THE COMPLEX NORTH TRANSITION REGION OF CENTAURUS A: A GALACTIC WIND

The radio source Cen A was first identified with the galaxy NGC 5128 by the radio emission from its Outer Lobes (Bolton et al. 1949), which extend ∼600 kpc end-to-end on the sky (e.g., Junkes et al. 1993; Feain et al. 2011). The outer radio lobes are also detected as extended γ-ray sources (Abdo et al. 2010), thought to result from inverse Compton scattering of cosmic microwave background photons. Eilek (2014) argues that the dynamical age of Cen A's outer lobes is on the order of ∼1 Gyr (assuming the lobes lie at some finite angle to the sky plane), but that they must have been last energized no more than ∼30 Myr ago to keep shining in radio and γ-rays. Any energy supplied to the Outer Lobes by the inner galaxy must, of course, have moved through the middle (∼50 kpc scale) regions which we study in this paper.

The massive black hole in the nucleus of NGC 5128 is currently active, as indicated by parsec-scale radio jets, a northern few-kiloparsec-scale radio and X-ray jet, and inner lobes extending ∼7 kpc NE and ∼5.5 kpc SW of the galactic core (e.g., Burns et al. 1983). Both brightness and proper motion asymmetries of the inner jet (e.g., Tingay et al. 2001) and Faraday depolarization of the South Inner Lobe (Clarke et al. 1992) suggest that the northern jet is approaching us while the South Inner Lobe is aimed away from us. X-ray observations of a shock driven into the interstellar medium (ISM) by the Southern Inner Lobe show the Inner Lobes are only ∼1–2 Myr old (Croston et al. 2009, Paper 1). This suggests the AGN has only recently restarted as a collimated outflow. The power currently produced by the AGN, several × 10⁴³ erg s⁻¹, is interestingly close to that required to power the outer lobes over their ∼1 Gyr lifetime (Eilek 2014, Paper 1).

The iconic dust lane in NGC 5128 is part of a thin (∼250 pc), multiply warped gas and dust disk. It is a site of active star formation. UBV photometry of young stars (van den Bergh 1976; Dufour et al. 1979) and young clusters (Minniti et al. 2004), together with Hα and [Nii] imaging of Hii regions (Bland et al. 1987) and far-infrared (FIR) spectroscopy of the disk (Unger et al. 2000), indicate ongoing active star formation in the disk that has continued for at least 50 Myr. The disk is thought to result from a merger or stripping interaction, between a large elliptical galaxy and a smaller gas-rich system, that occurred 250–750 Myr ago (Quillen et al. 1993; Sparke 1996; Struve et al. 2010).

The North Transition Region (NTR) is much brighter in the radio than the equivalent region to the south. The radio-bright northern region, which we studied in Paper 1, is called the NML in the literature. In addition to the extended, diffuse radio emission, Morganti et al. (1999) identified a linear radio structure in the NML, and suggested that it might be a "large-scale jet" connecting the inner and outer radio lobes. However, in Paper 1 we showed that the linear feature does not appear to be a radio jet. Rather, it is part of a knotty, radio-bright ridge embedded in the larger, overpressured, diffuse structure of the NML.

The NML seems to have pushed aside the hot ISM of NGC 5128; X-ray images reveal a cavity north of the galaxy, which is approximately coincident with the NML (e.g., Kraft et al. 2009; Paper 1). We showed in Paper 1 that the radio-loud NML also needs ongoing energization: without a driver, the NML will fade in ≲20 Myr due to expansion and high-frequency synchrotron losses.

In addition to unusual radio emission and the disturbed galactic ISM, the NTR hosts an X-ray-loud filament with embedded clouds, optical emission-line filaments, recently formed stars, and a broken ring of cold atomic and molecular gas—all in close proximity. As noted in Paper 1, we refer to this complex mixture of radio emission, disturbed multiphase ISM, and star formation as the "weather system."

A trail of emission line filaments, young star clusters, and ultraviolet emission (UV, λ1344–3000 Å) stretches outward from the galaxy. Most previous optical work has focused on two bright regions close the galaxy, known as the "Inner" and "Outer" Filaments. The Inner Filament is ∼8 kpc from the core, while the Outer Filament is ∼15 kpc from the core. Both Filaments contain ionized gas, with high internal/turbulent velocities (≳100 km s⁻¹, (e.g., Graham 1998), and so should not last more than 10–15 Myr. Both regions also contain young stars, with ages ≲10 Myr (e.g., Mould et al. 2000, Crockett et al. 2012), thus recent and ongoing star formation. In this paper, we show that an extended "weather ribbon" of UV and Hα emission extends much further, reaching from the North Inner Lobe to at least ∼35 kpc from the galaxy's core (see also Graham & Price 1981).

