Radio Imaging of the Very-High-Energy γ-Ray Emission Region
in the Central Engine of a Radio Galaxy
The VERITAS Collaboration, the VLBA 43GHz M87 Monitoring Team,
the H.E.S.S. Collaboration, and the MAGIC Collaboration∗
The full author list with affiliations can be found at the end of this paper
The accretion of matter onto a massive black hole is believed to feed the relativistic
plasma jets found in many active galactic nuclei (AGN). Although some AGN accelerate
particles to energies exceeding 1012electron Volts (eV) and are bright sources of very-high-
energy (VHE) γ-ray emission, it is not yet known where the VHE emission originates. Here
we report on radio and VHE observations of the radio galaxy M87, revealing a period of
extremely strong VHE γ-ray flares accompanied by a strong increase of the radio flux from
its nucleus. These results imply that charged particles are accelerated to very high energies
in the immediate vicinity of the black hole.
Active galactic nuclei (AGN) are extragalactic objects thought to be powered by massive
black holes in their centres. They can show strong emission from the core, which is often
dominated by broadband continuum radiation ranging from radio to X-rays and by substantial
flux variability on different time scales. More than 20 AGN have been established as VHE γ-ray
emitters with measured energies above 0.1 tera electron Volts (TeV); the jets of most of these
sources are believed to be aligned with the line-of-sight to within a few degrees. The size of the
VHE γ-ray emission region can generally be constrained by the time scale of the observed flux
variability [1, 2] but its location remains unknown.
We studied the inner structure of the jet of the giant radio galaxy M87, a known VHE γ-ray
emitting AGN [3, 2, 4, 5] with a (6.0±0.5)×109M?black hole , scaled by distance, located
16.7Mpc (54 million light years) away in the Virgo cluster of galaxies. The angle between its
plasma jet and the line-of-sight is estimated to lie between 15 − 25deg (see supporting online
text). The substructures of the jet, which are expected to scale with the Schwarzschild radius
firstname.lastname@example.org (H.E.S.S.), email@example.com, or firstname.lastname@example.org (MAGIC).
arXiv:0908.0511v1 [astro-ph.HE] 4 Aug 2009
Rsof the black hole1, are resolved in the X-ray, optical and radio wavebands  (Fig. 1). High-
frequency radio very long baseline interferometry (VLBI) observations with sub-milliarcsecond
(mas) resolution are starting to probe the collimation region of the jet . With its proximity,
bright and well-resolved jet, and very massive black hole, M87 provides a unique laboratory
in which to study relativistic jet physics in connection with the mechanisms of VHE γ-ray
emission in AGN.
VLBI observations of the M87 inner jet show a well resolved, edge-brightened structure
extending to within 0.5mas (0.04pc or 70Rs) of the core. Closer to the core, the jet has a
wide opening angle suggesting that this is the collimation region . Generally, the core can
be offset from the actual location of the black hole by an unknown amount , in which case it
could mark the location of a shock structure or the region where the jet becomes optically thin.
However, in the case of M87 a weak structure is seen on the opposite side of the core from the
main jet, which may be the counter-jet, based on its morphology and length [10, 11]. Together
with the observed pattern in opening angles, this suggests that the black hole of M87 is located
core (see supporting online text). Along the jet, previous monitoring observations show both
near-stationary components  (pc-scale) and features that move at apparent superluminal
speeds [12, 13] (100 pc-scale). The presence of superluminal motions and the strong asymmetry
of the jet brightness indicate that the jet flow is relativistic. The near-stationary components
could be related to shocks or instabilities, that can be either stationary, for example if they are
the result of interaction with the external medium, or slowly moving if they are the result of
instabilities in the flow.
A first indication of VHE γ-ray emission from M87 was reported by the High Energy
Gamma-Ray Astronomy (HEGRA) collaboration in 1998/99 . The emission was confirmed
on time scales of days. M87 was detected again with the Very Energetic Radiation Imaging
Telescope Array System (VERITAS) in 2007  and, recently, the short-term variability was
confirmed with the Major Atmospheric Gamma-Ray Imaging Cherenkov (MAGIC) telescope
during a strong VHE γ-ray outburst  in February 2008. Causality arguments imply that the
emission region should have a spatial extent of less than ≈ 5δRs, where δ is the relativistic
Doppler factor. This rules out explanations for the VHE γ-ray emission on the basis of (i) dark
matter annihilation , (ii) cosmic-ray interactions with the matter in M87 , or (iii) the
knots in the plasma jet (Fig. 1C). Leptonic [16, 17] and hadronic  VHE γ-ray jet emission
models have been proposed. However, the location of the emission region is still unknown.
