The VLA Frontier Fields Survey: Deep, High-resolution Radio Imaging of the MACS Lensing Clusters at 3 and 6 GHz

Four images were produced for each cluster, 12 in total. These are a high-resolution S-band image (S-HIGH), a second S-band image tapered to provide an approximately 3'' synthesized beam (S-LOW), a full-resolution C-band image (C-HIGH), and finally, a lower-resolution C-band image (C-LOW) designed to match the angular resolution of the S-HIGH image. This is mainly for the determination of matched-resolution component flux densities and brightnesses, and subsequently, the determination of component spectral indices. The main properties of each image are summarized in Table 2. The S-HIGH images are shown in Figure 1, and Figure 2 shows contours of the S-HIGH image overlaid on a three-color HST image for MACS J0717. Further details on the optical catalogs are provided in Section 3.3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The "master" map for component extraction is taken to be the S-HIGH image for each cluster, as this provides both a higher surface density of detections and a larger area compared to the C-band images. The first step involved running the PyBDSF (Mohan & Rafferty 2015) source finder on this image using a peak threshold of 5σ and an island threshold of 3σ, where σ is taken to be the source finder's own estimate of the background noise level. This is determined by stepping a box across the image and computing the standard deviation of the pixels within that region, and the resulting position-dependent values are recorded and interpolated. This produces a background noise image that is used for local estimates of σ. Peaks above five times this value are identified, and then a flood-fill method is used to delineate regions of contiguous emission down to the secondary 3σ island threshold. Regions of significant emission are then decomposed into a series of points and Gaussian components. Default box size settings were used in each case. This procedure was performed on each of the three S-HIGH images.

Following this, the resulting catalogs were manually examined to remove spurious components from image artifacts. These were primarily associated with brighter sources away from the pointing center. Components belonging to sources with extended morphologies were also visually identified and placed into a separate catalog (see Section 3.7 and Appendix D). Following these steps, we obtain a catalog containing only the compact S-band radio sources for each cluster.

Determining optical identifications for the radio sources relies on three existing resources for each cluster. The first two of these is a set of HST photometric catalogs created by the ASTRODEEP ¹³ project, by Castellano et al. (2016) for MACS J0416, and by Di Criscienzo et al. (2017) for MACS J0717 and MACS J1149, as well as the more recent catalogs from the DeepSpace project (Shipley et al. 2018). These catalogs also include spectroscopic redshifts where available, from the observing campaign by Ebeling et al. (2014). The third resource consists of the source catalogs derived from wider-area imaging with the Subaru telescope by the CLASH team, complete with photometric redshifts as presented by Medezinski et al. (2013) for MACS J0717 and Umetsu et al. (2014) for MACS J0416 and MACS J1149.

A cutout image around each compact component was created. The radio contours were overlaid on either the HST F140W image or the Subaru z-band image if the region was outside the HST footprint. Entries from either the ASTRODEEP HST or Subaru catalogs were overlaid on the cutout image, and visual inspection determined whether or not a cataloged optical/near-infrared association for the radio component was present, based on positional and morphological considerations. Note that we validate the astrometric alignment between radio and optical in Appendix A, and the cross-matching process is unaffected by this. In addition to verifying potential hosts for the radio components, manually inspecting each of the compact components also allowed flags to be applied to sources that (i) could be well represented by single Gaussian components but had been decomposed into several components by PyBDSF, and (ii) had significantly extended structures associated with them that had been missed in the first pass. Components belonging to the first category were force-fit with a single Gaussian using the CASA imfit task, and the multiple entries belonging to this source in the component catalog were replaced by this single component. Radio components belonging to the second category were removed from the compact catalog and placed into the extended catalog. Note that this process only cross-matched radio components with cataloged optical sources from the HST or Subaru catalogs.

At this stage, the compact S-band catalog for each cluster exists in two subsets: those with optical/near-infrared matches (the "radio-optical" catalog) and those without (the "radio" catalog). Relevant properties derived from the HST observations (e.g., optical IDs, positions, stellar mass and star formation rate estimates, spectroscopic and photometric redshifts) are merged with the radio-optical catalog, using the more recent catalogs from the DeepSpace project (Shipley et al. 2018), using positional coincidence matches between these catalogs and the ASTRODEEP catalogs. Additional entries present in the DeepSpace catalogs that were not visually matched as part of the ASTRODEEP process were assigned to radio components by conducting nearest-neighbor cross-matching for the additional sources that the DeepSpace catalogs contain. Following, e.g., Ivison et al. (2007), the match radius was set according to

$$\begin{eqnarray}&&{{\rm{\Delta }}}_{{\rm{R}}.{\rm{A}}.}={{\rm{\Delta }}}_{\mathrm{Decl}.}=0.66\displaystyle \frac{\theta }{{\rm{S}}/{\rm{N}}},\end{eqnarray} \tag{ 1 }$$

which predicts how Δ_R.A. and Δ_Decl. (the scatter in R.A. and decl. of the radio component) scale with the signal-to-noise ratio (S/N) of the detection for a synthesized beam with an FWHM of θ. Evaluating Equation (1) for S/N = 5 for the S-HIGH images (according to the noise values reported in Table 2) results in match radii of 022, 016, and 012 for MACS J0416, MACS J0717, and MACS J1149, respectively. Note that prior to cross-matching, the offsets described in Appendix A were removed to further align the cataloged positions. A complete description of the columns for both the radio-optical and the radio catalog, including the properties derived in the sections that follow, is provided in Appendix C. Stellar masses, star formation rates, and specific star formation rates from DeepSpace are included in the radio-optical catalog where appropriate. For a detailed analysis of radio versus optical star formation rates, we refer the reader to the companion paper to this work by Jiménez-Andrade et al. (2021).

Although PyBDSF provides the deconvolved source sizes, the associated uncertainties in the major and minor, and position angles that are returned are the same as the apparent sizes in the map plane (at the time of writing). We thus estimate the intrinsic source (major axis) sizes and the associated uncertainties according to the method presented in Appendix B.