The knotty, radio-bright ridge reported in Paper 1 sits close to the SE edge of the NML. It also sits very close to a similar, knotty, X-ray bright ridge (Feigelson et al. 1981; Kraft et al. 2009). We showed in Paper 1 that the radio knots and X-ray knots are spatially close to each other, but not exactly coincident; both sets of knots are ∼1 kpc in size and—if they are diffuse plasma—are at ≳20 times higher pressure than their surroundings. In this paper we show that the knotty radio/X-ray ridge is also nearly coincident with the outer part of the string of young stars and emission-line filaments mentioned above.

The Transition Regions also contain clouds of dusty, cold gas (e.g., Schiminovich et al. 1994), which form a broken ring orbiting NGC 5128. CO emission and warm dust have been detected from the densest parts of the Hi clouds (Charmandaris et al. 2000; Auld et al. 2012). One edge of the outermost Hi cloud appears spatially coincident with the Outer Filament and with the radio and X-ray ridges. Osterloo & Morganti (2005) found that the cloud edge shows anomalous, turbulent velocities, and speculate that gas stripped from the cloud provides seed material for the emission line filaments NE of the cloud.

There is near-consensus in the literature that an outflow must be causally related to the emission line clouds and young stars in the weather ribbon NE of the galaxy. Many authors assume the narrow feature seen by Morganti et al. (1999) is a collimated jet which has energized the line emitting clouds and induced star formation in the ribbon. Detailed models include weak shocks in dense clouds photoionizing the ambient gas (Sutherland et al. 1993) and shocks causing cloud fragmentation and collapse leading to star formation (e.g., Fragile et al. 2004; Croft et al. 2006). However, we showed in Paper 1 that the linear feature does not have the properties of a radio jet. Although the lack of a detected jet seems to challengethe jet-driven shock models, we argue in this paper that a broad wind outflow can play the same role as a narrow jet in energizing the clouds and inducing star formation.

We noted in Paper 1 that models proposed to explain the radio-loudness of the NML have either assumed the presence of a radio jet through the region, or attributed the NML to a slowly rising, bouyant bubble. We argued there that none of these models can explain all observed properties of the region—in particular the short-lived nature of several radio-related phenomena, the need for ongoing energy input to both Outer Lobes, and the differences in radio properties between the NTRs and STRs. We suggest in this paper that a broad wind outflow can explain the radio properties of the NML, just as it can energize the emission-line clouds and drive star formation in the weather ribbon.

Both optical and FUV data reveal a complex mix of young stars, active star formation, and emission-line gas extending throughout the ∼30 kpc eastern part of the NTR. Figure 4 shows a side-by-side comparison of Hα and FUV images of the weather system. The left panel shows part of an Hα image, taken by I. Evans and A. Koratkar, using the MOSAIC2 camera on the 4 m Blanco telescope at CTIO. The right panel of Figure 4 shows the same field, extracted from our full-field FUV image (Figure 1). The FUV and Hα emission appear to be coincident, with the same extent and the same morphology.

Zoom In Zoom Out Reset image size

The FUV/Hα ribbon extends over 25' (28 kpc) starting near the galaxy center and falling below our detection levels ∼33' (37 kpc) from the galaxy center. The comparison makes it clear that both Hα and FUV emission extend well beyond the bright emission line filaments previously known (Inner and Outer Filaments, indicated in Figures 2 and 3). The FUV/Hα ribbon contains the well-studied Inner and Outer Filaments, as well as Hα emission regions farther away from the galaxy which were examined by Graham & Price (1981).⁷ FUV and Hα emission, apparently part of the same large structure, also extends inward from the fields shown in Figure 4 toward the galaxy center (starting ≲7' (8 kpc) from the galaxy center, as can be seen in Figure 1).

The gas in the Inner and Outer Filaments, and in a few brighter regions in the outer ribbon, has a complex velocity structure (Graham & Price 1981; Morganti et al. 1991; Graham 1998). The velocity of the filaments in the line of sight is ∼300–400 km s⁻¹ away from the galaxy and toward us. In addition, turbulent internal velocities, of hundreds of km s⁻¹, are seen in the filaments, as well as gradients of ≳200 km s⁻¹ over only few hundred parsecs. Although no spectroscopy of fainter parts of the FUV/Hα emission has been published, R. G. Sharp (2013, private communication) reports that the line-emitting gas just outside the Outer Filament is largely constrained to tight filaments in space and tight ranges in velocity.

We are not aware of any discussion in the literature on the origin of these emission-lines' high velocities in NGC 5128. However, we note that clouds in several starburst galaxies (e.g., NGC 253, NGC 839, M82) have similar velocities. In those objects the clouds are thought to be carried out by a starburst-driven hot wind.

The diffuse, twisted ribbon appears nearly identical when seen in Hα and FUV; it is also visible in a very deep GALEX NUV image (Auld et al. 2012). It is probably a combination of warm line-emitting gas (e.g., Morganti et al. 1991 or Sutherland et al. 1993) and very recent star formation (recent enough that some of the newly formed stars still maintain Hii regions, e.g., Graham 1998; Mould et al. 2000; Crockett et al. 2012). Our current data do not allow us to discriminate between the two possibilities, but it seems likely that both processes are involved.

