Radio Galaxies at TeV Energies

Abstract

Unlike blazars, radio galaxies have jets that are misaligned relative to our line-of-sight. This misaligned geometry provides us with a unique view of both the jet and super massive black hole. To date, four radio galaxies have been detected at TeV energies with an additional two active galactic nuclei shown to exhibit both radio galaxy and BL Lac-type properties. TeV observations of radio galaxies have revealed these objects to be fascinating, displaying ultra-fast variability and often relatively hard spectral energy distributions. This work aims to provide a review of the current state of radio galaxy observations within the context of very-high-energy $\gamma$-ray astronomy, while also highlighting that radio galaxies are excellent targets for multi-wavelength observations. A number of motivations for the continued study of radio galaxies are provided, and these are discussed with a focus on the key observational results, including implications for future observations with next-generation instruments soon to be operational.

1. Introduction

Active Galactic Nuclei (AGNs) can be defined as the compact regions at the centre of a galaxy in the process of accreting gas through a disk onto the supermassive black holes (SMBHs) residing at their cores. Detected across a broad band of electromagnetic frequencies/wavelengths, AGNs are typically characterised by their high non-stellar luminosity, and these objects can form jets of plasma extending outward from their SMBHs. Strong radio emission from a small fraction of AGNs indicates the presence of a relativistic jet [1], but it is still unknown as to why only some fraction of AGNs produce jets, while others do not. It is believed that a few AGN properties—mass, spin, accretion rate and mode, and spin orientation—along with the environment, determine whether an AGN is jetted or not, but a clear consensus is still to be reached [2,3,4,5,6,7].

Radio galaxies are a subclass of AGNs that are powerful emitters of radio waves, delivering approximately a million times more power ($\sim {10}^{38}\mathrm{watts}$) than our own Milky Way Galaxy ($\sim {10}^{32}\mathrm{watts}$). Typically, radio galaxies are characterised as having a bright compact core that is coincident with the visible nucleus of the galaxy. However, the radio emission extends beyond this core along jets that propagate outward from the central core to distances of pc or even kpc scales, as seen in NGC 6251 for example [8]. At the termination site of these jets, large plumes are often visible far beyond the central confines of the core. For a comprehensive discussion of relativistic jets from AGNs, which is beyond the scope of this work, please see the excellent review by [9] and all the references therein.

A large number of radio-loud AGN sub-classes are defined based on observational properties seen in specific wavebands, and it is suggested that radio galaxies form the parent population of blazars (AGNs with a relativistic jet pointing toward us). The radio galaxies with an edge-darkened morphology known as FR-Is (Fanaroff and Riley classification, [10]) correspond to BL Lacs observed at large jet viewing angles, while edge-brightened lobe morphology FR-II radio galaxies are associated with flat-spectrum radio quasars (FSRQs) [11]. The distinction between FR-I and FR-II radio galaxies is largely their radio power [12], where FR-Is are weaker, but more common and, hence, more likely to be in a sample of nearby galaxies. Doppler boosting decreases as the jet viewing angle ($\mathrm{i}$) to the line-of-sight increases. Therefore, within the context of VHE $\gamma$-ray astronomy, we refer to radio galaxies as radio-loud AGNs with jet viewing angles beyond the blazar condition ($\mathrm{i}<{10}^{\circ }$), yet the angle remains small enough for the detection of Doppler-boosted $\gamma$-rays. In practice, this includes objects falling near the boundary, and these may exhibit properties of both radio galaxies and blazars (but less extreme). While this distinction between blazars and radio galaxies is somewhat arbitrary, due to relativistic effects and strong Doppler boosting along the line-of-sight, blazars may exhibit features (such as Doppler factors $>10$ and apparent speeds $>10 \mathrm{c}$ [13]) that one would not expect to see in radio galaxies. A review of blazar characteristics is beyond the scope of this work; instead, the reader is directed to the following references [14,15,16] also within this Special Edition publication of the Galaxies journal.

To date, four radio galaxies (all of them FR-Is) have been detected at TeV energies using ground-based imaging atmospheric Cherenkov telescopes (IACTs) such as the High Energy Stereoscopic System (H.E.S.S.) [17], the Major Atmospheric Gamma Imaging Cherenkov Telescopes (MAGIC) [18,19], and the Very Energetic Radiation Imaging Telescope Array System (VERITAS) [20,21]. Of the 84 AGNs detected at TeV energies, all but 6 are blazars. Of these, the four detected radio galaxies are: M 87, Cen A, NGC 1275, and 3C 264. The other two AGNs (PKS 0625-35 and IC 310) show both radio galaxy and blazar properties [22]. In addition to the extragalactic sources highlighted above, IACTs have also detected two starburst galaxies, M 82 [23] and NGC 253 [24]. Unlike the AGN-powered radio galaxies and blazars, these extragalactic objects are characterised by their high star-formation rate, where gamma-ray emission is believed to originate from the acceleration of cosmic rays within the numerous shock waves caused by the high abundance of supernova remnants around massive young stars.

Over the past decade or so, $\gamma$-ray astronomy has seen considerable advances. This results from two critical factors: improved analysis techniques for IACTs to extract the maximum information from their data and the existence of the Large Area Telescope aboard NASA’s Fermi satellite (Fermi-LAT) [25]. The situation is set to improve further at very high energies (VHE) with the advent of the Cherenkov Telescope Array (CTA) [26], providing coverage from a few 10s of GeV to several hundred TeV with unprecedented sensitivity. Seamless coverage of the electromagnetic spectrum spanning eight orders of magnitude in energy is now within our grasp.

Furthermore, as Meyer et al. [27] reported, observation baselines on the order of 20 years for the Hubble Space Telescope (HST) [28,29,30] and 40 years for the Very Large Array (VLA) [31,32], coupled with major advances in astrometry techniques, have led to great improvements in our understanding of jet kinematics. Additionally, with very-large-baseline interferometry (VLBI) [33] observations, we have seen superluminal proper motions (discovered for example in the quasars 3C 279 and 3C 273 [34]) within the inner regions of extragalactic jets, but how jet velocities evolve from these parsec scales to kiloparsec (and beyond) scales is still something of a mystery. Understanding how these jets evolve along their paths toward interacting with the intergalactic medium is critical for our understanding of jet structures and how much of the energy budget jets are able of carrying and depositing out in the AGN’s large-scale environment. Meyer et al. [27] also reported that by pairing long-time-baseline observations of the kpc-scale jets with HST (capable of achieving accuracies for proper motion measurements of a tenth of a milliarcsecond per year), with VLA and with VLBI measurements of the parsec-scale jet, astronomers are now able to map the full velocity profile of jets from close to their SMBHs to beyond the extents of their host galaxy. Furthermore, with new radio and optical instruments coming online soon, all of which should be operational concurrently with CTA, the accuracy and extent to which we can measure the proper motions of extragalactic jets is expected to improve further still.

The aim of this work is to review the current state of radio galaxies relating to VHE $\gamma$-ray astronomy. Focus is placed on the four TeV-detected AGNs often classified as FR-Is, and for the purpose of this review, these are referred to as radio galaxies (in line with the convention of the VHE community), even though these AGNs may be close to the boundary with blazars. For an earlier review of radio galaxies addressing TeV challenges, please see [35]. Section 2 provides some context by introducing some key characteristics of radio galaxies in terms of blazar studies and long-term $\gamma$-ray observations at GeV energies, as well as highlighting some motivations for the continued study of radio galaxies at VHE. Section 3 focuses on the results and findings from observations of radio galaxies detected at VHE, while finally, Section 4 discusses the implications of these key results, as well as addressing some of the future opportunities relating to the observation of radio galaxies with the next-generation of radio and TeV $\gamma$-ray observatories.

2. Radio Galaxies in Context

2.1. Blazar Studies

Understanding how jets are formed, collimated, and accelerated to relativistic speeds is a key goal of blazar studies. The jet emission from blazars and radio galaxies is characterized by a double-peaked, non-thermal spectral energy distribution (SED) [36]. The lower-frequency peak, which in blazars has a peak frequency ranging from ${10}^{12}$ to ${10}^{18}$ Hz, is well described as synchrotron emission. The higher-frequency peak (${10}^{18}\mathrm{Hz}–{10}^{27}\mathrm{Hz}$) is generally attributed to inverse-Compton (IC) emission arising from synchrotron self-Compton (SSC) [37,38,39] or external Compton (EC) processes1, or a combination of the two. Typically, the sources with higher-frequency synchrotron peaks have IC peaks in the $\sim 10–100\mathrm{GeV}$ energy range. Thus, high-synchrotron-peak (HSP) blazars (${\mathsf{\nu }}_{\mathrm{peak}}>{10}^{15}\mathrm{Hz}$) are the brightest and best-studied at VHE (currently, 53 of the AGNs detected at VHE), although they are the least powerful. Only nine blazars in the current TeV catalogue are low-synchrotron-peak (LSP) objects (${\mathsf{\nu }}_{\mathrm{peak}}⩽{10}^{14}\mathrm{Hz}$), including both FSRQs and BL Lacs.