In addition to the component peak brightnesses and integrated flux densities measured by the source finder, we also provide a "best" flux density estimate (S_*) based on an assessment of how reliably resolved the Gaussian components are. This largely follows the method adopted by Murphy et al. (2017), modified for the considerations above that cater for the fact that the restoring beam in our maps was not forced to be circular. If the criterion

$$\begin{eqnarray}&&{\phi }^{{\prime} }-{\theta }_{\mathrm{beam}}^{{\prime} }\,\geqslant 2{\sigma }_{\phi }^{{\prime} }\end{eqnarray} \tag{ 2 }$$

is satisfied, then we deem a source to be reliably resolved, where ${\theta }^{{\prime} }$ is the FWHM of the elliptical beam as projected along the source major axis, and ${\phi }^{{\prime} }$ and ${\sigma }_{\phi }^{{\prime} }$ are the source major axis and associated uncertainty as measured by PyBDSF. In this case, S_* is set to the integrated flux density of the component as determined by the source finder. For sources that are marginally resolved (those that do not satisfy the Equation (2) criterion), we set S_* to be the geometric mean of the integrated flux density and peak brightness. The uncertainties in these two quantities are anticorrelated when fitting Gaussians, and hence, the geometric mean provides the best flux density measurement for a Gaussian fit to a faint component (Condon 1997).

The PyBDSF source finder was also run on the C-LOW images, and the results were cross-matched with the "radio-optical" and "radio" catalogs described above. The C-LOW image matches the S-HIGH images in terms of angular resolution, making for more robust flux density comparisons for compact features between the two frequencies, and thus for more reliable spectral index estimates. All components in the S-band catalog that had a C-band match within a radius of 1'' were visually examined (via S-band contours over the C-band image) to confirm that the C-band component was neither spurious nor fragmented into multiple components. Genuine C-band matches were then associated with the relevant S-band components. Spectral index (α) ¹⁴ estimates for compact components were derived from the dual-frequency "best" flux density measurements (S_*) and added to the catalog.

The Frontier Fields have publicly available lensing models, ¹⁵ derived independently by five different teams. Further details are given by Lotz et al. (2017), with the mapping teams making use of data from Schmidt et al. (2014), Vanzella et al. (2014), Diego et al. (2015), Merlin et al. (2015, 2016), Castellano et al. (2016), Jauzac et al. (2016), Kawamata et al. (2016), Limousin et al. (2016), and Caminha et al. (2017).

The models take the form of images that capture the spatial variation in mass surface density (κ) and lensing shear (γ) over each cluster. The magnification (μ) for a point-like source at a given redshift (z_s ) behind the lensing cluster (at redshift z_l ) can be modeled using these two parameters via the relationship

$$\begin{eqnarray}&&1/\mu ={\left(1-{\kappa }_{z}\right)}^{2}\,-\,{\gamma }_{z}^{2},\end{eqnarray} \tag{ 3 }$$

where κ_z and γ_z are the model values for mass surface density and shear scaled by the ratio D_ls /D_s , where D_ls is the angular diameter distance between the cluster and the source, and D_s is the angular diameter distance from redshift zero to the source.

Magnification estimates were determined for radio components that have either spectroscopic or photometric redshifts by evaluating Equation (3), using the source redshift estimate. The preferred order of redshifts when calculating lensing is (i) spectroscopic redshifts from the DeepSpace catalogs, (ii) photometric redshifts from the DeepSpace photometry fitting as derived using the EAZY code (Brammer et al. 2008), and (iii) Bayesian photometric redshifts from the CLASH catalogs. This process was repeated for each lensing model, and the median magnification over all available models is evaluated and recorded in the catalog. Median absolute deviations of μ across all available models are also tabulated in order to provide some measure of the magnification uncertainty. Note that when extracting the values from the maps of κ and γ, the position of the radio component was used, not the position of the optical host, naturally leading to some differences between the magnifications in the radio-optical catalog and those derived from the DeepSpace catalogs. Note that we do not track inverted parity (μ < 0) sources in the catalog as we only present the median absolute value over all available lensing models.

As mentioned in Section 3.2, the visual confirmation of optical hosts in conjunction with the component models derived by PyBDSF allowed a thorough identification of radio sources with extended morphologies. These features were subdivided into three categories: (i) extended radio sources with cataloged optical hosts, (ii) extended radio sources with no cataloged optical host, and (iii) diffuse structures (e.g., relics and halos) that are likely arising due to processes in the foreground clusters. Optical IDs were associated with the first category via the same method as for the compact components, and categories (i) and (ii) were distinguished from (iii) using morphological and optical-ID considerations.

In order to determine the integrated flux density of the extended sources, we employ the ProFound software package (Robotham et al. 2018). Designed primarily with large optical surveys in mind, the software has been found to provide superior photometric estimates to other common source-finding packages when applied to both simulated and real radio interferometer data (Hale et al. 2019). ProFound does not perform component fits to regions of extended emission, instead adopting an approach whereby thresholded regions are iteratively dilated until the surface brightness measurement converges.

Cutout images for each extended region from both the S-LOW and S-HIGH images were processed using ProFound, the former also being used to better gauge the total integrated flux density for sources that may be partially resolved out by the A-configuration observations. The resulting catalogs from ProFound were pruned of any measurements of additional sources in the field in order to only derive properties from the extended source being considered.

Optical/near-infrared and radio contour composite images, notes on individual extended sources, and their tabulated radio properties are presented in detail in Appendix D. Note that the well-studied foreground Fanaroff–Riley Type-1 (FR-I) source in the MACS J0717 field and the narrow-angle-tail (NAT) galaxy embedded in the radio halo are not included.