The Inner and Outer Filaments are known to contain massive stars, some of them very young (≲4 Myr), and many of them exciting local Hii regions (Graham 1998; Mould et al. 2000; Graham & Fassett 2002; Rejkuba et al. 2002; Crockett et al. 2012). Our Hα and FUV images of the outer ribbon show compact bright spots intermixed with the more diffuse ribbon. Some of these spots can be identified in both images; this suggests they are Hii regions around very young stars, embedded in an extended warm ISM. Some FUV-bright regions do not have Hα counterparts (e.g., the extended west arm of the V-shaped Outer Filament).

FUV and Hα emission may arise from diffuse clouds of warm gas; the close structural agreement of the FUV and Hα ribbons suggests cospatial extended emission. Part of the ribbon may be composed of multiphase ISM structures, containing 10⁴ K gas—emitting Hα , [Nii], and [Oiii] lines—and possibly 10⁵ K gas, radiating in [Civ] at 1549 Å (e.g., Sparks et al. 2004; Werner et al. 2013). High sensitivity FUV spectroscopic observations could test this suggestion.

Some of the ionized gas cannot be excited by young stars alone. The well-studied Inner and Outer Filament gas is characterized by a very high ionization level (Graham & Price 1981; Morganti et al. 1991).⁸ Young stars cannot account for this, because they reach at most ∼15,000 K, while some of the ionized gas requires temperatures of ∼10⁵ K.

There is disagreement in the literature about how the gas in the Inner and Outer Filaments has been ionized. Possibilities include photoionization from the AGN (Morganti et al. 1992), flow-induced shocks from a buoyant bubble (Saxton et al. 2001), or shock-produced X-ray photoionization driven by flow of a possible "large-scale jet" (Sutherland et al. 1993). We note that many of the ionizing effects attributed to a nearby jet could equally well be caused by a more diffuse fast flow, such as a galactic wind.

The well studied Inner and Outer Filaments contain very young stars (age ∼4−15 Myr; Mould et al. 2000; Rejkuba et al. 2002, 2011; Crockett et al. 2012). Some of these (e.g., those in the Inner Filament and the eastern branch of the Outer Filament) have associated Hii regions; others (western branch of the Outer Filament) do not. Stars hot enough to ionize Hii regions have typical lifetimes ≲10 Myr.

The FUV emission places less stringent constraints on stellar ages: we expect to see FUV emission from massive O stars for ∼10 Myr and from B stars for up to 100 Myr. Although we have no direct information on the nature of the outer FUV/Hα ribbon, the same lifetime arguments will apply if the ribbon emission comes from recently formed stars.

Individual emission-line clouds are subject to several disruptive effects. To estimate the associated timescales, which depend on cloud parameters, we note that the relatively large (≳100 pc) emission-line "clouds" studied in the Inner and Outer Filaments (e.g., Graham & Price 1981) are actually aggregations of much smaller clouds (Morganti et al. 1991, also the Hα image in the left panel of Figure 4). Following the analysis of Morganti et al. (1991), who studied individual clouds in the Inner Filament and some parts of the V-shaped Outer Filament, we define a "typical" emission line cloud: size a_c ∼ 10 pc, density n_c ∼ 30 cm⁻³ and temperature T_c ∼ 10⁴ K. We assume these numbers are also representative of emission line clouds further out in the weather system.

These clouds cannot last long in their present form, for at least two reasons.

• (1)  
  A typical emitting cloud requires ongoing energization to offset rapid cooling and recombination. The radiative cooling time is extremely short: estimating the cooling function as Λ(T_c) ∼ 10⁻²² erg cm³ s⁻¹ for T_c ≳ 10⁴ K, we see t_cool = k T_c/nΛ(T_c) ∼ 15 yr. This is a short enough timescale for changes to to be observable, if one has sufficient angular resolution. For instance, a 10 pc cloud would subtend ∼0 5, so that secular changes in such clouds would be accessible to current telescopes over the time that the filaments have been studied.
• (2)  
  A typical emission line cloud is at much higher pressure than the surrounding ISM. The cloud has p_c = 2 n_c k T_c ∼ 8 × 10⁻¹¹dyn cm⁻²—significantly higher than the pressure of the local galactic ISM (8.5 × 10⁻¹³ dyn cm⁻², Kraft et al. 2009) If these clouds are not confined, they will expand at their internal sound speed (∼15 km s⁻¹); thus a 10 pc cloud will dissipate in only ∼0.6 Myr (see also Morganti et al. 1991).

Emission-line clouds within both the Inner and Outer Filaments show high random velocities (∼150 − 400 km s⁻¹; Graham & Price 1981; Morganti et al. 1991) within small regions (∼100 pc). These structures will disperse in ≲10–15 Myr unless they are somehow confined. Less is known about velocities in the weather ribbon beyond the Outer Filament. Graham & Price (1981) measured velocities in a few of the brightest regions of the ribbon, and found velocity dispersions similar to those in the Inner and Outer Filaments.