Unlike blazars, radio galaxy jets are thought to be oriented at larger angles to the line-of-sight (≥10${}^{\circ }$), implying they are are much less Doppler-boosted than blazars, thus exhibiting SEDs similar to LSP blazars [40]. There does appear to be an observed relationship between jet and accretion power [41]. It is also postulated that, in addition to differentiating source classes by accretion modes, a range of jet viewing angles is needed to explain a blazar sequence; low-power jets with inefficient accretion have structured jets [42], whereby the synchrotron peak frequency decreases as the “misalignment” increases and observations probe the lower-speed sheath of the jet. In this scenario, radio galaxies should then cluster at the low end of the synchrotron peak frequency distribution [40].

Keenan et al. [43] extended this idea, with a resulting catalogue of over 2000 radio-loud AGNs with well-sampled synchrotron SEDs (including over 1000 with jet power estimates) showing the distribution of sources on the synchrotron peak—luminosity (L) peak plane is driven by a dichotomy in jet strength where weak jet sources associated with inefficient accretion dominate at $\log \left({\mathsf{\nu }}_{\mathrm{peak}}\right)>15$ and strong jet sources with efficient accretion are primarily found with $\log \left({\mathrm{L}}_{\mathrm{peak}}\right)>45$ and $\log \left({\mathsf{\nu }}_{\mathrm{peak}}\right)<15$. This shows that the blazar sequence is at least in part due to a selection effect.

2.2. Radio Galaxies at GeV Energies

Over the past decade, Fermi-LAT has provided a high-quality view of the $\gamma$-ray sky at GeV energies, detecting a number of radio galaxies during the course of its mission. These observations have acted as a pathfinder for TeV observations, for using the current generation of ground-based IACTs, as well as for predicting future observations with CTA, e.g., [44]. In the recently released incremental version of the fourth Fermi-LAT point-source catalogue, which covers the first 12 years of the Fermi-LAT mission (4FGL-DR3) [45], there are a total of 43 non-blazar AGN classified as radio galaxies. The 4FGL-DR3 details for these radio galaxies are highlighted in

Table 1. The Fermi-LAT 4FGL-DR3 radio galaxies, sorted by right ascension, that exhibit SEDs best fitted with a power-law model. Although Cen A is listed in this table, its SED is actually best fitted with a broken power-law model, as discussed in this review and elsewhere, e.g., [46,47]. The columns include the 4FGL catalogue name, an alternative object name, the type of model best fitted to the SED where PL is for power-law, the spectral index value ($\mathsf{\Gamma }$), the photon flux between $1–100\mathrm{GeV}\left[\mathrm{photons}{\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}\right]$, the 4FGL-DR3 variability index where an index value >18.48 indicates a <1% chance of being a steady source, and a TeVCat flag indicating whether the object is detected at TeV energies.

4FGL Name	Object	Model	$\mathsf{\Gamma }$	${\mathbf{F}}_{\mathbf{E}=1-100\mathbf{GeV}}$	Variability	TeVCat
4FGL J0009.7-3217	IC 1531	PL	2.24	$2.02\times {10}^{-10}$	27.9362	N
4FGL J0038.7-0204	3C 17	PL	2.84	$1.77\times {10}^{-10}$	17.5947	N
4FGL J0057.7+3023	NGC 315	PL	2.38	$2.74\times {10}^{-10}$	8.81993	N
4FGL J0153.4+7114	TXS 0149+710	PL	1.93	$2.60\times {10}^{-10}$	13.9387	N
4FGL J0237.7+0206	PKS 0235+017	PL	2.10	$1.36\times {10}^{-10}$	6.75128	N
4FGL J0308.4+0407	NGC 1218	PL	2.00	$8.44\times {10}^{-10}$	37.8178	N
4FGL J0312.9+4119	B3 0309+411B	PL	2.49	$1.74\times {10}^{-10}$	21.8156	N
4FGL J0316.8+4120	IC 310	PL	1.85	$2.52\times {10}^{-10}$	31.8913	Y
4FGL J0322.6-3712e	Fornax A	PL	2.05	$5.59\times {10}^{-10}$	14.2643	N
4FGL J0334.3+3920	4C +39.12	PL	1.85	$2.14\times {10}^{-10}$	14.2155	N
4FGL J0519.6-4544	Pictor A	PL	2.54	$2.95\times {10}^{-10}$	11.9384	N
4FGL J0627.0-3529	PKS 0625-35	PL	1.91	$1.26\times {10}^{-09}$	25.0043	Y
4FGL J0708.9+4839	NGC 2329	PL	1.72	$1.04\times {10}^{-10}$	13.1158	N
4FGL J0758.7+3746	NGC 2484	PL	2.23	$1.39\times {10}^{-10}$	7.38446	N
4FGL J0931.9+6737	NGC 2892	PL	2.28	$4.75\times {10}^{-10}$	49.7003	N
4FGL J0958.3-2656	NGC 3078	PL	2.16	$1.68\times {10}^{-10}$	7.16341	N
4FGL J1116.6+2915	B2 1113+29	PL	1.61	$3.54\times {10}^{-11}$	11.7614	N
4FGL J1144.9+1937	3C 264	PL	2.02	$2.95\times {10}^{-10}$	5.42454	Y
4FGL J1149.0+5924	NGC 3894	PL	2.19	$2.20\times {10}^{-10}$	10.8819	N
4FGL J1219.6+0550	NGC 4261	PL	2.09	$1.83\times {10}^{-10}$	13.2929	N
4FGL J1236.9-7232	PKS 1234-723	PL	2.40	$2.34\times {10}^{-10}$	6.54389	N
4FGL J1306.3+1113	TXS 1303+114	PL	1.86	$1.37\times {10}^{-10}$	14.3789	N
4FGL J1325.5-4300	Cen A	PL	2.59	$3.59\times {10}^{-09}$	33.403	Y
4FGL J1443.1+5201	3C 303	PL	2.15	$1.59\times {10}^{-10}$	9.88656	N
4FGL J1449.5+2746	B2 1447+27	PL	1.46	$5.24\times {10}^{-11}$	12.7277	N
4FGL J1516.5+0015	PKS 1514+00	PL	2.51	$3.03\times {10}^{-10}$	16.249	N
4FGL J1518.6+0614	TXS 1516+064	PL	1.79	$1.10\times {10}^{-10}$	6.96701	N
4FGL J1521.1+0421	PKS B1518+045	PL	2.04	$1.26\times {10}^{-10}$	21.5818	N
4FGL J1530.3+2709	LEDA 55267	PL	2.01	$8.30\times {10}^{-11}$	12.2306	N
4FGL J1724.2-6501	NGC 6328	PL	2.56	$1.87\times {10}^{-10}$	6.27477	N
4FGL J1843.4-4835	PKS 1839-48	PL	2.03	$1.46\times {10}^{-10}$	4.58083	N
4FGL J2227.9-3031	PKS 2225-308	PL	1.80	$1.11\times {10}^{-10}$	17.3154	N
4FGL J2302.8-1841	PKS 2300-18	PL	2.26	$2.00\times {10}^{-10}$	14.3889	N
4FGL J2326.9-0201	PKS 2324-02	PL	2.59	$2.27\times {10}^{-10}$	17.579	N
4FGL J2329.7-2118	PKS 2327-215	PL	2.45	$4.00\times {10}^{-10}$	18.1362	N
4FGL J2341.8-2917	PKS 2338-295	PL	2.24	$1.74\times {10}^{-10}$	13.6434	N

and

Table 2. The Fermi-LAT 4FGL-DR3 radio galaxies, sorted by right ascension, that exhibit SEDs best fitted with a log-parabola model. The columns include the 4FGL catalogue name, an alternative object name, the type of model best fitted to the SED where LP is for log-parabola, the log-parabola spectral index value ($\mathsf{\alpha }$), the log-parabola spectral curvature value ($\mathsf{\beta }$), the photon flux between $1–100\mathrm{GeV}\left[\mathrm{photons}{\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}\right]$, the 4FGL-DR3 variability index where an index value >18.48 indicates a <1% chance of being a steady source, and a TeVCat flag indicating whether the object is detected at TeV energies.