There are 1262/55 radio sources with a photometric/spectroscopic redshift across the three fields, including the host galaxies of radio sources with complex morphologies. The distribution of these redshifts is shown in Figure 3, with photometric measurements shown in gray and spectroscopic measurements shown in red. The median redshift measured from the cataloged sources from all three clusters is 0.88. This is lower than generally reported from other radio-selected deep-field studies (e.g., Smolčić et al. 2017); however, the overdensities of galaxies in the clusters themselves will pull this median value down. On a per-cluster basis, the median redshifts are 0.9, 0.75, and 1.03 for MACS J0416, MACS J0717, and MACS J1149, respectively. If we exclude sources within ±0.05 of the cluster redshift, the median redshifts of the radio-selected objects become 0.92, 0.85, and 1.13, having excluded 10, 58, and 38 objects from MACS J0416, MACS J0717, and MACS J1149, respectively. There appear to be significantly more sources out to z ∼ 1.5 in MACS J1149, and the detections in this field appear to be less clustered along the line of sight (see also Figure 7). Despite the limiting magnitudes for the Subaru observations being the same, the surface density of strongly lensed sources varies by a factor of >2 across the three fields (Umetsu et al. 2014). This suggests that sample variance is the likely explanation for the boxy (or possibly bimodal) redshift distribution of the radio sources in MACS J1149.

Zoom In Zoom Out Reset image size

Accurate radio spectral index (α) measurements are essential for deriving rest-frame quantities such as spectral luminosity and, by extension, parameters such as star formation rates (e.g., Jiménez-Andrade et al. 2021). The use of measured values of α as opposed to adopting characteristic or canonical values (typically assumed to be −0.7 to −0.8 for synchrotron emission) can significantly reduce the scatter and biases in studies involving the radio/far-infrared correlation, as demonstrated by Gim et al. (2019), who also demonstrate the importance of deriving values of α from dual-frequency images that are resolution-matched.

Figure 4 shows the 3 GHz integrated flux densities of the compact radio sources against their 3–6 GHz spectral index. Note that the plot only shows sources that have a C-band detection within the 50% gain region of the VLA C-band primary beam at 6 GHz (for a total of 169 objects). Within this region, both the C-band and especially S-band primary beam responses do not impart significant attenuation, and thus, the selection biases introduced by the frequency-dependent antenna primary beam pattern over the broad bandwidths of both observing bands are reduced. Selection biases persist, however, introduced by the unknown distribution of source spectral indices and the slightly different depths of the S-HIGH and C-LOW images, which are used to measure α for the cataloged sources.

Zoom In Zoom Out Reset image size

We investigate the selection function imposed on the S–α plot using a Monte Carlo simulation. The spectral index range −2 to +2 is partitioned into 400 bins (width 0.01). For each of these bins, a source with the corresponding spectral index is created and assigned a peak brightness at 3 GHz. The corresponding peak brightness at 6 GHz is calculated according to the assigned value of α. Both the 3 and 6 GHz values are then perturbed with noise drawn from a Gaussian distribution with a mean of zero and a standard deviation corresponding to the lowest S-HIGH (0.7 μJy beam⁻¹) and C-LOW (1.0 μJy beam⁻¹) rms noise values across the three clusters, as listed in Table 2. If the noisy simulated brightness measurements at 3 and 6 GHz exceed five times the rms noise values in both bands, then the simulated source is considered to be detected and in the real-world case would have a legitimate α measurement in the catalog. Because we are interested in the sources we may be missing from Figure 4, the above procedure is repeated with a steadily declining 3 GHz brightness. When the simulated source does not meet the peak brightness detection criteria, i.e., either the 3 or 6 GHz brightness drops below the threshold, the source is considered undetected and the peak brightness and α values are noted. This process is repeated 400 times for all 400α bins, and the resulting distribution of simulated sources that have dropped below the detection threshold is plotted as the red cloud of points in Figure 4, where the density of sources increases from dark red to white.

The fact that there is a detection criterion at both 3 and 6 GHz, but the plot shows only the 3 GHz measurements, imparts two distinct regions to this distribution. Sources to the left of the distribution have steep spectra that render them too faint for detection at 6 GHz, but the corresponding full range of noisy 3 GHz brightness measurements is present. Sources to the right of the distribution have inverted spectra that cause them to drop below the 3 GHz detection threshold, delineated by the hard upper limit on the right of the distribution. Essentially we are seeing the manifestation of Eddington bias as it applies to a combination of two peak brightness-limited samples. Our spectral index selection function is thus only complete for −2 < α < 2 above a peak 3 GHz brightness of 30 μJy beam⁻¹.

The median spectral index of the 169 objects plotted in Figure 4 is −0.5; however, for the 74 sources with peak brightnesses above 30 μJy beam⁻¹ at 3 GHz, the median value of α is −0.63. For the canonical synchrotron radiation spectral index of −0.7, our peak S-band completeness limit is equal to 50 μJy beam⁻¹ at 1.4 GHz. Our median value is consistent with previous dual-band studies at comparable depths (e.g., Huynh et al. 2015; Gim et al. 2019; Huynh et al. 2020), as well as in-band measurements made using the 1–2 GHz L-band receivers of the VLA (Heywood et al. 2016, 2020). We note also that the dual-band studies involving measurements at ∼5 GHz and above also revealed a significant fraction of flat and inverted spectrum sources in addition to those with typical synchrotron spectra, something that is also evident in Figure 4.

Figure 5 shows a histogram of the lensing magnification estimates, the derivation of which is given in Section 3.6. As with Figure 3, sources with spectroscopic redshifts are marked in red. We detect 13 objects with magnification factors greater than 2, 7 of which have spectroscopic redshifts. Figure 6 shows the cutouts for each of these sources. The ID, redshift, and median lensing magnification are given above and on the numbered panel for each source. The background of each image is a three-color rendering of the F814W, F606W, and F435W HST filters overlaid with the S-HIGH contours, with contour levels provided in the caption.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The majority of these sources are blue star-forming galaxies with spiral or disturbed morphologies and prominent dust lanes. The mean redshift is 1.73 (standard deviation 0.81). For each of the radio components shown in Figure 6, we provide demagnified, intrinsic values for the integrated flux density, peak brightness, and effective noise levels in Table 3. The source with the highest magnification factor (μ_median = 6.45) is VLAHFF-J071736.66+374506.4, which has an intrinsic peak brightness of 0.9 μJy beam⁻¹ at an effective noise level of 140 nJy beam⁻¹. This makes it a candidate for the faintest radio source detected to date (see also Jackson 2011). The total star formation rate for this source is modest, at 10 M_⊙ yr⁻¹ (for more details, see Jiménez-Andrade et al. 2021).