The nucleus [19, 20], the inner jet  or larger structures in the jet, such as the knot HST-1
(Fig. 1C), have been discussed as possible sites . Because the angular resolution of VHE
experiments is of the order of 0.1deg, the key to identifying the location of the VHE γ-ray
emission lies in connecting it to measurements at other wavebands with considerably higher
spatial resolutions. An angular resolution more than six orders of magnitude better (less than
1The Schwarzschild radius of a black hole with the mass m is defined as Rs= 2Gm/c2, G is the gravitational
constant, and c is the speed of light. The Schwarzschild radius defines the event horizon of the black hole.
6 × 10−8degrees, corresponding to approximately 30Rsin case of M87) can be achieved with
radio observations (Fig. 1).
We used the H.E.S.S. , MAGIC  and VERITAS  instruments to observe M87
during 50 nights between January and May 2008, accumulating over 95h of data (corrected
for the detector dead times) in the energy range between 0.1TeV and several 10’s of TeV.
with a resolution of 0.21 × 0.43mas , corresponding to about 30 × 60Rs, see . During
the first half of 2008, three X-ray pointings were performed with the Chandra satellite . Our
light curves are shown in Fig. 2.
We detected multiple flares at VHE in February 2008 with denser sampling, following a
trigger sent by MAGIC [∼23 h of the data published in ]. The short-term VHE variability,
first observed in 2005 , is clearly confirmed and the flux reached the highest level observed
so far from M87, amounting to more than 10% of that of the Crab Nebula. At X-ray frequencies
the innermost knot in the jet (HST-1) is found in a low state, whereas in mid February 2008 the
nucleus was found in its highest X-ray flux state since 2000 . This is in contrast to the 2005
VHE γ-ray flares , which happened after an increase of the X-ray flux of HST-1 over several
years , allowing speculation that HST-1 might be the source of the VHE γ-ray emission
; no 43GHz radio observations were obtained at that time. Given its low X-ray flux in
2008, HST-1 is an unlikely site of the 2008 VHE flaring activity.
Over at least the following two months, until the VLBA monitoring project ended, the
43GHz radio flux density from the region within 1.2mas of the core rose by 30% as compared
with its level at the time of the start of the VHE flare and by 57% as compared with the average
level in 2007 (Fig. 2). The resolution of the 43GHz images corresponds to 30 × 60Rsand
the initial radio flux density increase was located in the unresolved core. The region around the
core brightened as the flare progressed (Fig. 3), suggesting that new components were emerging
from the core. At the end of the observations, the brightened region extended about 0.77mas
from the peak of the core, implying an average apparent velocity of 1.1c (c is the speed of
light), well under the approximately 2.3c seen just beyond that distance in the first half of 2007.
Astrometric results obtained as part of the VLBA monitoring program show that the position
of the M87 radio peak, relative to M84, did not move by more than ∼6 Rsduring the flare,
suggesting that the peak emission corresponds to the nucleus of M87.
Because VHE, X-ray and radio flares of the observed magnitude are uncommon, the fact
that they happen together (chance probability of P < 0.5%, supporting online text) is good
evidence that they are connected. This is supported by our joint modeling of the VHE and radio
light curves: The observed pattern can be explained by an event in the central region causing
the VHE flare. The plasma travels down the jet and the effect of synchrotron self-absorption
causes a delay of the observed peak in radio emission because the region is not transparent
at radio energies at the beginning of the injection (supporting online text, Sec. 3). The VLBI
structure of the flare along with the timing of the VHE activity, imply that the VHE emission
occurred in a region that is small when compared with the VLBA resolution. Unless a source
of infrared radiation is located very close to the central black hole, which is not supported by
current observations , TeV γ-ray photons can escape the central region of M87 without
being heavily absorbed through e+e−pair production [19, 20].