We showed in Paper 1 that if the radio and X-ray knots are diffuse plasma clouds, they are strongly overpressured, ≳20 p_ISM. Thus, they will dissipate quickly if they are not confined. The X-ray knots will survive no longer than ∼3 Myr (Paper 1, also Kraft et al. 2009), while the radio knots will last no more than ∼0.5 Myr (Paper 1).

Alternatively, we noted in Paper 1 that the the X-ray and radio knots might come from star-forming regions. Hard X-rays, if observed, could be an indicator of short-lived, massive stars with lifetimes ∼10 Myr. Because the region has not been observed in hard X-rays, we cannot definitely say that the knots must be sites of recent star formation; but evidence for such young stars elsewhere in the weather ribbon (e.g., Mould et al. 2000; Crockett et al. 2012) suggests that the X-ray/radio knots could contain similarly young stars if they are star-forming regions.

The survival timescales for filaments and knots are much shorter than the time it would take for gas from either the 17 kpc Hi cloud or from the inner galaxy to have reached the weather system. The weather ribbon extends ∼35 kpc from NGC 5128, and ∼18 kpc from the Hi cloud (both distances measured in the sky plane). The optical filaments are moving away from NGC 5128 and toward us at ∼350 km s⁻¹(line of sight velocity; Section 5.1). If the cloud speed in the sky plane is comparable, and if the ribbon material originated in the 17 kpc Hi cloud, it would take ∼50 Myr for the line-emitting gas to travel from the 17 kpc Hi cloud to the outer end of the ribbon. Alternatively, it would take ≳85 Myr for the material in the weather ribbon to travel from the center of the galaxy to its present location.

It follows that any young stars in the outer regions of the weather ribbon must have formed there. They are too young to have been formed close to the galaxy and transported out. Similarly, the emission-line clouds and radio/X-ray knots are unlikely to have lived long enough to survive the journey from the galaxy in their present form. They too must be energized in situ in the weather ribbon.

In Paper 1 we pointed out three lines of further evidence for recent energy transport through, and dissipation within, the NTR. (1) We found there that the NML is at a higher pressure than the surrounding ISM of the galaxy; if it were a static structure, it would expand away in ≲16 Myr. (2) We noted that the detection of the NML at 90 GHz (by WMAP; Hardcastle et al. 2009) requires the electrons radiating at those frequencies to have been re-energized no more than ∼20 Myr ago. (3) Because the Outer Lobes must have been re-energized no more than ∼30 Myr ago to keep them shining in γ-rays as well as radio (Eilek 2014), the necessary energy must have flowed through both Transition Regions at least this recently.

We used the FUV image to characterize the unobscured part of the starburst. The burnt-out display in Figure 1 shows that FUV emission can be detected for ∼15 above and below the dust disk plane, and for a total distance of ∼8' along the dust plane. The disk is quite thin, and warped multiple times (Quillen et al. 1993); Figure 7 shows the detailed structure of the central 13 kpc of the NGC 5128 as seen in FUV.

We used the AIPS program BLSUM to integrate the emission from the central galaxy in a hand-drawn region of size ∼73 × 29 (∼8.3 × 3.3 kpc), enclosing all of the FUV emission shown in Figure 7 but excluding the base of the inner jet and nearby bright stars. We measured the local background in several nearly star-free boxes near the galaxy (within 10') and subtracted that level from the total measured flux. Integrating across the GALEX bandpass, we derive the observed FUV flux of the central starburst disk, $F_{{\rm FUV}}^{{\rm obs}}\gtrsim 3.2\times {{10}^{-12}}$ erg s⁻¹-cm² Å⁻¹. Multiplying by the effective GALEX bandwidth (268 Å for FUV, Morrissey et al. 2007), and taking a 3.8 Mpc distance, we estimate the directly observed FUV luminosity⁹ as $L_{{\rm FUV}}^{{\rm obs}}\sim 1.5\times {{10}^{42}}$ erg s⁻¹.

To correct for foreground Galactic extinction, we use A_FUV = 0.91 mag (Gil de Paz et al. 2007). Thus, the unobscured FUV luminosity from the galaxy is boosted by a factor 2.3, to $L_{{\rm FUV}}^{{\rm corr}}\sim 3.5\times {{10}^{42}}$ erg s⁻¹. We convert this to a star formation rate (SFR) with the proxy relation SFR = 4.4 × 10⁻⁴⁴ L_FUV (Murphy et al. 2011; units of SFR are M_☉/yr and units of L_FUV are erg s⁻¹ in GALEX FUV bandpass). We thus estimate the unobscured SFR in the disk of NGC 5128: ${{({\rm SFR})}_{{\rm naked}}}\sim .15{{M}_{}}$ yr⁻¹. If UV light escapes preferentially above and below the disk, and is not scattered into our line of sight, (SFR)_naked may be higher.

