We thank Isaac Magolego and Dakalo Phuravhathu for insightful conversations and assistance in data reduction and point source subtraction. GDL and RPD acknowledge funding from the South African Radio Astronomy Observatory (SARAO), which is a facility of the National Research Foundation (NRF), an agency of the Department of Science, Technology and Innovation (DSTI). RPD acknowledges funding by the South African Research Chairs Initiative of the DSI/NRF (Grant ID: 77948). The MeerKAT telescope is operated by the South African Radio Astronomy Observatory, which is a facility of the National Research Foundation, an agency of the Department of Science and Innovation. We acknowledge the use of the ILIFU5 cloud computing facility - www.ilifu.ac.za, a partnership between the University of Cape Town, the University of the Western Cape, Stellenbosch University, Sol Plaatje University, the Cape Peninsula University of Technology and the South African Radio Astronomy Observatory. The ILIFU facility is supported by contributions from the Inter-University Institute for Data Intensive Astronomy (IDIA - a partnership between the University of Cape Town, the University of Pretoria and the University of the Western Cape), the Computational Biology division at UCT and the Data Intensive Research Initiative of South Africa
(DIRISA). This work has made use of the Cube Analysis and Rendering Tool for Astronomy (CARTA, Comrie et al. 2021). This research made use of Astropy6, a community-developed core Python package for Astronomy. This research makes use of the semi-automated MeerKAT data calibration pipeline, oxkat7.
Data Availability
The raw visibilities used in this work are from project ID SCI-20241101-GL-01 on the MeerKAT archive8. The authors may make raw visibilities and data products available upon reasonable request.
This paper has been typeset from a TeX/LaTeX file prepared by the author.
References
- [42]Józsa G. I. G., et al., 2020, CARAcal: Containerized Automated Radio Astronomy Calibration pipeline, Astrophysics Source Code Library, record ascl:2006.014 (ascl:2006.014)
- [55]Peebles P. J. E., 1980, The large-scale structure of the universe Planck Collaboration et al., 2016, A&A, 594, A27
- [68]van Weeren R. J., Intema H. T., de Gasperin F., Tonk J. B. R., Röttgering H. J. A., Clarke T. E., 2009, A&A, 508, 1269
- [69]van Weeren R. J., de Gasperin F., Akamatsu H., Brüggen M., Feretti L., Kang H., Stroe A., Zandanel F., 2019, Space Sci. Rev., 215, 16