4FGL Name	Object	Model	$\mathsf{\alpha }$	$\mathsf{\beta }$	${\mathbf{F}}_{\mathbf{E}=1-100\mathbf{GeV}}$	Variability	TeVCat
4FGL J0319.8+4130	NGC 1275	LP	2.05	0.07	$3.38\times {10}^{-08}$	5460.94	Y
4FGL J0418.2+3807	3C 111	LP	2.62	0.23	$7.23\times {10}^{-10}$	40.5743	N
4FGL J0433.0+0522	3C 120	LP	2.63	0.21	$7.77\times {10}^{-10}$	308.796	N
4FGL J1230.8+1223	M 87	LP	2.01	0.04	$1.62\times {10}^{-09}$	23.8056	Y
4FGL J1306.7-2148	PKS 1304-215	LP	1.86	0.24	$5.22\times {10}^{-10}$	16.9549	N
4FGL J1630.6+8234	NGC 6251	LP	2.25	0.12	$1.26\times {10}^{-09}$	7.34	N
4FGL J2156.0-6942	PKS 2153-69	LP	2.47	0.48	$1.41\times {10}^{-10}$	5.73628	N

. Table 1 shows the radio galaxies that have Fermi-LAT-recorded SEDs best fitted with a power-law model, whereas Table 2 shows the radio galaxies that have Fermi-LAT-recorded SEDs best fitted with a log-parabola model.

The photon indices of the 4FGL-DR3 radio galaxies best fitted with a power-law model are found to be in the range between 1.4 and 2.9. In addition, the photon fluxes of all the radio galaxies at GeV energies vary between $\sim {10}^{-11}\mathrm{photons}{\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}$ and $\sim {10}^{-8}\mathrm{photons}{\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}$. As can be seen in Table 1 and Table 2, the majority of Fermi-LAT-detected radio galaxies have an SED adequately described by a simple power-law model, at least over a long time period such as the 12 years covered by the 4FGL-DR3 catalogue. As defined in the original release of the 4FGL eight-year point source catalogue [48], a source is considered to be variable over yearly binned timescales if its Variability Index (shown in Table 1 and Table 2) is larger than $18.48$. Based on this measure, ∼72% of the Fermi-LAT radio galaxies are not considered variable and $\sim 28\%$ are, with NGC 1275 and 3C 120 displaying the strongest variability of all the Fermi-LAT-detected radio galaxies. However, given that many of these radio galaxies are faint (i.e., weak) $\gamma$-ray sources, it cannot be concluded for certain that these objects are not variable, but rather that Fermi-LAT cannot measure their variability.

2.3. Motivations for Studying Radio Galaxies at TeV Energies

Within the context of TeV $\gamma$-ray astronomy, radio galaxies are typically thought of as “misaligned” AGNs with much reduced Doppler boosting factors. The detection of these objects at TeV energies was therefore somewhat of a surprise and opened up many questions about the $\gamma$-ray emission processes, $\gamma$-ray emission sites, and our general understanding about the nature of these extragalactic objects. Observations of radio galaxies are not overwhelmed with very strong Doppler-boosted emission (as is the case with blazars), and instead, the misaligned geometry provides us with a unique view of the AGN’s jet and SMBH. Despite the small sample size, the detection of $\gamma$-rays from radio galaxies provides a number of motivations for future observations and studying these objects further. The following is a small selection of the outstanding questions relating to radio galaxies at TeV energies:

• Do all TeV-detected radio galaxies exist in dense clumpy environments, for example? What role does the environment play in decelerating radio galaxy jets, for example, and does this interaction lead to $\gamma$-ray emission and/or $\gamma$-ray orphan flares? (See [49] and the references therein.)
• Do TeV-detected radio galaxies exhibit any common kinematic characteristics? How do radio galaxies accelerate particles in their jets, and can regions with different apparent velocity profiles (over all jet scales) be correlated with $\gamma$-ray emission? What mechanisms give rise to features in radio galaxy jets, e.g., internal shocks, and are these responsible for gamma-ray emission?
• Is it right to classify the TeV-detected radio galaxies as misaligned blazars, or are these objects simply blazar-like, with much smaller jet inclination angles to the line-of-sight than currently derived or assumed?
• In TeV-detected radio galaxies, where does gamma-ray emission occur? Is it close to the black hole in the inner-jet region or at kpc scales further down the jet? Are there multiple $\gamma$-ray emission sites and components, for example, both jet and large-scale emission from the up-scatter of external photons?
• Are TeV-detected radio galaxies best described by a singular leptonic, hadronic, or mixed emission model? Which process dominates?
• Are TeV-detected radio galaxies cosmic ray sources? If $\gamma$-ray emission from radio galaxies can be correlated with neutrino observations, can the population of as-yet-undetected radio galaxies in our universe make a substantial contribution towards the cosmic ray background?

Section 4 attempts to address some of the themes covered in these questions, but in order to resolve any of them, it is important to state that TeV observations alone cannot fully answer all of these questions. Instead, multi-wavelength observations from radio to optical to $\gamma$-rays are all needed, and we live in an exciting time with many new large and improved observatories on the horizon, all capable of making major contributions toward a better understanding of AGN science. Taking into account unification considerations [50], there is a motivation to explore these objects deeper and with higher precision. For example, to construct better theoretical emission models for radio galaxies, there is a need for well-sampled broadband spectra of these objects in different states. In order to do this at the highest energies, deeper observations are required. In addition, TeV observations provide us with the best opportunity to probe the fastest variability at the highest energies, as this will enable us to gain a better understanding of the nature of these objects and tell us more about the location of gamma-ray emission sites, for example whether the observed gamma-ray emission originates close to the core or further down the jet. Furthermore, it is suggested that given radio galaxies are now detected at both GeV and TeV energies, their large population likely provides a significant contribution toward the cosmic rays that permeate our universe. Therefore, multi-messenger observations of radio galaxies may provide us with insights into their overall contribution toward the cosmic ray background.

3. Results

As already mentioned above, four radio galaxies have been detected with ground-based IACTs. The discovery of these objects with the current generation of TeV instruments is exciting as this paves the way for some promising observations with future IACTs to provide new insights into the mechanisms driving gamma-ray emission. In this section, we highlight the key findings from the observations of the four TeV-detected radio galaxies including the additional two non-blazar AGN that display both radio galaxy and blazar characteristics.

Table 3. Highlighted here are the key characteristics of the 6 non-blazar AGNs detected at TeV energies. The table is sorted by redshift in ascending order. The column details are as follows: the common name given to the AGN (Name), an alternative name for the AGN (Alt. Name), redshift (z), the jet inclination angle to the line-of-sight (${\mathsf{\theta }}_{\mathrm{i}}$ (degrees)), jet power (${\mathrm{P}}_{1.4\mathrm{GHz}}^{\mathrm{jet}}$ (W/Hz)), radio luminosity at 5 GHz (${\mathrm{L}}_{5\mathrm{GHz}}$ (ergs/s)), synchrotron peak frequency (${\mathsf{\nu }}_{\mathrm{peak}}$ (Hz)), VHE discovery date (${\mathrm{dd}}_{\mathrm{VHE}}$ (year-month)), and finally, a column containing references for each object (Refs.).

Name	Alt. Name	z	${\mathsf{\theta }}_{\mathbf{i}}$	$\mathbf{log}{\mathbf{P}}_{1.4\mathbf{GHz}}^{\mathbf{jet}}$	${\mathbf{L}}_{5\mathbf{GHz}}$	$\mathbf{log}{\mathsf{\nu }}_{\mathbf{peak}}$	${\mathbf{dd}}_{\mathbf{VHE}}$	Refs.
Cen A	NGC 5128	0.0018	12${}^{\circ }$–45${}^{\circ }$	21.97	$1.91\times {10}^{40}$	12.81	2009-03	[22,43,51,52]
M 87	3C 274	0.0042	13${}^{\circ }$–27${}^{\circ }$	23.27	$5.83\times {10}^{39}$	9.5	2003-05	[22,43,52,53]
NGC 1275	3C 84	0.0175	30${}^{\circ }$–55${}^{\circ }$	-	$1.39\times {10}^{42}$	13.39	2010-10	[22,54]
IC 310	B3 0313+411	0.0188	$\sim 8^{\circ }$	-	$9.94\times {10}^{39}$	$\sim 17.0$	2010-03	[22,55,56]
3C 264	NGC 3862	0.0216	$\sim {50}^{\circ }$	23.67	$9.54\times {10}^{40}$	16.30	2018-03	[22,43,52,57,58]
PKS 0625-35	TXS 0625-354	0.0562	<53${}^{\circ }$	-	$7.67\times {10}^{41}$	$\sim 16.0$	2015-07	[22,59,60]

highlights the key characteristics of these objects.

3.1. NGC 1275

NGC 1275 (also commonly known as 3C 84) is the brightest cluster galaxy at the centre of the Perseus cluster of galaxies. It is located at $\sim 75$ Mpc and was detected by both MAGIC [61] and VERITAS [62]. Radio observations of NGC 1275 show a sub-parsec-scale radio jet oriented between $\sim {30}^{\circ }$ and 60${}^{\circ }$ with possible bending in the jet. Figure 1 shows NGC 1275 at 43 GHz (left panel) and 22 GHz (right panel) radio frequencies. Here, we see the sub-parsec-scale jet appears to have a complex structure and is dominated by two features2: the core (A0) and a knot (C2) [63]. There does not appear to be evidence of superluminal apparent speeds in the parsec-scale jet; indeed; C2 moves at subluminal speeds of $\sim 0.2$ c relative to the core. The left panel of Figure 1 also highlights a number of other knots (C1…C8), but the one with the highest recorded proper motion is C6 at $\sim 0.36$ c. Knot C3 appears to move upstream, while knot C4 appears to be at a significantly different, yet stable position angle. Recent observations suggest the component C3 is a remnant of previous emission [49,64].