The strength of an obscured starburst is commonly determined from its bolometric IR luminosity (L_FIR: 8–1000 μ). To estimate L_FIR for NGC 5128, we started with its IRAS fluxes, ∼217 Jy at 60 μ, and ∼501 Jy at 100 μ (Rice et al. 1988). We converted these to 40–120 μ flux, using ${\rm FIR}=1.26\times {{10}^{-14}}\left[ 2.58{{S}_{60\mu }}+{{S}_{100\mu }} \right]$ (e.g, Helou et al. 1985). We then multiplied (FIR) by the ∼1.5 conversion factor typical for FIR-selected galaxies (Yun et al. 2001) finding that the bolometric flux L_FIR ∼ 9.3 × 10⁹ L_☉ yr⁻¹.

Both Kennicutt (1998) and Murphy et al. (2011) give proxy relations to convert L_FIR to a SFR, with small differences reflecting different assumptions detailed conditions within the starburst. Kennicutt gives SFR ∼ 1.7 × 10⁻¹⁰ L_FIR; Murphy et al. give SFR ∼ 1.47 × 10⁻¹⁰ L_FIR; here, SFR is measured in (M_☉ yr⁻¹) and L_FIR is measured in solar luminosities. Taking a rough mean of these, we estimate the rate of obscured star formation in NGC 5128 as SFR_obsc ∼ 1.6 M_☉yr⁻¹

We also used our 90 cm radio image (Paper 1) as an independent check of the SFR in the central gas/dust disk. Because radio luminosity does not suffer from extinction, it should give the total SFR (obscured plus unobscured). We pointed out in Paper 1 that we detect a "ruff" of radio emission, transverse to the Inner Lobes and approximately coincident with the gas/dust disk (compare, e.g., Figure 1 or 6 of Paper 1 to Figure 7 of this paper). We used the flux of this ruff as a separate proxy for the SFR, as follows. We measured the mean flux density of the NW and SE regions of the disk (outside of the bright central AGN) as 165 mJy/beam, which we estimate includes a background of 25 mJy/beam. We estimated the total disk area as 53 × 15; thus the total radio flux from the disk ∼(140 mJy/beam) × (56 beams) ≃7.84 Jy at 327 MHz. Because of the uncertainties in estimating the radio background and interpolating across the AGN, this flux estimate is probably only good to a factor ∼2.

We scaled this flux estimate to 1 GHz using the mean spectrum of star-forming galaxies in this frequency range, S(ν) ∝ ν^−0.55 (Marvil et al. 2015). This gives a 1 GHz radio luminosity L_{1 GHz} ∼ 7 × 10²⁸ erg s⁻¹ Hz⁻¹ for the star-forming disk in Cen A. We then used the proxy relation from Equation (16) of Murphy et al. (2011), which combines thermal and non-thermal radio emission to estimate a total rate as SFR = 5.8 × 10⁻²⁹ L_1 GHz. In this relation, the units of SFR are M_⊙ yr⁻¹, the units of L_{1 GHz}are erg s⁻¹ Hz⁻¹; we have assumed T = 10⁴ K and ν = 1 GHz. From this we estimate the radio-derived SFR in the disk of Cen A as ∼4 M_⊙ yr⁻¹. Although this rate is formally larger than the net SFR we derive from FUV and FIR, ∼1.75 M_⊙ yr⁻¹, we do not take the difference seriously because of the large uncertainties in our measurement of the radio flux. We thus argue that the radio data support the FUV+FIR data, showing that the central disk in NGC 5128 is currently forming stars at a rate ∼2 M_⊙ yr⁻¹.

As we discuss in Paper 1, the inner radio source is ≲2 Myr old (Croston et al. 2009). One possibility is that the AGN has been dormant for much or all of the duration of the starburst. This could be causal, perhaps because supernovae-driven turbulence within the starburst disk disrupts the accretion flow, and/or removes substantial amounts of material that would otherwise have accreted onto the central black hole (e.g., Wild et al. 2010, Davies et al. 2007).

To be specific, we assume the AGN in NGC 5128 has been quiet for ∼50–100 Myr (the likely duration of the central starburst, also the delay that Davies et al. (2007) suggest between start of a starburst and the restarted feeding of the AGN). If this is the case, any radio-loud plasma from previous active episodes of the AGN would long ago have reached and/or dispersed within the Outer Lobes. For instance, if the disconnected Inner Lobe plasma kept a coasting speed ∼3000 km s⁻¹(comparable to the deprojected advance speed of the Inner Lobes; cf. Paper 1), it would reach 150 kpc in 50 Myr—well beyond the Transition Regions discussed here. In this case, energy and mass flow within the Transition Regions will be limited to that of the basic, starburst-driven wind described above.