The light curve might indicate a rise in radio flux above the range of variations observed
in the past, starting before the first VHE flare was detected. This could imply that the radio
emission is coming from portions of the jet launched from further out in the accretion disk than
that responsible for the VHE emission. However, it is difficult to derive a quantitive statement
on this, because no VHE data were taken in the week previous to the flaring. Thus, an earlier
start of the VHE activity cannot be excluded, either.
A possible injection of plasma at the base of the jet observed at optical and X-ray energies
with a delayed passage through the radio core ∼104Rsfurther down the jet – interpreted as
a standing shock and accompanied by an increase in radio emission – has been discussed in
the case of BLLac  (with evidence for VHE emission, see supporting online text for more
details). M87 is much closer than BLLac and has a much more massive black hole, allowing
the VLBA to start resolving the jet collimation region whose size, from general relativistic
magnetohydrodynamic simulations , is thought to extend over ∼1000 Rs. In case of M87
the radio core does not appear to be offset by more than the VLBA resolution of ∼50 Rsfrom
the black hole (see supporting online text) and the jet has a larger angle to the line-of-sight than
in BLLac. Thus the coincidence of the VHE and radio flares (separated in photon frequency by
16 orders of magnitude), constrains the VHE emission to occur well within the jet collimation
Acknowledgements: H.E.S.S.: The support of the Namibian authorities and of the University of Namibia in
facilitating the construction and operation of H.E.S.S. is gratefully acknowledged, as is the support by the Ger-
man Ministry for Education and Research (BMBF), the Max Planck Society, the French Ministry for Research,
the CNRS-IN2P3 and the Astroparticle Interdisciplinary Programme of the CNRS, the U.K. Science and Tech-
nology Facilities Council (STFC), the IPNP of the Charles University, the Polish Ministry of Science and Higher
Education, the South African Department of Science and Technology and National Research Foundation, and by
the University of Namibia. We appreciate the excellent work of the technical support staff in Berlin, Durham,
Hamburg, Heidelberg, Palaiseau, Paris, Saclay, and in Namibia in the construction and operation of the equipment.
MAGIC: The collaboration thanks the Instituto de Astrof´ ısica de Canarias for the excellent working conditions at
the Observatorio del Roque de los Muchachos in La Palma, as well as the German BMBF and MPG, the Italian
INFN and Spanish MICINN. This work was also supported by ETH Research Grant TH 34/043, by the Polish
MniSzW Grant N N203 390834, and by the YIP of the Helmholtz Gemeinschaft. VERITAS: This research is sup-
ported by grants from the U.S. Department of Energy, the U.S. National Science Foundation and the Smithsonian
Institution, by NSERC in Canada, by Science Foundation Ireland and by the STFC in the U.K. We acknowledge
the excellent work of the technical support staff at the FLWO and the collaborating institutions in the construction
and operation of the instrument. VLBA: The Very Long Baseline Array is operated by the National Radio Astron-
omy Observatory, a facility of the U.S. National Science Foundation, operated under cooperative agreement by
Associated Universities, Inc.
Figure 1: M87 at different photon frequencies and length scales. (A): Comparison of the different length scales.
(B): 90cm radio emission measured with the VLA. The jet outflows terminate in a halo which has a diameter
of roughly 80kpc (15?). The radio emission in the central region is saturated in this image. Credit: F.N. Owen,
J.A. Eilek and N.E. Kassim , NRAO/AUI/NSF. (C): Zoomed image of the plasma jet with an extension of
2kpc (20??), seen in different frequency bands: X-rays (Chandra, upper panel) optical (V band, middle) and radio
(6cm, lower panel). Individual knots in the jet and the nucleus can be seen in all three frequency bands. The
innermost knot HST-1 is located at a projected distance of 0.86arcseconds (60pc, ≈ 105Rs) from the nucleus.
Credit: X-ray: NASA/CXC/MIT/H. Marshall et al., radio: F. Zhou, F. Owen (NRAO), J. Biretta (STScI), optical:
NASA/STScI/UMBC/E. Perlman et al., . (D): An averaged, and hence smoothed, radio image based on 23
images from the VLBA monitoring project at 43GHz. The color scale gives the logarithm of the flux density in
units of 0.01mJy/beam. The indication of a counter-jet can be seen, emerging from the core towards the lower
left side. Image from .