The knot C2 emerged following a radio outburst in 2005 located in the central 1pc region [65]. After this outburst, limb-brightening of the jet was discovered, indicating that the transverse structure of the jet had changed. This morphological change appears to have coincided with an increased $\gamma$-ray flux level at GeV energies [66,67]. Furthermore, the radio emission from knot C2 was significantly polarized, and there appeared to be a location change from southwest to southeast. It is suggested that this movement and enhanced polarization indicated that the jet jittered in an inhomogeneous ambient medium [68]. A $\sim 0.4\mathrm{mas}$ directional flip of the knot C2 was confirmed with radio observations taken in 2015 and 2016, providing the first evidence of jet interaction with the environment [49]. It is believed the jet of NGC 1275 interacted/collided with a compact dense cloud, which led to a year-long frustration of the jet before breaking out in 2018. Subsequently, the knot C2 continued to move southward until eventually becoming distorted and returning NGC 1275 to a darkened lobe morphology state [64].

The high-resolution radio observations shown in the right panel of Figure 1 provide evidence of a limb-brightened jet. The large intensity ratio between the bright outer layer and the dim central region can be explained by transverse velocity structures and supports $\gamma$-ray emission scenarios via spine–sheath-structured jet models [42,69]. Sub-arcsecond optical images of NGC 1275 reveal a possible recent merger event [70], which is interesting because galaxy mergers are postulated to be potential sources of cosmic rays, neutrinos, and $\gamma$-rays [71].

NGC 1275 exhibits significant flux variability with flux doubling times of around 10 h observed, suggesting $\gamma$-ray emission originates from a compact region. In its quiescent state, the spectral shape of NGC 1275, determined from MAGIC observations, appears to follow a log-parabola (or power-law with sub-exponential cut-off) with a somewhat steep spectral index of $\mathsf{\Gamma }\sim -4.1$ and no excess $\gamma$-ray photons above $\sim 650$ GeV detected. In January 2017, significant increases in the flux levels seen from NGC 1275 were reported by MAGIC ($\sim 1.5$-times the flux from the Crab Nebula) [72] and VERITAS ($\sim 65\%$ the flux from the Crab Nebula) [73]. Significant variability was observed in the day-to-day light curve of NGC 1275 with a fastest flux-doubling timescale of $(611\pm 101)\min .$ seen [74]. Such a flux-doubling timescale implies that the emission region responsible for VHE $\gamma$-rays is larger than the event horizon light crossing time, but it still represents a compact region of the jet as opposed to a large-scale emission zone.

From these MAGIC observations of NGC 1275 during a flaring state, the MAGIC Collaboration et al. [74] attempted to model the observed SED assuming a classical SSC leptonic scenario. However, this was not possible due to the very-fast-variability scale constraining the size of emission zone at one to two orders of magnitude below that typically used in the SSC scenario adopted by the authors. In attempting to model the $\gamma$-ray emission from NGC 1275 by assuming an SSC leptonic scenario, and as hinted elsewhere, a higher Doppler factor would be needed to circumvent absorption, thus necessitating a jet inclination angle (to the line-of-sight) that is at odds with the jet inclination angle determined from radio observations.

Figure 1. The left panel taken from [63] shows the total intensity image of NGC 1275 at 43 GHz radio frequencies. The parsec-scale jet shows a complex structure and is dominated by two features: the core (A0) and a knot (C2). Knot C6 has the highest recorded proper motion of $\sim 0.36$ c. The right panel taken from [67] shows the $22\mathrm{GHz}$ space-VLBI image of NGC 1275’s jet from the core to $\sim 1\mathrm{pc}$. These high-resolution radio observations provide evidence of a limb-brightened jet. The large intensity ratio between the bright outer layer and the dim central region can be explained by transverse velocity structures and supports $\gamma$-ray emission scenarios via spine–sheath-structured jet models.

Figure 1. The left panel taken from [63] shows the total intensity image of NGC 1275 at 43 GHz radio frequencies. The parsec-scale jet shows a complex structure and is dominated by two features: the core (A0) and a knot (C2). Knot C6 has the highest recorded proper motion of $\sim 0.36$ c. The right panel taken from [67] shows the $22\mathrm{GHz}$ space-VLBI image of NGC 1275’s jet from the core to $\sim 1\mathrm{pc}$. These high-resolution radio observations provide evidence of a limb-brightened jet. The large intensity ratio between the bright outer layer and the dim central region can be explained by transverse velocity structures and supports $\gamma$-ray emission scenarios via spine–sheath-structured jet models.

3.2. Cen A

Cen A (NGC 5128) is the nearest ($\mathrm{d}\sim 3.8$ Mpc [75]) and one of the most well-studied extragalactic objects over a wide frequency domain. A misaligned FR-I radio galaxy, Cen A is often considered the prototypical Fanaroff–Riley FR-I classification. Cen A exhibits a complex and extended radio morphology with two giant lobes, as shown by the radio contours (black lines) in Figure 2. The latest sub-parsec-scale VLBI observations of Cen A obtained with the Event Horizon Telescope (EHT) at 228 GHz reveal a highly collimated, asymmetrically edge-brightened jet along with a faint counter-jet [76]. This strong limb-brightening seen in the sub-parsec-scale jet of Cen A is similar to that seen in other TeV-detected radio galaxies M 87 and NGC 1275, which suggests the dominant emission originates from the jet sheath and could potentially indicate that some component of $\gamma$-ray emission from Cen A takes place via spine–sheath-structured jet models.

Faint TeV $\gamma$-ray emission from Cen A was detected by H.E.S.S. after approximately $\sim 120$ h of observations. The TeV emission was detected with a statistical significance of $5\mathsf{\sigma }$ above the background coincident within a region including both the radio core and the inner kiloparsec jet. The integrated flux above $\sim 250$ GeV was estimated to be $\sim 0.8\%$ that of the flux from the Crab Nebula. A power-law shape was best fitted to the spectral energy distribution with a photon index of $2.7\pm 0.5_{\mathrm{stat}}\pm 0.2_{\mathrm{sys}}$, and no significant flux variability was detected [77].

Using the first four years of Fermi-LAT data, Sahakyan et al. [46] first reported evidence (albeit not statistically significant at $<3\mathsf{\sigma }$) for an observed hardening of Cen A’s energy spectrum. They reported a break in the spectrum at ${\mathrm{E}}_{\mathrm{b}}\sim 4\mathrm{GeV}$ with a spectral index of ${\mathsf{\Gamma }}_{\mathrm{E}<4\mathrm{GeV}}=2.74\pm 0.03$ below and ${\mathsf{\Gamma }}_{\mathrm{E}>4\mathrm{GeV}}=2.09\pm 0.20$ above the break. Sahakyan et al. [46] suggested that the best-fitted broken power-law model could be caused by the contribution of an additional high-energy component beyond common synchrotron self-Compton jet emission. With eight years worth of Fermi-LAT data, Brown et al. [47] confirmed the existence of a statistically significant spectral hardening in Cen A’s spectra, a likelihood-ratio test with $\mathrm{TS}=28.6\approx 5.3\mathsf{\sigma }$ demonstrating that Cen A’s spectra are best described by a broken power-law model compared to a standard power-law model. The spectral index values of their study agreed (within the reported errors) with those reported by Sahakyan et al. [46] both above and below the break energy. In addition, Brown et al. [47] found no evidence for significant flux variability from Cen A in either the low- or high-energy spectral components when considering the full eight-year dataset. However, they did note that their results showed some flux variability (at the $3.3\mathsf{\sigma }$ level) in the low-energy spectral component during the first 3.5 years of Fermi-LAT data. The lack of variability, in particular for the high-energy spectral component, rules out the possibility that the spectral feature of Cen A is a result of jet-induced leptonic processes. Instead, Brown et al. [47] suggested that the feature seen in the combined GeV-TeV SED is consistent with a spike in the dark matter halo profile, but also postulated that a population of unresolved millisecond pulsars or some other population of energetic particles could be responsible for the spectral feature.

Single-zone SSC models do not adequately describe Cen A’s observed broadband spectral energy distribution. As pointed out in [78], a variety of different scenarios exist, and none of them mutually exclusive, for the physical origin of the observed spectral energy break. To avoid going into great detail for each scenario here, the appropriate references are included again:

• Magnetospheric pulsar-like scenarios [79,80];
• Inner-jet models involving a number of leptonic SSC-emitting components that propagate at different angles relative to the line-of-sight [39], inverse-Compton emission from structured jets as discussed elsewhere in this article [42], photo-meson decay processes [81,82,83,84], lepto-hadronic emission models [85,86], and $\gamma$-ray-induced pair cascades from different regions [87,88,89];
• Extended emission scenarios such as the interaction of energetic protons with ambient matter (proton–proton interactions) at kiloparsec scales [46], millisecond pulsar contributions as already highlighted above [47], the inverse-Compton upscattering of photons on kiloparsec scales [90], or host galaxy starlight [91];
• The self-annihilation of dark matter, as already highlighted above [47].