Table 3. Magnifications and Demagnified Integrated Flux Densities, Peak Brightnesses, and Effective Noise Levels for the 13 Lensed Sources Presented in Figure 6

Panel	ID	μ	${S}_{\mathrm{int}}^{\mathrm{demagnified}}$	${S}_{\mathrm{peak}}^{\mathrm{demagnified}}$	σdemagnified
			(μJy)	(μJy beam−1)	(μJy beam−1)
1	VLAHFF-J041606.36–240451.2	3.03	6.71	4.76	0.33
2	VLAHFF-J041606.62–240527.8	2.26	7.75	5.06	0.44
3	VLAHFF-J041611.67–240419.6	2.26	5.75	2.91	0.44
4	VLAHFF-J071725.85+374446.2	2.21	4.04	3.03	0.41
5	VLAHFF-J071730.65+374443.1	2.84	2.44	1.85	0.32
6	VLAHFF-J071733.14+374543.2	2.11	8.43	3.1	0.43
7	VLAHFF-J071734.46+374432.2	5.84	7.57	1.14	0.15
8	VLAHFF-J071735.22+374541.7	3.61	4.61	4.56	0.25
9	VLAHFF-J071736.66+374506.4	6.45	1.96	0.9	0.14
10	VLAHFF-J071740.55+374506.4	2.18	4.17	3.78	0.41
11	VLAHFF-J114932.03+222439.3	2.11	9.99	3.04	0.43
12	VLAHFF-J114934.46+222438.5	2.16	6.05	3.61	0.42
13	VLAHFF-J114936.09+222424.4	3.13	4.26	2.11	0.29

Note. The first column refers to the panel number for each source in that figure.

Download table as:  ASCIITypeset image

We note some differences between our lensed source catalog and that of van Weeren et al. (2016). The principal reason for this is our use of median rather than mean magnifications. The former is far more robust to strong outliers that may be present in the range of lensing models used. For example, when using mean magnifications for our catalogs, one source had an average magnification factor of 150. Examination of the nine individual magnification values from each of the models revealed one model predicting a magnification factor of 1305.8, whereas the median of the other eight values was 6.22. The use of median values also results in fewer sources with μ > 2.

Here we bring together the results of Sections 4.1–4.3 to compute the demagnified 3 GHz radio luminosities for the compact radio components. These are plotted against redshift for each cluster in Figure 7. The vertical lines show the redshifts of the foreground clusters in each panel, and the dashed line shows the 5σ detection limit based on the noise measurements of the S-HIGH images as listed in Table 2. Measured spectral index values are used where available, and our median value of −0.63 is used otherwise. The preferential order for redshifts is as before, namely (i) DeepSpace spectroscopic redshifts, (ii) DeepSpace photometric redshifts, and (iii) CLASH photometric redshifts. As elsewhere, the last dominate the counts. The magnification corrections demonstrate that we are detecting high-redshift galaxies below the formal detection threshold, although the extreme outlier visible in the MACS J0717 field is likely the result of an improper photometric redshift fit. Figure 7 also shows evidence for more distant clustering in redshift along the line of sight, particularly in the MACS J0416 field.

Zoom In Zoom Out Reset image size

The MACS J0416 luminosity plot reveals a source at z = 4.06 (VLAHFF-J041559.99–240132.5), with a 3 GHz rest-frame luminosity of 4.1 × 10²⁵ W Hz⁻¹. The 3–6 GHz spectral index of this source is steep (or ultra-steep by some definitions, α = −1.13 ± 0.05). This is a characteristic known to be an effective method to search for powerful radio sources at high redshift, although the physical explanation is not conclusively understood (e.g., Singh et al. 2014). The source is deemed to be resolved, although it is well characterized by a single Gaussian component, and there is no evidence for large-scale jet emission in either the S-HIGH or S-LOW images. This source is thus likely a newly discovered powerful high-z radio galaxy and worthy of follow-up observations.

The C-HIGH images with angular resolutions of ∼03–05 will offer the best means for estimating intrinsic source sizes from this project. To examine this, we present the distribution of the FWHM deconvolved source major axes in the upper panel of Figure 8. The median deconvolved source size (θ_M; for entries in the catalog that have nonzero values) is 027 ± 025. For the assumed cosmology, the radio components with associated redshift measurements have a median physical size of 1.9 ± 1.2 kpc. The corresponding distribution in physical units is shown in the lower panel of Figure 8.

Zoom In Zoom Out Reset image size

Higher angular resolution observations reaching ∼0.6 μJy beam⁻¹ have been made of the GOODS North field at the X band (10 GHz; Murphy et al. 2017). This study also used long-track A- and C-configuration VLA observations, resulting in a measurement of 〈θ_M〉 = 0167 ± 0032 (with an rms scatter of 091), corresponding to a median linear size of 1.3 ± 0.28 kpc (with rms scatter 0.79 kpc). This was shown to match both the sizes of the dust emission regions measured at submillimeter wavelengths and the sizes measured in extinction-corrected Hα imaging.