While we cannot prove or disprove this picture, we find it unlikely for several reasons. It would require us to be catching the AGN at a special time, only 1–2 Myr since its restart after a long silence. In addition, we argue below (in Sections 6.2 and 6.4) that a wind driven only by the starburst does not have enough power to support the emission line clouds in the weather system, or the radio emission from the NML. Finally, essentially every double radio galaxy known hosts a currently active, radio-loud AGN. Even the rare radio galaxies without detected jets connecting the AGN to the radio lobes have detectable jets on kpc scales (e.g., Fornax A, van Breugel & Fomalont 1984; 3C310, Fomalont et al. 1989). If it were common for the AGN supporting a large radio galaxy to go dormant for ∼10% of the age of the source, we would expect to see a similar fraction of the radio galaxy population with no radio core—which is not the case.

An alternative possibility is that the AGN, recently restarted, has only been quiet for a time short compared to the duration of the starburst. Perhaps the AGN "burbles along," dying down and then reactivating every few Myr. This may happen, for instance, if most of the starburst is sufficiently spread out within the inner few kpc of the star-forming disk that it has little effect on the accretion flow close to the AGN. As we discussed in Paper 1, as the restarted jets propagate into the wind, different evolution paths are possible when they grow to tens of kpc scales. They may continue to be identifiable as jets, without much effect on the ambient wind flow. In this case, we might expect "jet fragments" to continue to coast within the ambient wind when the AGN again becomes quiet. Alternatively, the jets may turn into more diffuse, turbulent plumes, with the potential to impact and energize the local wind flow. In this case we might expect the jet plasma to lose its identity as a jet, and to mix with the ambient wind.

As an example, assume the AGN has been dormant for only ∼5 Myr. If this is the case, the jet plasma from the previous cycle would still be within the Transition Regions (3000 km s⁻¹ × 5 Myr = 15 kpc; 3000 km s⁻¹is the estimated advance speed of the Inner Lobes, Paper 1), and would still be radio-loud (estimating the synchrotron lifetime at several tens of Myr; e.g., Paper 1). Because we do not see any such jet remnants within the Transition Regions in Cen A, we speculate that jets from the previous cycle must have disrupted, dispersed and mixed with the ambient wind plasma. It follows that the mass and energy from the from AGN may have been deposited locally in the wind—perhaps quite close to the base of the wind—and thus became part of the diffuse wind flow.

Morganti et al. (1991, 1992), suggested that the Inner and Outer Filaments are photoionized by beamed radiation from a blazar. However, Hardcastle et al. (2003) suggest that the radio and X-ray brightness ratios in the inner jet appear inconsistent with a misdirected blazar model. Sutherland et al. (1993) point out that the ionization gradient noted by Morganti et al. (1992) does not point toward the nucleus, as might be expected if the Inner Filament were ionized by beamed EUV-X-ray nuclear radiation. Together with Viegas & Prieto (1992), Sutherland et al. (1993) also mention difficulties reproducing the observed [Oiii]I₄₃₆₃/(I₅₀₀₇ + I₄₉₅₉) ratio using pure photoionization models. Crockett et al. (2012) note that a beamed photoionization model also fails to account for the highest excitation lines (e.g., Graham 1998; Evans & Koratkar 2004). Furthermore, ionization by beamed radiation, by itself, does not explain the complex velocity field of the ionized gas in the Filaments.

Relative orientations also do not support the idea of beamed photoionization. The inner jet is oriented at an an angle of 30–41° (depending on distance from the nucleus), with an opening angle of ∼8–10°. The Inner Filament is roughly aligned with the inner jet, with an orientation of ∼28–33° degrees, and thus it might might plausibly be ionized by beamed radiation from the nucleus or inner jet. However, the jet is not aligned with the Outer Filament (orientation ∼44–53°), or with most of the outer ribbon structure (orientation ∼40–56°).

Another possibility is that the FUV and emission-line regions are illuminated by the star-forming disk. However, the agreement between radio-determined SFRs and UV/IR-determined SFRs (Section 4) precludes the possibility of significant amounts of hidden FUV emission being emitted along the dust-disk's axis.

Filaments of emission-line gas and young stars are also found close to central radio galaxies in cooling-core clusters (e.g., Fabian 2012; Canning et al. 2014). These filaments are similar to the weather system in Cen A: they need ongoing energization, and they sit in a dynamic, hot (T ∼ 10⁷ K) ambient medium. What can we learn from them? Although the filaments in cooling-core clusters sit in a denser ambient medium than that which surrounds the weather system in Cen A, there are interesting similarities. The thermal state of filaments in cooling cores is far from understood, but there are several possibilities. They may be energized by energy transfer from the surrounding medium (e.g., heat conduction, Sparks et al. 2004, or cosmic ray penetration, e.g., Fabian 2012). The filaments may also be energized dynamically, either by infall of the residual cooling flow, or by expansion work done by the radio galaxy if the filaments happen to sit close to the radio lobe edges.

The cooling-core analogy is not perfect, however. The energetics of cooling cores are ultimately driven by a combination of slow cooling inflow and radio lobe expansion. Neither of these are clearly available for the Cen A weather system. There is no sign of cooling cores in small galaxy groups such as that which hosts NGC 5128. The radio-loud plasma in both Transition Regions is more likely to expand, or flow, into the tenuous channels provided by the pre-existing Outer Lobes. Thus, a different driver is needed for the weather system; the wind we suggest is flowing through both Transition Regions can provide that driver. If such a wind exists, the mechanisms by which energy is transferred to the cool gas clouds may be quite similar in Cen A and in cooling cores. We discuss wind energization of the weather system in Cen A in more detail in Section 6.