With the current generation of IACTs, the jets of Cen A cannot be resolved. Therefore, distinguishing between these scenarios is currently difficult, and the observed $\gamma$-ray emission has tended to be associated with a region close to the black hole. Recent observations of Cen A by H.E.S.S. have focused on looking for extended emission and by comparing different models using a test statistic, and they found a strong preference for an elliptical Gaussian model ($\mathrm{TS}=19.4$) compared to a radially symmetric Gaussian model ($\mathrm{TS}=6.1$) [92]. This result, shown in Figure 2, implies that a major part of the VHE $\gamma$-ray emission originates on large scales at great distances from the SMBH.

Figure 2. Taken from [92], the 21 cm VLA radio intensity map of Cen A after convolution with the H.E.S.S. point spread function (PSF) and additional oversampling (arbitrary units labelled as A.U.). The solid black lines show the unconvolved VLA radio contours for the inner lobes of Cen A. The white dashed line shows the VHE $\gamma$-ray morphology of Cen A derived from H.E.S.S. observations, and the best fit elliptical Gaussian is shown by the $1\mathsf{\sigma }$ blue ellipse. The green dashed line shows the H.E.S.S. PSF $68\%$ containment area.

Figure 2. Taken from [92], the 21 cm VLA radio intensity map of Cen A after convolution with the H.E.S.S. point spread function (PSF) and additional oversampling (arbitrary units labelled as A.U.). The solid black lines show the unconvolved VLA radio contours for the inner lobes of Cen A. The white dashed line shows the VHE $\gamma$-ray morphology of Cen A derived from H.E.S.S. observations, and the best fit elliptical Gaussian is shown by the $1\mathsf{\sigma }$ blue ellipse. The green dashed line shows the H.E.S.S. PSF $68\%$ containment area.

3.3. M 87

M 87 (3C 274, NGC 4486) is a radio galaxy located in the Virgo cluster at a distance of $\mathrm{d}\sim 16.7\mathrm{Mpc}$ [93], making this the second closest radio galaxy after Cen A, but the closest kiloparsec-scale radio-optical jet. The jet of M 87 is misaligned with an angle $\mathrm{i}\sim$15${}^{\circ }$–25${}^{\circ }$ to the line-of-sight. M 87 is classified as a low-excitation weak-power FR-I-type radio galaxy displaying a radiatively inefficient accretion mode.

HST observations of M 87 provide evidence for superluminal speeds up to $\sim 6 \mathrm{c}$ out to $\sim 0.5\mathrm{kpc}$ [94], and radio observations of the feature known as HST-1 (located at a deprojected distance $>100\mathrm{pc}$) show superluminal speeds of $\sim 4 \mathrm{c}$ [95,96,97]. However, Kovalev et al. [98] found no evidence for motions faster than $0.07 \mathrm{c}$ within the inner $\sim 1.6\mathrm{pc}$ of the jet, and using radio images from the European VLBI Network, Asada et al. [99] suggested that the jet of M 87 is gradually accelerated from these subluminal speeds seen at the milliarcsecond scales of the inner-jet out to the superluminal speeds seen at HST-1. Mertens et al. [53] raised concerns with this scenario including the implied need for a jet-to-counter-jet intensity ratio of 1 when the jet of M 87 is essentially one-sided. Furthermore, the observed limb-brightened morphology of the jet strengthens the argument that the subluminal apparent speeds reported for the inner-jet point to a pattern speed or a wind from an outer sheath in the jet.

Using the $10\mathrm{m}$ Whipple Cherenkov telescope, in the years 2000/2001, early efforts were made by the VERITAS collaboration to detect VHE $\gamma$-ray emission from M 87 [100]. The question of whether M 87 emitted TeV $\gamma$-rays was again raised by Aharonian et al. [101] after finding excess photons (with a statistical significance of $4\mathsf{\sigma }$) at energies $\mathrm{E}>730\mathrm{GeV}$ coincident with the position of the radio galaxy using data taken with HEGRA [102]. VHE $\gamma$-ray emission from M 87 was detected by H.E.S.S. with a statistical significance of $13\mathsf{\sigma }$ [103] and thereby establishing M 87 as the first non-blazar extragalactic TeV source. In addition, the observations providing this first detection of M 87 at TeV energies also provided evidence for statistically significant spectral variability on the timescale of days. This early detection strongly constrained the size of the $\gamma$-ray-emitting region and challenged most scenarios of VHE $\gamma$-ray production theorised at that time for extragalactic sources. The VERITAS collaboration also reported the detection of VHE $\gamma$-rays from M 87 showing a correlation between the VHE $\gamma$-ray flux and M 87’s X-ray emission detected by the ASM RXTE observatory [104]. Observations by MAGIC also confirmed the detection of M 87 as a TeV source [105]. Again, these early observations seemed to suggest that the $\gamma$-ray emission was most likely associated with the core of the radio galaxy as opposed to a knot in the jet, for example. However, a detailed multi-wavelength study of M 87 by Abramowski et al. [106] showed two of three detected flares exhibited rapid variability, suggesting $\gamma$-ray emission close to the core and the remaining flare coincident with the brightening of an HST-1 knot in the jet. This study used both the historical VHE observations, as well as new VHE data obtained during several coordinated monitoring campaigns between 2008 and 2010.

Figure 3 shows a composite image of M 87 viewed across a broadband spectrum from radio to $\gamma$-ray frequencies. The image demonstrates the huge scales between images of the black hole at the heart of M 87 versus images encompassing the whole jet and extent of the radio galaxy itself. Figure 3 should also provide the reader with some insight into the challenges faced by astronomers when trying to understand and model the mechanisms driving the non-thermal emission from radio at one end of the electromagnetic spectrum to $\gamma$-rays at the other. Current instruments, at $\gamma$-ray energies, are not able to spatially resolve emission from the core and the jet.

3.4. 3C 264

The fourth radio galaxy detected at VHE energies is 3C 264. This was detected by VERITAS [20] between February 2017 and May 2019. At a distance of $\sim 93\mathrm{Mpc}$, 3C 264 is the most distant radio galaxy detected at TeV energies, but is also one of the brightest and closest radio-optical jets known [108]. The excess $\gamma$-rays coincident with the location of 3C 264 were detected with a statistical significance of $7.8\mathsf{\sigma }$ above the background. The detection of 3C 264 came from deep observations comprising some 57 h of data and is believed to have occurred during an elevated flux state during the 2018 observations [109]. Like M 87, 3C 264 exhibits superluminal motions on kiloparsec scales. When comparing the jet structure, 3C 264 closely resembles M 87. Both galaxies have one-sided FR-I-type jets with a kinetic luminosity of $\sim {10}^{43.8}\mathrm{erg}{\mathrm{s}}^{-1}$ [40]. In addition, their jets display quite similar morphological traits, for example they both have multiple knots. Indeed, a one-to-one correspondence between radio and optical features was found in the jet of 3C 264 [110]. Like the other TeV-detected radio galaxies, 3C 264 exhibits a limb-brightened jet [111] indicating the possibility of $\gamma$-ray emission via spine–sheath-structured jet models. Using HST observations, Meyer et al. [112] found apparent speeds as high as ${\mathsf{\beta }}_{\mathrm{app}}=(7.0\pm 0.8)\mathrm{c}$, exceeding the apparent speeds measured in the jet of M 87 highlighted in Section 3.3 above. Thus, 3C 264 could be considered as a more distant analogue of M 87, based on both its beamed and unbeamed radio emission, as well as its kinematic profile. 3C 264 is most likely observed at a smaller angle to the line-of-sight however.

The VHE flux observed by VERITAS during 2017–2019 shows slight variability, which suggests the detection did not result from a strong flare, but instead from a modestly enhanced flux state over a period of time. At GeV energies, 3C 264 shows some variability on six-monthly timescales, but compared to other Fermi-LAT-detected radio galaxies such as NGC 1275 or even 3C 120, it does not exhibit strong variability on these timescales [113]. At VHE energies, Archer et al. [109] reported 3C 264 showing variability on annual and monthly timescales. When comparing HST images of the jet in 3C 264 between 1994 and 2014, Meyer et al. [112] found two superluminal knots to be on a collision course. Indeed, the authors reported that these two knots began interacting during 2014 with the collision still ongoing today. Such an ongoing collision between two knots could be responsible for the observed VHE $\gamma$-ray emission. However, Archer et al. [109] concluded that the unresolved core was the most likely source of the enhanced VHE flux, particularly as there is no evidence for any notable changes in any of the contemporaneous high-resolution Chandra or HST images. The authors also stressed that no large changes in the core flux were seen at lower frequencies.