Although consistent with the 1σ measurement errors, and despite the high angular resolution of the C-HIGH images, our median linear size is larger than that measured by Murphy et al. (2017). This suggests that even half-arcsecond resolution is suboptimal for robust size measurements of high-redshift star-forming galaxies at radio wavelengths, although a larger angular size is expected to be seen at lower radio frequencies due to cosmic-ray propagation effects. For a more detailed analysis of radio source sizes using our data, we again refer the reader to Jiménez-Andrade et al. (2021). The interferometers will not resolve out significant emission on these scales due to insufficient short spacings. Ignoring AGN contamination, the radio emission at both 6 GHz and 10 GHz should be arising from the star-forming regions. Given that the angular resolution of both of these studies exceeds that of Band 2 of SKA-MID (0.95–1.76 GHz), X-band studies with the A configuration of the VLA are likely the best path to such intrinsic size measurements prior to the arrival of the Next-Generation VLA (ngVLA; Murphy et al. 2018; McKinnon et al. 2019).

Figure 9 shows the apparent magnitude differences of bands $B-z^{\prime}$ against $R-z^{\prime}$ from the Subaru catalogs. The hosts of compact radio sources are shown in blue. The rest of the entries in the Subaru catalogs are as a gray 2D histogram. The use of apparent instead of intrinsic magnitudes causes the significant scatter in the latter distribution; however, red (upper) and blue (lower) branches are evident. With reference also to Figure 3, the compact radio catalog is mostly detecting galaxies with redshifts higher than those of the clusters (although not necessarily on sight lines that pass through the mass distribution of the cluster). Despite the use of apparent magnitudes, the color distributions of the galaxies with radio detections are also visible. The radio data are deep enough to detect large numbers of galaxies in the blue cloud, i.e. regular star-forming spiral galaxies.

Zoom In Zoom Out Reset image size

The diffuse radio emission associated with galaxy clusters is generally observed at low radio frequencies due to the low surface brightness and steep radio spectra; however, our 3 GHz observations are deep enough to provide images of these structures at high angular resolution.

The central right-hand panels of Figures 1 and 2 are powerful illustrations of the usefulness of high angular resolution for relic and halo studies. MACS J0717 hosts one of the brightest radio halos known (e.g., van Weeren et al. 2017; Bonafede et al. 2018) and although the S-HIGH image resolves out the largest scales of the diffuse halo, the filamentary structures surrounding the central NAT radio galaxy can be seen with detail not possible with current low-frequency instruments. The central tailed radio source has a host galaxy with a redshift that places it within the cluster; however, Rajpurohit et al. (2021) use spectral modeling to suggest that it is not interacting with the halo itself and is merely seen in projection.

Bonafede et al. (2012) present observations of MACS J1149 at 323 MHz and 1.4 GHz, concluding the cluster hosts a double radio relic. Double relic structures are relatively rare, thought to arise when an ongoing cluster merger is seen from a favorable viewpoint (Bonafede et al. 2017). Figure 10 shows the view of the relevant regions afforded by our 3 GHz data. The grayscale is the S-HIGH image, overlaid with the contours of the 3'' resolution S-LOW image (please refer to the figure caption for further details). The left-hand panel shows the eastern relic, which is revealed to be a complex structure that is resolved into two distinct, almost perpendicular components. The bright peak seen at lower frequencies is actually an embedded NAT source (VLAHFF-J114922.34+222327.7, panel number 60 in Figure 13(a)). At z = 0.545, this source could truly be embedded in the relic, rather than being a foreground or background source seen in projection. The right-hand panel shows the region of the reported second relic in MACS J1149, which in our high-resolution observations is revealed to be an FR-I source, VLAHFF-J114922.34+222327.7 (panel number 53 on Figure 13(a)). The high-resolution radio data also allow us to provide an unambiguous optical identification for the host galaxy, which has a photometric redshift of 0.24, and so this entire radio structure is not actually associated with the cluster.

Zoom In Zoom Out Reset image size

The three sections of Figure 13(a) show the S-HIGH contours for the 66 sources identified as having complex radio morphologies, overlaid on an RGB image formed from the Subaru B-, R-, and z-band images. Please refer to the caption of Figure 13(b) for further details. Note that these images are not primary beam corrected, in the interests of achieving a uniform contouring scheme, and are presented primarily for morphological classification. The properties of each of the complex sources are provided in order of decreasing total (primary-beam-corrected) flux density in Tables 4–6 for MACS J0416, MACS J0717, and MACS J1149, respectively. Here we provide brief comments on the radio morphology and optical host.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 5. Positions and Integrated Flux Densities of the Extended Radio Sources in MACS J0717

ID	R.A.	Decl.	CLASH ID	zBPZ	${S}_{\mathrm{int}}^{\mathrm{HIGH}}$	σSint HIGH	${S}_{\mathrm{int}}^{\mathrm{LOW}}$	σSint LOW
	(deg)	(deg)			(mJy)	(mJy)	(mJy)	(mJy)
VLAHFF-J071632.28+373912.5	109.1345	37.65348	26801	2.813	301.87	0.34	396.62	0.47
VLAHFF-J071632.65+374251.6	109.13604	37.71434	⋯	⋯	25.63	0.14	89.57	0.19
VLAHFF-J071644.28+373956.1	109.18451	37.66561	43871	0.076	74.06	0.14	245.25	0.18
VLAHFF-J071723.46+374529.8	109.34777	37.75829	⋯	⋯	23.85	0.03	68.43	0.03
VLAHFF-J071724.97+375331.3	109.35408	37.89205	71157	0.626	24.64	0.08	56.79	0.12
VLAHFF-J071725.06+374714.7	109.35442	37.78742	54902	0.431	1.8	0.03	4.09	0.03
VLAHFF-J071725.70+373717.3	109.35712	37.62148	21414	0.709	16.3	0.1	32.24	0.08
VLAHFF-J071725.95+373352.8	109.35816	37.56468	18841	0.067	8.62	0.15	28.77	0.22
VLAHFF-J071730.66+374651.1	109.37778	37.78089	49833	1.524	2.51	0.05	6.12	0.05
VLAHFF-J071732.20+375619.7	109.38419	37.93882	86311	0.882	24.46	0.2	69.94	0.24
VLAHFF-J071735.48+374444.8	109.39787	37.7458	44706	0.555	1.06	0.03	12.51	0.14
VLAHFF-J071738.28+374650.4	109.40954	37.78067	50671	0.605	48.96	0.05	69.63	0.04
VLAHFF-J071741.15+374313.7	109.42149	37.72047	39947	0.56	214.0	0.72	631.53	0.11
VLAHFF-J071751.07+374440.3	109.46282	37.74454	44825	0.537	6.82	0.06	13.55	0.06
VLAHFF-J071752.69+374527.0	109.46956	37.75753	47767	0.502	2.95	0.02	16.92	0.04
VLAHFF-J071753.50+374209.3	109.47295	37.70261	36924	0.563	328.89	0.73	993.4	0.21
VLAHFF-J071803.45+375203.4	109.5144	37.86762	68329	0.4	225.45	0.62	393.2	0.28
VLAHFF-J071804.74+374852.3	109.51977	37.81453	56662	0.955	6.52	0.08	21.18	0.13
VLAHFF-J071806.37+373558.1	109.52657	37.59949	⋯	⋯	141.17	0.44	130.6	0.20
VLAHFF-J071810.75+374926.7	109.54479	37.82411	59007	0.642	396.47	0.68	1320.51	0.84
VLAHFF-J071815.16+374556.3	109.56318	37.76565	48526	0.906	6.78	0.05	19.25	0.07