If there were no pre-existing radio lobe, the wind would expand spherically into the undisturbed galactic ISM at density 10⁻³cm⁻³ and pressure ∼8.5 × 10⁻¹³dyn cm⁻² (revealed by X-ray data; Kraft et al. 2009). In this situation, the termination shock sits at a position given by $R_{{\rm shock}}^{2}\simeq {{P}_{w}}/4\pi {{p}_{o}}$. In a pure-starburst wind the shock sits at only R_shock ∼ 7 kpc from the core; in an AGN-enhanced wind it sits at R_shock ∼ 13 kpc from the core. Thus, by the time the wind plasma reaches the Transition Regions, it carries the same amount of power, but becomes a subsonic breeze rather than a supersonic wind.

The situation in Cen A is likely to be more complicated. The low-density radio lobes (previously evacuated by the AGN) may constrain the wind to flow within a relatively small solid angle, ${\Omega }\ll 4\pi$. If this is the case—and if the wind flow remains laminar within the channel—the termination shock sits at a position given by $R_{{\rm shock}}^{2}\simeq {{P}_{w}}/8\pi f{{p}_{o}}$, where f = Ω/4π, and we assume half of the total power goes to each side of the galaxy. Based on radio images, we estimate the wind has expanded laterally to a width of ∼24 kpc when it is 30 kpc from the galaxy in the NTR. Thus, the "open channel" occupies only a fraction f ∼ 5% of the full solid angle that would be available to a wind expanding spherically into an isotropic medium.

To estimate pre-wind conditions in the radio lobe, we follow Eilek (2014), who argued that the Outer Lobes on large scales are in pressure balance with the local IGM (thus the pressure drops from ∼8.5 × 10⁻¹³dyn cm⁻² close to the galaxy, to ∼3.2 × 10⁻¹³dyn cm⁻² past ∼100 kpc from the galaxy). In this scenario, the termination shock of a pure-starburst wind sits at R_shock ∼ 30 kpc from the core, nearly reaching the outer end of the weather ribbon. In an AGN-boosted wind, the shock sits at R_shock ∼ 65 kpc. Thus if the galactic wind is channeled into the low-density radio lobes, it may still be supersonic when it reaches the NTR and STR.

Finally, one might ask if these termination shocks could be detected observationally. Detection in X-rays seems unlikely; the density jump in such tenuous plasma would be very faint, and X-ray imaging would be severely challenged by the large angular size, and variations in Galactic foreground emission. One might also expect the shock to be radio-bright, due to particle acceleration and magnetic field amplification local to the shock. Eilek (2014) has in fact suggested that the bright filaments within the outer lobes (seen in ATCA image from Feain et al. 2011) could be internal shocks, driven by transonic outflow within the lobes. The complex nature of the radio-bright filaments makes it difficult to identify any one structure as "the" terminal wind shock; but if the wind flow becomes turbulent as it propagates into the Outer Lobes, it may be the origin of the observed radio-bright filaments.

Although the detailed physics remains unclear, it has often been suggested that plasma flow past cold clouds will induce star formation in the clouds (e.g., Fragile et al. 2004, Gaibler et al. 2012). The flow could, of course, be either a jet or a wind. The "typical emission-line cloud" we use in Section 3.4—to represent the emission line clouds studied by Morganti et al. (1991)—is not gravitionally unstable. However, we can speculate that some of the clouds, when perturbed and/or fragmented by the passing flow, will cool rapidly enough to become unstable and collapse to form stars (e.g., Mellema et al. 2002; Cooper et al. 2009).

When a supersonic wind encounters slower clouds, bow shocks will form upstream of the clouds. Some clear examples are apparent in simulations from Cooper et al. (2009). Hα images of the Outer Filament in Cen A strongly suggest that such bow shocks exist upstream of the emission-line clouds there (Section 5.2). This has several interesting consequences for the local weather. (1) The upstream ram pressure, moderated by the bow shock, will help confine the high-pressure emission-line clouds and radio/X-ray knots (if the latter are diffuse gas). (2) The hot shocked wind plasma around the clouds will be an X-ray source (e.g., Marcolini et al. 2005; Cooper et al. 2009), consistent with the observed spatial correlation of X-rays with some emission line clouds. (3) The shocks are likely to accelerate electrons to relativistic energies, and also enhance the local magnetic field (as we discuss in Section 6.3), consistent with the observed spatial correlation of nonthermal radio emission with the FUV/Hα ribbon.