Figure 4 shows the broadband SED of 3C 264 compared to that of M 87. Archer et al. [109] were able to model the SED of 3C 264 successfully using a single-zone SSC model, although they did point out that a better fit could be achieved using a multi-zone model. However, from simple visual inspection, it is immediately noticeable that 3C 264 has a relatively high-frequency synchrotron peak near the X-ray frequency band ($\sim {10}^{17}\mathrm{Hz}$), which appears to be somewhat unusual for a radio galaxy, as discussed in Section 2. In another study of 3C 264, the broadband SED and jet power were found to be reasonably well described by simple one-zone SSC leptonic models, but only when the high-energy component was associated with a second emitting region located at the end of the acceleration region ($\sim 11\mathrm{pc}$) in the spine or sheath of the jet. Furthermore, when the SED was modelled assuming synchrotron emission from the core, the emission at higher energies was significantly underpredicted [111].

3.5. PKS 0625-35 and IC 310

The two remaining non-blazar TeV-detected AGNs, PKS 0625-35 and IC 310, display both radio galaxy and blazar characteristics. TeV $\gamma$-rays (above an energy threshold of $\sim 250\mathrm{GeV}$) from PKS 0625-35 were first detected in 2012 by H.E.S.S. with a statistical significance of $\sim 6\mathsf{\sigma }$ above the background [114]. Located at a distance $\mathrm{d}\approx 235\mathrm{Mpc}$ ($\mathrm{z}=0.055$), PKS 0625-35 is much further away relative to other TeV-detected radio galaxies. Radio observations revealed a one-sided jet that extends to a distance of $\sim 143\mathrm{pc}$ and exhibits superluminal apparent velocities of $\sim 3 \mathrm{c}$ [115,116,117]. Evidence for blazar characteristics was found in optical observations where the presence and intensity of the [OIII] emission line was consistent with those of other BL Lac objects [118]. In further work by the H.E.S.S. Collaboration et al. [119], it was suggested that PKS 0625-35 displays other unusual characteristics for a radio galaxy such as the synchrotron peak lying between ultraviolet and X-ray frequencies, which is typical for BL Lac objects. Furthermore, X-ray variability studies constrain the maximum possible misalignment to $\mathsf{\theta }\approx {15}^{\circ }$, and the authors therefore argued that the object be classified as a blazar or moderately misaligned blazar.

IC 310 is located in the Perseus cluster along with NGC 1275 and is at a distance of $\mathrm{d}\approx 80\mathrm{Mpc}(\mathrm{z}=0.019)$ from Earth. MAGIC made a serendipitous discovery of this head–tail radio galaxy during 2009/2010 while observing NGC 1275 located in the same field of view ($\sim 0.6^{\circ }$ away). An initial detection corresponding to a statistical significance of $7.6\mathsf{\sigma }$ above the background was obtained from 20.6 h of stereoscopic observations [120,121]. The observed energy spectrum as seen by MAGIC was relatively hard with a spectral index $\mathsf{\Gamma }=-2.00\pm 0.14$. MAGIC estimated the average flux above $300\mathrm{GeV}$ to be $(3.1\pm 0.5)\times {10}^{-12}{\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}$, which corresponds to $\sim 2.5\%$ of the Crab Nebula flux [121]. An extensive multi-wavelength campaign observing IC 310 was conducted between 2012 and 2013. After data quality cuts, MAGIC acquired an additional ∼39 h on IC 310 including ∼3.7 h when MAGIC detected a bright flare. Thus, MAGIC was able to separate out this flare data to estimate energy spectra for two distinct states; during the low state, the energy spectrum is best described by a standard power-law model with normalisation at $1\mathrm{TeV}$ of ${\mathrm{N}}_0=(3.12\pm 0.91)\times {10}^{-13}{\mathrm{TeV}}^{-1}{\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}$ and spectral index $\mathsf{\Gamma }\approx (2.36\pm 0.30)$, and during the bright flare, the energy spectrum reached a peak normalisation at $1\mathrm{TeV}$ of ${\mathrm{N}}_0=(44.2\pm 2.1)\times {10}^{-12}{\mathrm{TeV}}^{-1}{\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}$ with spectral index $\mathsf{\Gamma }\approx (1.51\pm 0.06)$ [56].

Therefore, at TeV energies, the spectral index appears to display harder-when-brighter behaviour. Contemporaneous soft X-ray data taken during this multi-wavelength campaign appear to show a similar harder-when-brighter trend, and the authors concluded that this relationship is consistent with an SSC emission scenario, successfully fitting a single-zone SSC model to the broadband SED during the non-flaring state. Similarly, at GeV energies, IC 310 appears highly variable in both flux and spectral state [122].

As mentioned above, the classification of IC 310 is still the subject of some debate. Originally classified as a narrow-angle tail (NAT) radio galaxy, early $\gamma$-ray detections of IC 310 categorized this object as an AGN of unknown type before being assigned as an HBL in the Fermi-LAT 2nd AGN catalogue. However, the discovery of a one-sided core–jet structure with blazar-like, beamed radio emission oriented along the kiloparsec tail was not consistent with the NAT classification. Instead, IC 310 appears to be a low-luminosity FR-I radio galaxy that is moderately misaligned $\mathrm{i}\sim$10${}^{\circ }$–20${}^{\circ }$ still revealing its BL Lac-type central engine [123]. Moreover, Figure 5 shows new low-frequency radio observations that provide a resolved view of the bent jets and have led to a new estimation of the misalignment to be $\mathrm{i}\sim 8^{\circ }$. This evidence means that the $\gamma$-rays most likely originate from a small region close to the SMBH and are beamed toward us along the line-of-sight, within a region of the jet before it is bent into the narrow tail-like structure [55]. Furthermore, rapid minute-scale variability seen at TeV energies is consistent with such an interpretation.

3.6. Radio Galaxies with Water Cherenkov Detectors

In addition to next-generation IACTs such as CTA, other techniques for exploring the VHE space, such as using water Cherenkov detectors (WCDs), were used in the past such as the VHE surveys conducted by Milagro [124,125] and ARGO [126]. At present, the High Altitude Water Cherenkov Observatory (HAWC) [127] located in central Mexico and a hybrid array called the Large High Altitude Air Shower Observatory (LHAASO) [128] in Sichuan province of China are operational, and another future observatory using WCDs called the Southern Wide-field Gamma-ray Observatory (SWGO) [129] (formerly the Southern Gamma-ray Survey Observatory) is also proposed. Observatories employing WCDs typically have a larger angular resolution compared to IACTs. However, compared to IACTs, their sensitivity for larger extended sources is better, and they are able to operate both day and night under all weather conditions with a large instantaneous field of view $>2\mathrm{sr}$. HAWC’s angular resolution (defined as the $68\%$ photon containment radius) is both energy- and zenith-angle-dependent and ranges from $0.1^{\circ }$–1${}^{\circ }$ [127]. Furthermore, WCDs such as HAWC (and the proposed SWGO) are more sensitive to sources that transit directly overhead and also to sources with a hard energy spectrum. Thus, HAWC’s energy threshold for detection is source-dependent. Naturally, this imposes a selection bias on the observable radio galaxies, but may be extremely useful in detecting any steady VHE $\gamma$-ray emission from radio galaxies, such as Cen A, that display spectral hardening in the GeV/TeV band and also display extended emission. Unfortunately, Cen A is not visible to HAWC, so undoubtedly will be a priority target for SWGO.

The relative closeness of radio galaxies means smaller $\gamma$-ray photon attenuation on the extragalactic background light (EBL), making these potentially great targets for studying the far-infrared band of the EBL above $\sim 10\mathrm{TeV}$. In the latest results from the HAWC survey of active galaxies, six radio galaxies with favourable transits for HAWC were considered: NGC 1218, IC 310, NGC 1275, 4C +39.12, 3C 264, and M 87. At GeV energies, all of these radio galaxies have hard spectra; in the case of NGC 1275, this is true for active states. From live data spanning a period of approximately 4.5 years, HAWC sees only marginal evidence of $\gamma$-ray emission from M 87 with a statistical significance of $\sim 3.6\mathsf{\sigma }$ at $\mathrm{E}>0.5\mathrm{TeV}$ [130]. Similarly, M 87, NGC 1275, and IC 310 are all nearby sources within the LHAASO field of view [131], and since the observatory was only completed in 2021, observational findings for these sources will undoubtedly be published in the near future. Finally, even if WCDs require a longer integration time to reach the accepted statistical threshold to claim a detection, these types of detectors have a part to play in improving our understanding of radio galaxies, particularly if SWGO is able to bring advances that greatly improve the achievable angular resolution of current WCDs.

4. Discussion

The above sections have provided some context for radio galaxies within $\gamma$-ray astronomy by highlighting some key characteristics of these objects relating to wider blazar studies, as well as highlighting key findings from multi-wavelength observations of radio galaxies detected at VHE. This section will address some of the themes highlighted in Section 2.3 by discussing the principal findings from these observations and what implications if any these have for future observations of these fascinating extragalactic objects.