Note. Please refer to the caption of Table 4 for further details.

Download table as:  ASCIITypeset image

Table 6. Positions and Integrated Flux Densities of the Extended Radio Sources in MACS J1149

ID	R.A.	Decl.	CLASH ID	zBPZ	${S}_{\mathrm{int}}^{\mathrm{HIGH}}$	σSint HIGH	${S}_{\mathrm{int}}^{\mathrm{LOW}}$	σSint LOW
	(deg)	(deg)			(mJy)	(mJy)	(mJy)	(mJy)
VLAHFF-J114911.81+222049.6	177.29924	22.34713	47811	0.107	3.07	0.05	11.64	0.05
VLAHFF-J114912.61+222114.4	177.30256	22.354	52412	0.175	3.57	0.06	19.8	0.08
VLAHFF-J114915.00+222123.2	177.31251	22.35646	50105	0.488	7.77	0.04	33.26	0.06
VLAHFF-J114917.65+221725.6	177.32358	22.29046	34122	0.708	27.24	0.03	63.5	0.08
VLAHFF-J114919.43+222621.8	177.33099	22.4394	68009	0.191	1.48	0.02	6.672	0.05
VLAHFF-J114922.34+222327.7	177.34311	22.39104	62337	0.24	1.80	0.03	⋯	⋯
VLAHFF-J114933.09+222036.7	177.3879	22.34353	46902	0.558	33.74	0.07	57.73	0.08
VLAHFF-J114933.62+221307.7	177.39009	22.21883	⋯	⋯	291.76	0.28	1316.9	1.15
VLAHFF-J114935.51+222403.4	177.39796	22.40095	58850	0.553	0.49	0.02	2.09	0.04
VLAHFF-J114936.51+222559.2	177.40217	22.43313	65767	0.739	5.37	0.02	74.12	0.05
VLAHFF-J114936.83+222609.9	177.40349	22.43609	67239	0.563	13.49	0.04	74.21	0.06
VLAHFF-J114939.36+222430.7	177.41402	22.40853	60954	0.561	17.16	0.02	69.36	0.03
VLAHFF-J114942.53+222037.6	177.42725	22.34378	46364	0.545	7.77	0.04	⋯	⋯
VLAHFF-J114952.24+222500.4	177.46767	22.4168	62088	1.123	1.48	0.02	6.27	0.03
VLAHFF-J114957.22+222018.8	177.48843	22.33856	44764	0.986	139.16	0.1	545.25	0.21
VLAHFF-J115003.87+221711.9	177.51616	22.28667	34839	0.229	1.54	0.07	5.95	0.1
VLAHFF-J115014.53+221734.3	177.56056	22.29288	⋯	⋯	8.16	0.13	40.94	0.21
VLAHFF-J115015.28+222052.7	177.56368	22.34797	48740	0.54	5.13	0.08	25.57	0.13
VLAHFF-J115029.89+222524.5	177.62458	22.42348	64436	0.848	5.49	0.18	28.8	0.29

Note. Please refer to the caption of Table 4 for further details.

Download table as:  ASCIITypeset image

The properties of radio sources with extended morphologies are presented in Tables 4–6 for MACS J0416, MACS J0717, and MACS J1149, respectively. The equatorial (J2000) positions are the flux-weighted centroids of the emission associated with each extended source, as determined by ProFound. Optical IDs from the CLASH catalogs together with any available photometric redshifts are also listed, together with the integrated flux densities (and associated uncertainty) as measured from the S-HIGH and S-LOW images. The latter provides a more robust estimate of the total integrated flux density of these sources at 3 GHz. As can be seen from the tables, the extended A configuration of the VLA (coupled with the weighting scheme required to deliver the high angular resolution required for the primary goal of these observations) resolves out a significant amount of extended emission.

(1) VLAHFF-J041514.28–240934.3: This is a low surface brightness isolated radio lobe, with a compact feature that may be a Fanaroff–Riley Type-2 (FR-II; Fanaroff & Riley 1974) hotspot. It is associated with the compact core in the lower right of the panel, which also hosts a lobe to the west; however, this source is not cataloged as it falls outside the primary beam cutoff. The host appears to be an elliptical.

(2) VLAHFF-J041528.24–240913.7: Double-lobed structure associate with an elliptical galaxy. The host galaxy has no cataloged optical/near-infrared ID.

(3) VLAHFF-J041549.94–235437.3: FR-I (Fanaroff & Riley 1974) radio source. The host is not visible in the Subaru imaging, so is possibly dust obscured.

(4) VLAHFF-J041553.04–240716.5: This is an FR-I structure, or possibly due to the low axial ratio of the lobes, a relic radio galaxy. No compact hot spots are evident. The host is an elliptical with a redshift of 0.499.