In Section 3.3, we suggested that cold Hi close to NGC 5128—in particular the "17 kpc cloud" and smaller structures closer to the galaxy (Figure 6)—might be the source of the material in the extended weather ribbon. If this is the case, gas clouds torn from the parental Hi clouds must maintain their identity for at least 50 Myr, the travel time from the 17 kpc cloud to the end of the weather ribbon. This may be difficult, because the clouds are subject to destructive forces as they are carried outwards by a faster, hot galactic wind. They can be be evaporated by thermal conduction as they sit in the hot wind. They can also be shredded by the Kelvin–Helmholtz (KH) instability as the hot wind flows past them.

To check this, we estimate lifetimes for our typical emission-line clouds (Section 3.4) against both effects (details and references are given in the Appendix). We find the answer is uncertain. Simple estimates suggest the clouds are destroyed very rapidly, in only a few Myr. This seems to imply that Hi closer to NGC 5128 cannot be the only source of material in the weather ribbon. However, recent simulations identify mitigating effects which may slow down the destruction, and retain the cloud's identity, for longer than the lifetimes predicted by the simple arguments.

Thus, while it is possible that known Hi clouds can be the source for the entire weather ribbon, the question is not settled. Other more local mechanisms may be necessary. Some of the gas in the extended weather system may have come from other cold clouds—not yet detected, or perhaps now destroyed—-which happen to lie farther than 17 kpc from the galaxy and in the path of the wind. Another possibility is that star formation in the weather ribbon close to the galaxy created young star clusters which continue to move outwards at the speed of their parental gas clouds. Winds from these young clusters, driven by normal stellar evolution, may supply new gas to the weather system which we now detect as part of the FUV/Hα ribbon.

The NTR contains the diffuse, radio-bright NML as well as the radio-loud knotty ridge that coincides with the optical/FUV/X-ray weather system. Synchrotron radio emission from the region will be enhanced when the interaction of the supersonic wind with the cold gas generates shocks (as the wind is forced to change direction and/or decelerate) and MHD turbulence (easily generated by any perturbation to an MHD flow). Shocks can quickly energize relativistic particles and amplify magnetic fields (e.g., Drury 1983; Schure et al. 2012). Alfvén waves within MHD turbulence can also accelerate relativistic particles (e.g., Lacombe 1977; Eilek 1979).

If this is the case, the radio-bright ridge and knots coincident with the weather system are due to localized particle acceleration and field amplification where the wind encounters cool gas in the weather system. The enhanced radio surface brightness which continues to the northwest—and defines the larger NML—could be due to advection of shock-accelerated plasma within the wind and/or local particle acceleration due to MHD turbulence in the wind.

No comparable radio structure has been detected in the STR, which contains only weak, diffuse radio emission (perhaps a factor ∼10 fainter than in the NML; Paper 1). If the radio-brightness of the NML is due to to the interaction of the galactic wind with cold gas clouds which happen to be in the region, then the asymmetry of the NTR and STR must be due to the lack of corresponding cold Hi to the south of the galaxy. We therefore suggest that the radio-faint STR region shows the basic, unperturbed wind flow. That is, some relativistic, magnetized plasma (from the starburst, possibly enhanced by the AGN) is advected out with the wind. It radiates faintly as it goes, but its radio emission is not amplified by any interaction with cold gas clouds in its path.

In contrast to the NML, which we have argued is at a higher pressure than the ambient galactic ISM (Paper 1, also Section 3.4), there is no evidence that the radio-loud plasma in the STR is over-pressured. In Paper 1 we found that the minimum pressure in the radio-loud NML is a factor ∼4 larger than that of the ambient ISM. Because in Paper 1 we could only put upper limits on the radio power from the STR, we could not estimate its volume emissivity (which is required to calculate the minimum pressure; e.g., Paper 1 Appendix). We can, however, use the lower-resolution image of Junkes et al. (1993; kindly provided by N. Junkes) to estimate the mean surface brightness in the STR to be ∼10% of that in a comparable area of the NTR. Because the minimum pressure is approximately proportional to the square root of the surface brightness, it is a factor ∼3 lower, on average, in the STR than in the NTR. We therefore argue that current observations are consistent with the STR being in pressure balance with the ambient ISM—and thus with the STR not being energized by interaction with dense gas clouds in its path.

This argument requires a particular geometry for the cold Hi clouds shown in Figure 6. If the gas in the northern weather ribbon is indeed being dragged out of the 17 kpc Hi cloud by a wind (which we envision as filling a flaring cylindrical region above and below the starburst disk), most of that cloud must be located outside the wind; otherwise it would seed warm emission-line gas everywhere along its length. Thus, the observed arc of Hi clouds must be either in front of, or behind, the suspected wind flow. Since only one edge of the 17 kpc cloud appears to be impacted by the wind, that edge of the cloud must be rotating into the wind flow. Figure 6 shows there is also an Hi cloud SW of the galaxy. If this southwestern cloud were located in the wind flow, we would expect to detect a Southern Middle Lobe and an equivalent southern weather system. Taken together, these arguments suggest the southwestern Hi cloud, and most of the ring of Hi clouds (e.g., to north and northwest of center), are well away from the wind flow.