4.1. Radio Galaxy Environment

The first and probably most obvious finding from the TeV observations is that all four radio galaxies detected at VHE energies have low-power FR-I-type jets. This may seem counter-intuitive for accelerating particles to VHE, but equally, this may tell us that FR-I-type jets live in denser environments and that it is the interaction of the jet with these dense environments, rather than the intrinsic power of the jet, that provides the conditions for accelerating particles to VHE. For example, it is suggested that changes in the direction and flux of NGC 1275’s bright C2 component (as shown in Figure 1) during August–September of 2015 are consistent with hydrodynamical simulations of jets in a clumpy ambient medium [49]. Furthermore, observations of 3C 264 show that the radio galaxy is located within a large region exhibiting low-surface-brightness emission, which could indicate that the radio galaxy resides in a dense environment [111]. Therefore, if radio monitoring reveals potential jet interactions/collisions with dense clumps in the environment (as in the case of NGC 1275 [49,64]), this could indicate an increased likelihood for enhanced $\gamma$-ray emission via internal shock models, leading to flares or possibly even orphan flares at TeV energies.

At the time of writing, no high-power FR-II-type radio galaxies are listed on TeVCat [22], but it is reasonable to assume that the current generation of ground-based IACTs have attempted to detect these objects without success. Perhaps this is not completely surprising given that FR-IIs are systematically farther away [132,133] and their VHE $\gamma$-ray photons are more likely to be absorbed by the EBL, coupled with the fact that the current generation of IACTs simply does not possess the necessary sensitivity to detect $\gamma$-rays produced at their jet termination sites. If CTA is able to detect FR-II radio galaxies, this will expand our understanding of these objects further and provide better insight into the emission mechanisms at play, the emission locations, and perhaps, even their environments.

4.2. Kinematic Studies and Classification Considerations

Another obvious question one might ask is whether the TeV-detected radio galaxies are indeed radio galaxies in the “true sense”, particularly as most of these objects exhibit properties such as a one-sided jet, fast variability, and superluminal motion, all characteristics of blazars, but less extreme in these cases. This of course is an argument of classification, and as already highlighted above, classification is often a product of the observed wavelength/frequency. Given that TeV emission from these extragalactic AGNs requires a relatively high Doppler factor, any blazars with jets not aligned closely to the line-of-sight are designated misaligned blazars. Thus, in the context of VHE $\gamma$-ray observations, this includes both “true” radio galaxies, as well as quasars, but also less extreme blazars falling near the radio galaxy boundary. There is no absolute angle used to precisely define the exact boundary of when a blazar becomes a radio galaxy; instead, if the jet is inclined with respect to the line-of-sight at angles $\mathsf{\theta }\gtrsim {10}^{\circ }$, then it is generally accepted that these are not BL Lacs, nor FSRQs in the strict sense. This corresponds to the critical angle used to distinguish the division between BL Lacs and FR-I-type radio galaxies [134].

In the case of IC 310, perhaps our understanding of the true nature of these objects is not complete. Perhaps $\gamma$-rays are emitted in regions that are closely aligned to our line-of-sight where large Doppler factors are attainable. What this result shows us is that high-resolution parsec- and kiloparsec-scale radio images are needed (in particular for radio galaxies that display rapid VHE variability) in order to better determine line-of-sight angles. Indeed, blazar multi-wavelength monitoring programs such as the MOJAVE survey program [135], the BEAM-ME program (successor to the VLBA-BU-BLAZAR program) [136], and the TANAMI program [59] are crucial as we enter a new era of observations with new instruments coming online soon, such as the James Webb Space Telescope (JWST) [137], the Square Kilometer Array (SKA) [138], and CTA [26], or planned, such as the Next-Generation Very Large Array (ngVLA) [139], as well as successors of Chandra [140] and Fermi-LAT.

Programs such as those highlighted above include many of the Fermi-LAT-detected radio galaxies and allow regular monitoring of radio amplitudes that enable correlation studies between radio and $\gamma$-ray frequencies. In addition, these monitoring programs can provide polarisation data that underline the magnetic field structure of the radio-emitting regions. All of this is crucial information that is needed when attempting to associate $\gamma$-ray emission with a particular region or emission process for example. Additionally, given the relatively poor angular resolution of $\gamma$-ray instruments, both from space and ground, very-long-baseline interferometry (VLBI) monitoring programs play a vital role in charting the evolution of parsec-scale radio emission, which again enables the study of any relationship between radio and $\gamma$-ray emission.

A further point for consideration is that we might only detect these four radio galaxies at TeV energies due to their proximity, i.e., even though they are misaligned, the combination of their proximity coupled with strong-enough Doppler boosting means they are detectable at TeV energies. However, it should be noted that at least two other Fermi-LAT-detected radio galaxies reside closer than 3C 264, and these are not detected at TeV energies. Furthermore, even though PKS 0625-35 displays blazar characteristics, it appears misaligned nonetheless, and there are many radio galaxies undetected at TeV energies closer than this object.

One last observation is that all four of the TeV-detected radio galaxies exhibit limb-brightened jets. Perhaps this is an important feature that could indicate an increased likelihood for gamma-ray emission via spine–sheath-structured jet models, where a small-enough beaming angle is maintained to provide sufficient Doppler boosting.

4.3. Emission Sites and Variability

The detection of synchrotron X-ray emission from the jets of Cen A and M 87 implies the presence of electrons at and above TeV energies, and it is possible that the TeV $\gamma$-ray emission is produced (or partly produced) by inverse-Compton scattering of photons by the high-energy electrons producing the synchrotron emission on kiloparsec scales [90]. Indeed, the detection of extended gamma-ray emission in Cen A with Fermi-LAT [141] coupled with a lack of evidence for strong variability means that an extended leptonic origin for its emission cannot be ruled out. As highlighted in Section 3.2, the latest H.E.S.S. observations of Cen A suggest that a major part of the VHE emission arises on large scales [92]. Moreover, the H.E.S.S. findings are consistent with (and provide the first evidence of) VHE $\gamma$-ray emission resulting from the inverse-Compton scattering of low-energy photons by ultra-relativistic electrons in the jet. This inverse-Compton emission potentially contributes toward the spectral hardening (discussed above) observed in Cen A and also suggests that radio galaxy jets are potentially a steady source of VHE gamma-rays.

However, in the case of M 87, the picture is not so clear. As highlighted above, two out of three TeV-detected flares exhibit rapid variability with timescales on the order of approximately half a day, which suggests that $\gamma$-ray emission originates from a region close to the core. Yet, the remaining TeV-detected flare was correlated (using multi-wavelength data) with the HST-1 knot located downstream in the jet of M 87 [106].

Fermi-LAT observations of M 87 found flux variability on six-monthly timescales at energies below $\mathrm{E}<10\mathrm{GeV}$ [142]. Similarly, in a study of 26 radio galaxies at GeV energies using Fermi-LAT data [113], only 7 were seen to exhibit any statistically significant flux variability on six-monthly timescales. Of these seven, two (3C 264 and PKS 0625+35) were detected at TeV energies. On the other hand, NGC 1275 has displayed rapid variability at GeV energies during flares with flux-doubling timescales of $1.21\pm 0.22\mathrm{h}$ and $8.03\pm 1.24\mathrm{h}$ reported [143]. Such flux-doubling timescales are more usual for blazars rather than radio galaxies, but as already alluded to before, radio galaxies are weak $\gamma$-ray sources and the sensitivity of the current space- and ground-based $\gamma$-ray observatories limits us to seeing variability in the brightest radio galaxies.

4.4. Emission Mechanisms and Modelling

When modelling the broadband spectral energy distributions of blazars and indeed radio galaxies, often a single-zone synchrotron self-Compton (SSC) jet model is used to provide a reasonable description of the emission mechanisms. However, for radio galaxies such as Cen A with very modest Doppler boosting, it is clear that the single-zone SSC models do not adequately describe the VHE $\gamma$-ray emission [39,144,145]. Yet, for other radio galaxies such as 3C 264, its broadband emission spectrum can be described by a single-zone SSC model, as shown in Figure 4, but is probably better fit using multi-zone models [109].

When modelling the broadband SEDs of radio galaxies, the jet misalignment value used (or assumed) has a profound impact on the model used to describe the emission. Assuming TeV-detected radio galaxies displaying rapid variability are truly misaligned, then exotic scenarios such as “mini-jets in jets” [146] or “black-hole lightning” [147] are necessary in order to produce VHE $\gamma$-rays. In-depth discussion of all the models applicable to radio galaxies along with the intricacies of applying those models to SEDs is beyond the scope of this work. For a comprehensive review of blazar modelling in the context of TeV extragalactic astronomy please, see [16], and for an earlier review of radio galaxies at VHE energies that includes discussion of the models relevant to radio galaxies, please see [148]. Instead, what is important to take away is that radio galaxies are excellent targets for multi-wavelength observations, which provide us with an opportunity for correlation studies across wavebands in order to single out a preferred model (or set of models) that explain $\gamma$-ray emission from these objects.