(5) VLAHFF-J041554.13–240406.9: Two optical galaxies are enveloped by edge-brightened radio emission with a diffuse tail structure. The tabulated host is the central galaxy, at a redshift of 1.645.

(6) VLAHFF-J041556.77–240434.5: This radio source shows resolved emission apparently associated with a spiral galaxy at z = 1.695. The radio emission appears to be asymmetric with respect to the disk, so it may be a jet structure associated with a rare spiral AGN, rather than star-formation-driven radio emission. An alternative explanation is that the alignment with the spiral is a projection effect, and this is a radio lobe, possibly associated with the compact radio source associated with the elliptical galaxy also seen in this panel.

(7) VLAHFF-J041603.56–240429.2: A possible hybrid morphology source, the radio emission appears to be FR-I-like on the northern side, with a diffuse radio lobe on the southern side. No hot spots are visible. The optical host is a z = 0.945 galaxy with unclear or disturbed optical morphology.

(8) VLAHFF-J041603.63–240551.4: This is an NAT radio galaxy, hosted by an elliptical galaxy at z = 0.392. Tailed radio galaxies are very common in massive clusters, and the host galaxy redshift associates it with the MACS J0416 cluster.

(9) VLAHFF-J041604.63–240415.0: A comparatively low-redshift (z = 0.037) elliptical galaxy, likely hosting a one-sided core–jet structure.

(10) VLAHFF-J041604.84–241028.0: Another one-sided core–jet structure, hosted by an elliptical galaxy at z = 0.288. There are some artifacts associated with this source; however, we believe the jet structure to be real, due to its three-contour significance, and the lack of correspondingly bright positive or negative features at the expected 120° positions that one would expect from a PSF-like artifact in a VLA image.

(11) VLAHFF-J041605.95–235733.6: There is an issue with the green channel in the optical imaging across this source; however, it appears to be a pair of merging or conjunction of two galaxies. The radio emission exhibits a double peak and a tail.

(12) VLAHFF-J041609.06–240945.0: This is another comparatively low-redshift source (z = 0.025). The radio morphology exhibits two peaks; however, it is most probably a core–jet structure associated with the coincident elliptical galaxy.

(13) VLAHFF-J041609.16–240402.8: A double-peaked radio source with a diffuse envelope, coincident with an elliptical galaxy with a photometric redshift of 0.438. The two radio peaks are of comparable brightness, however only one of them is coincident with an optical peak, and the detected radio structure is small compared to the size of the host. This could be a young radio jet or a double AGN.

(14) VLAHFF-J041609.64–240555.3: This source has what appears to be a one-sided core–jet structure; however, it could also be a young FR-II source as the innermost radio peak is offset from the peak of the optical emission.

(15) VLAHFF-J041613.94–235645.0: This relatively low-redshift (z = 0.024) massive elliptical exhibits a feature in the optical imaging that resembles a spiral arm, possibly a tidal tail. The radio emission is a compact core with a diffuse envelope, possibly driven by both a central AGN and circumnuclear star formation.

(16) VLAHFF-J041620.05–241008.0: A strong compact radio source associated with a galaxy at z = 1.176. The image is dynamic range limited at this position. The linear diagonal features are likely residual sidelobes from the imperfect deconvolution; however, the diffuse FR-I-like jet structure to the west may be real.

(17) VLAHFF-J041621.02–235804.8: A FR-I type galaxy, or possible wide-angle tail (WAT) source at z = 1.438.

(18) VLAHFF-J041625.22–240011.6: A strong radio source associated with the core of an elliptical at z = 0.516. The source exhibits a possible core–jet extension, but is dynamic range limited.

(19) VLAHFF-J041628.34–235321.9: FR-I radio morphology associated with an elliptical galaxy at z = 0.358.

(20) VLAHFF-J041629.22–235903.1: Resolved radio emission associated with the disk of a spiral galaxy at z = 0.574.

(21) VLAHFF-J041632.03–240340.8: A core–jet source or possibly a resolved disk at z = 0.874

(22) VLAHFF-J041633.46–240857.7: One-sided core–jet structure at a redshift of 1.184.

(23) VLAHFF-J041646.28–235623.4: FR-II radio galaxy at z = 3.53.

(24) VLAHFF-J041648.84–240243.4: Another bright core with a possible jet extension, but the image is dynamic range limited. The host galaxy is at a redshift of 0.831.

(25) VLAHFF-J041649.20–235802.6: A double-lobed radio galaxy at z = 0.423. Host optical morphology is elliptical, with a possible merging counterpart.

(26) VLAHFF-J041656.65–240600.4: Bright compact core, with possible FR-I structure but the image is dynamic range limited. The host galaxy is at z = 1.148.

(27) VLAHFF-J071632.28+373912.5: Core–jet radio source associated with a compact elliptical galaxy at z = 2.813.

(28) VLAHFF-J071632.65+374251.6: This is likely a core–jet source, but due to the offset radio peak, we assume that the true optical host is obscured by a bright foreground object at z = 0.017.

(29) VLAHFF-J071644.28+373956.1: Resolved radio emission from an elliptical galaxy at z = 0.076, or possibly a compact FR-I source associated with the nucleus of the host.

(30) VLAHFF-J071723.46+374529.8: This source is an NAT radio galaxy. The compact core is prominent in the C-HIGH image; however, there is no compact component at that position in the S-HIGH image indicating that the core is synchrotron self-absorbed. The optical host is obscured by the foreground galaxy or star in Subaru imaging. The radio source visible to the south in the corresponding figure is unrelated.

(31) VLAHFF-J071724.97+375331.3: A highly asymmetric twin jet structure, or possibly a WAT with a resolved out or otherwise nondetected jet. The optical data place the host at z = 0.626.

(32) VLAHFF-J071725.06+374714.7: Resolved radio emission associated with a spiral galaxy at z = 0.431. The radio morphology shows a possible spiral arm structure.

(33) VLAHFF-J071725.70+373717.3: Diffuse double-lobed radio source at z = 0.709, with no clear sign of the hotspot emission at the lobe heads.

(34) VLAHFF-J071725.95+373352.8: An intriguing source at redshift 0.067, showing compact jets associated with either a large edge-on spiral, or an elliptical with a prominent dust lane.

(35) VLAHFF-J071730.66+374651.1: This source is likely a compact FR-II radio galaxy at z = 1.524.

(36) VLAHFF-J071732.20+375619.7: A tailed radio galaxy, or extremely one-sided source. At a redshift of 0.882, it is not an NAT or WAT associated with the foreground cluster.

(37) VLAHFF-J071735.48+374444.8: A compact core–jet source, seen to be embedded in the bright halo emission of the MACS0717 cluster. With a photometric redshift of 0.555, the host galaxy is likely to be a cluster member.

(38) VLAHFF-J071738.28+374650.4: A FR-I source at z = 0.605. The northern jet has a significant extension that changes direction with respect to the jet emission close to the core; however, there is no southern counterpart to this very extended feature.

(39) VLAHFF-J071741.15+374313.7: This source is an NAT with twisted jets that are clearly visible in the C-HIGH image. The host appears to be an elliptical galaxy with a photometric redshift of 0.56.

(40) VLAHFF-J071751.07+374440.3: A WAT source at z = 0.537, likely associated with a cluster member.

(41) VLAHFF-J071752.69+374527.0: This is a compact FR-I source at z = 0.502.

(42) VLAHFF-J071753.50+374209.3: A double-lobed source with a pair of diffuse inner hot spots, possibly a restarted AGN. The host is at z = 0.563.

(43) VLAHFF-J071803.45+375203.4: A head–tail radio galaxy at z = 0.4 with very diffuse, low axial ratio jet emission. Could possibly be an NAT or WAT seen in projection.

(44) VLAHFF-J071804.74+374852.3: Possible hybrid morphology (FR-I/FR-II) radio source at z = 0.955.

(45) VLAHFF-J071806.37+373558.1: A bent tail (NAT/WAT) radio source with a clear optical host in the Subaru images but no cataloged counterpart.

(46) VLAHFF-J071810.75+374926.7: This source is possibly a one-sided FR-I with a large opening angle, but the image is dynamic range limited due to the bright radio core. The host galaxy is at z = 0.642.

(47) VLAHFF-J071815.16+374556.3: The radio emission shows either a core–jet source, or possibly the resolved disk of the optical host galaxy at z = 0.906.

(48) VLAHFF-J114911.81+222049.6: This radio source is likely driven by circumnuclear star formation in an elliptical galaxy at z = 0.107.

(49) VLAHFF-J114912.61+222114.4: Resolved radio emission associated with the disk of a spiral galaxy at z = 0.175.

(50) VLAHFF-J114915.00+222123.2: FR-I structure associated with a redshift 0.488 elliptical galaxy.

(51) VLAHFF-J114917.65+221725.6: Double compact radio sources, both within the redshift 0.708 host. Possible double nucleus.

(52) VLAHFF-J114919.43+222621.8: The morphology of this source suggests it is resolved star-formation-driven radio emission at z = 0.191.

(53) VLAHFF-J114922.34+222327.7: This is a bright FR-I radio source on the periphery of the MACS1149 field; however, with a photometric redshift of 0.24 it is likely not associated with the cluster itself. The host appears to be a large, low surface brightness elliptical galaxy. We discuss this source further in Section 4.7.

(54) VLAHFF-J114933.09+222036.7: This source is a WAT with a redshift of 0.558, likely associated with the MACS J1149 cluster.

(55) VLAHFF-J114933.62+221307.7: There is no cataloged counterpart for the host galaxy of this source; however, a strong candidate is visible in Subaru imaging. The radio morphology is a disturbed, asymmetric FR-II structure with prominent radio plumes.

(56) VLAHFF-J114935.51+222403.4: This object is a one-sided core–jet source associated with a redshift 0.553 galaxy, likely a cluster member.

(57) VLAHFF-J114936.51+222559.2: A compact FR-II source with a z = 0.739 host galaxy.

(58) VLAHFF-J114936.83+222609.9: A tailed radio galaxy, likely an NAT or WAT seen in projection. The optical counterpart is clear, with a cataloged redshift of 0.563, likely associating it with the MACS J1149 cluster.

(59) VLAHFF-J114939.36+222430.7: This appears to be a bent FR-I source at z = 0.536.

(60) VLAHFF-J114942.53+222037.6: Another NAT at z = 0.545. This source appears to be embedded in the eastern radio relic associated with the MACS1149 cluster. See Section 4.7 for more details.

(61) VLAHFF-J114952.24+222500.4: Three components are visible in this compact source, which is possibly a young FR-II with a redshift of 1.123.

(62) VLAHFF-J114957.22+222018.8: The radio emission from this strong source appears to be resolved in multiple directions, but the image is dynamic range limited at this point, so this must be interpreted with care. The host galaxy is at a redshift of 0.986.

(63) VLAHFF-J115003.87+221711.9: This is diffuse radio emission at low S/N. Likely driven by star formation, it is associated with the core of a spiral galaxy at z = 0.229.

(64) VLAHFF-J115014.53+221734.3: No optical counterpart is visible for this source. Given the morphology, it is likely to be an isolated radio lobe. The morphology also suggests the most likely core counterpart is the compact source VLAHFF-J115008.52+221733.8 (z = 0.25, and not visible in this panel). At this redshift, the projected separation is 328 kpc, which is not unreasonable for an FR-II. No western counterpart lobe is visible on the other side of the putative core.

(65) VLAHFF-J115015.28+222052.7: This is an FR-I source associated with an elliptical galaxy at z = 0.54, likely associated with MACS J1149.

(66) VLAHFF-J115029.89+222524.5: The radio emission is aligned perpendicular to the major axis of the elliptical host galaxy (z = 0.848) and is likely a compact pair of radio jets.