4.5. Radio Galaxies as Cosmic Ray Sources

It is well established that cosmic ray particles permeating our galaxy cannot be traced back to their origins (except perhaps at the highest energies) as they are deflected by magnetic fields. We know that the origin of cosmic rays above a few EeV must be extragalactic, but as of yet, no specific type of source is positively identified as the primary cosmic ray contributor [149,150]. However, there is growing evidence in the Pierre Auger Observatory data (independent of the inconsistency arguments highlighted regarding the connection between ultra-high-energy cosmic rays (UHECRs) and AGNs) that Cen A is a potential source of UHECRs and neutrinos [150,151,152,153,154]. Indeed, the work by Koers and Tinyakov [155] highlights the possibility of Cen A being an exceptionally efficient neutrino source. This is important because if radio galaxies like Cen A are shown to be neutrino sources and can be correlated to TeV $\gamma$-ray emission in these objects, this provides very strong evidence that hadronic or lepto-hadronic emission processes are responsible.

Numerous authors have considered radio galaxies as candidates for neutrino events [156,157,158,159], and even FR-0-type radio galaxies were considered as well [160]. FR-0-type radio galaxies are a relatively new class of radio galaxy that display a different radio morphology to the observed FR-I- and FR-II-type radio galaxies. They lack the large-scale extended radio structures of the latter two classes, instead displaying jets that are both weak and short, extending only a few tens of kiloparsecs [161]. FR-0s share similar properties to those of FR-Is except the ratio of their core to total emission is $\sim 30$-times higher than that of FR-I-type radio galaxies [162]. Despite the abundance of FR-0s in the local universe ($\mathrm{z}<0.05$), these are difficult objects to study because of their weak and compact nature [163]. No TeV emission is currently detected from an FR-0-type radio galaxy, but there are many open questions relating to these radio galaxies, not least the fact that at least one, 4C +39.12, was detected with Fermi-LAT at GeV energies displaying a relatively hard spectrum similar to that of 3C 264.

There is also evidence that a non-negligible positional association exists between AGNs with strong parsec-scale cores and neutrino events. This implies that AGNs with strongly Doppler-boosted jets, as opposed to radio galaxies, are an important source of neutrinos [164]. In spite of this, radio galaxies are important multi-messenger targets, in particular the closest radio galaxies Cen A and M 87.

As already discussed above, single-zone SSC models can be inadequate to explain the observed $\gamma$-ray emission from radio galaxies. The rising TeV components seen in the broadband SEDs of Cen A (and IC 310 during a flaring state) have led to suggestions of either a hadronic origin or leptonic origin with multiple electron distributions [47,165]. Using Fermi-LAT data, Rulten et al. [113] analysed 26 AGNs classified as radio galaxies in the Fermi-LAT 4FGL catalogue [48]. The goal of this work was to search for Cen A-like spectral features; no other radio galaxies exhibited a Cen A-like break in their spectral energy distributions. Whether other radio galaxies possess such a break is still an open question because, given Fermi-LAT’s limited sensitivity at energies above $100\mathrm{GeV}$ coupled with the poor sensitivity of the current generation of IACTs for detecting weak $\gamma$-ray sources, the possibility still exists that other radio galaxies do indeed exhibit Cen A-like spectral breaks, but we just happen to see Cen A’s because of the AGN’s proximity to us. Discovery of such hardening amongst a greater number of radio galaxies would strongly implicate radio galaxies as a major source of cosmic rays within the universe.

4.6. Radio Galaxies with Future Instruments

It is imperative that TeV monitoring of radio galaxies continues, both with the current generation of IACTs, as well as with CTA when that instrument comes online. Deeper observations are necessary in order to better sample the energy spectra of these objects and, importantly, detect them in different states. Ideally, multi-wavelength campaigns should be run in order to capture data simultaneously, particularly important for correlating $\gamma$-ray emission with knots downstream in the jet. For example, a quasi-simultaneous multi-wavelength campaign monitoring M 87 produced the most comprehensive multi-wavelength spectrum of the radio galaxy. Combining data from radio, optical, ultraviolet, X-ray, and gamma-ray wavebands, a near-complete broadband picture was produced during a historically low state for the AGN. In this work, the authors concluded that a structured jet is necessary to explain the observed spectrum of M 87 [166]. From modelling of the spectrum, the authors were able to exclude simultaneous $\gamma$-ray emission from inverse-Compton emission originating in the same region that produces the mm-band emission seen by the EHT. Moreover, the authors also concluded that $\gamma$-rays originating from the inner-jets (inward of HST-1) are only produced in the presence of strong particle-dominated regions [166]. However, they were not yet able to exclude direct synchrotron emission from accelerated protons (and secondaries). Such campaigns allow better modelling of radio galaxy broadband SEDs and provide the necessary information to favour a particular emission scenario for different AGN states.

Furthermore, the unequalled performance of CTA (relative to the current generation of IACTs) will allow $\gamma$-ray astronomers to investigate VHE emission from Cen A, for example with much better spatial and temporal resolution and sensitivity than ever before. Undoubtedly, this will help to better constrain the ratio of $\gamma$-ray emission between the core and extended regions of this particular radio galaxy.

New observatories such as SKA, JWST, and CTA will bring a new era of astronomy that will provide coverage of radio galaxies across the electromagnetic spectrum from radio to $\gamma$-ray frequencies, advancing our understanding of AGN physics. For example, at $\gamma$-ray energies, we know radio galaxies and blazars in general exhibit variability. The current generation of radio telescopes have limited survey speed, restricting the number of AGN with detailed variability studies to approximately a thousand of the brightest objects. Phase 1 of SKA will provide variability information at a sub-day-scale cadence for hundreds of thousands of AGN [167]. Such coverage will enable vastly improved multi-wavelength variability studies of radio galaxies between CTA and SKA, and perhaps facilitating the first insights into a radio-VHE connection, which has so far proven elusive. Furthermore, SKA is likely to play a crucial role in counterpart identification and classification of the many unidentified Fermi-LAT extragalactic $\gamma$-ray sources [168]. One of the most exciting discovery spaces in this new era regarding radio galaxies will perhaps come from the very-long-baseline interferometry (VLBI) capability of SKA complementing ngVLA advances, which is expected to provide ten-times the continuum imaging sensitivity of e-MERLIN [169] at $5\mathrm{GHz}$ in $1/10\mathrm{th}$ the integration time, as well as achieving baselines of $1000\mathrm{km}$ [139]. Being able to conduct astrometry at the microarcsecond level out to tens of kiloparsecs will enable high-precision study of the proper motions of features inside radio galaxy jets [170]. Furthermore, with full linear and circular polarization capabilities, ngVLA will be able to study the magnetic field structure of radio galaxy lobes and hotspots with high precision, providing a better understanding of both the particle acceleration and jet deceleration and how this relates to $\gamma$-ray emission [139]. For example, do $\gamma$-ray flares occur after the radio emission peak and thus potentially indicating a jet “loading up” before emitting $\gamma$-rays, such as the creation of an emerging radio knot that interacts with an internal shock downstream in the jet [112]? Likewise, if no correlations are seen between $\gamma$-ray flares and knot brightening and the radio emission lags any $\gamma$-ray emission, this could be evidence that the primary $\gamma$-ray emission zone resides very close to the SMBH.

5. Conclusions

$\gamma$-ray observations of radio galaxies over the past decade or so have shown these objects to be fascinating, exhibiting ultra-fast variability or displaying relatively hard spectral energy distributions. These observations have also highlighted the need for instruments with improved sensitivity and resolution in order to better classify objects or to better resolve their morphological and timing characteristics, for example.

Although our understanding of radio galaxies as $\gamma$-ray emitters has advanced in the past decade or so, there still remains many outstanding questions, and a number of these were highlighted in Section 2.3. These relate to the radio galaxy environment, kinematic studies, classification considerations, $\gamma$-ray emission sites, $\gamma$-ray variability, $\gamma$-ray emission mechanisms, $\gamma$-ray emission modelling, and radio galaxies as cosmic ray sources. From the discussion in Section 4, it is clear that many of these outstanding questions will need to be addressed with multi-wavelength studies, employing a number of different instruments.

Looking forward, the near-term horizon appears very promising with many new and improved instruments arriving, such as JWST, SKA, and CTA, to name a few. They will provide astronomers with many opportunities for future observations of radio galaxies. What is more, the unprecedented coverage and resolution of these instruments will help to advance our understanding of AGN science, for example by identifying $\gamma$-ray emission sites with better precision, by constructing high-quality apparent velocity profiles over all jet scales or through deeper observations with well-sampled broadband spectra to enable better modelling of $\gamma$-ray emission.