29 Jan05 Feb 12 Feb
Chandra (2-10 keV)
02 Apr 200702 Jul 200701 Oct 200701 Jan 200801 Apr 2008
VLBA (43 GHz)
nucleus (r = 1.2 mas)
peak flux density
jet w/o nucleus (1.2-5.3 mas)
Figure 2: Combined M87 light curves from 2007 to 2008. (A): VHE γ-ray fluxes (E > 0.35TeV, nightly av-
erage), showing the H.E.S.S., MAGIC and VERITAS data. The fluxes with statistical errors (1 standard deviation)
were calculated assuming a power-law spectral shape of dN/dE ∝ E−2.3. Monthly-binned archival VERITAS
data taken in 2007 are also shown . The systematic uncertainty in the flux calibration between the experiments
was estimated to be on the order of 20% based on Crab Nebula data. The regular gaps in the light curve correspond
to phases of full moon during which no observations were possible. The inlay shows a zoomed version of the flar-
ing activity in February 2008; the time span is indicated by the grey vertical box in all panels. (B): Chandra X-ray
measurements (2 − 10keV) of the nucleus and the knot HST-1 . (C): Flux densities from the 43GHz VLBA
observations are shown for (i) the nucleus (circular region with radius r = 1.2mas = 170Rscentered on the peak
flux), (ii) the peak flux (VLBA resolution element), and (iii) the flux integrated along the jet between distances of
r = 1.2−5.3mas (compare with Fig. 3). The error bars correspond to 5% of the flux. The shaded horizontal area
indicates the range of fluxes from the nucleus before the 2008 flare. Whereas the flux of the outer regions of the
jet does not change substantially, most of the flux increase results from the region around the nucleus. Image from
between January and August, 2007, well before the VHE and radio flare. The contour levels start at 5, 10, 14.3,
and 20mJy per beam and increase from there by factors of√2. The restoring beam used for all of the images
is 0.21 × 0.43mas (30 × 60Rs) elongated in position angle −16deg, as shown by the ellipse in the upper left
corner. (B): Image from 5 April 2008, with the same contours and colors as in (A). The linear scale in light days
and Schwarzschild radii is also shown. (C - F): Difference images for observations during the period of the radio
flare showing its effects. These were made by subtracting the average image (A) from the individual epoch images.
The contours are linear with 10 (white) at intervals of 7mJy per beam followed by the rest (black) at intervals of
70mJy per beam; negative contours are indicated by dashed lines. At the time of the VHE flare, the core flux
density was already above the average but the region of the jet between −0.5 and −1.0mas RA offset was below
average, suggesting that there had been a period of below-normal activity leading up to the flare and that the radio
flare may have begun before the VHE flare. The sequence shows the substantial rise in the core flux density and
the appearance of enhanced emission along the inner jet. Image from .
VLBA images of M87 at 43GHz. (A): Average (hence smoothed) of 11 images from data taken
Supporting Online Material
1) The geometry of the M87 jet
angle of 1 milli-arc sec (mas) corresponds to 0.081pc = 2.5×1017cm. The mass of the central
black hole has been recently determined using detailed modeling of long slit spectra of the
central regions of M87 to be (6.0 ± 0.5) × 109M?, corrected for the distance we use2. The
Schwarzschild radius is Rs ≈ 1.8 × 1015cm; 1mas therefore corresponds to approximately
140Rs. The observed timing delay ∆tobsbetween the onset of the VHE flare and the radio peak
is about 50days (∼ 4×106s). The following equations hold: The observed velocity in units of
the speed of light c is calculated as βobs= [βintsinθ]/[1 − βintcosθ], the Doppler boost factor
is δ = [Γ(1 − βintcosθ)]−1, the observed time delay is ∆tobs= ∆tint[1 − βintcosθ)], and the
distance along the jet is rint= robs/sinθ ≈ 3robsfor θ = 20deg (see below).
The luminosity distance of M87 is D = 16.7Mpc , so that an
The jet orientation angle.
with superluminal speeds  of βobs≈ 6 (HST-1), βobs≈ 5 (knot D), and βobs≈ 4 (knot E).
1 indicate superluminal speeds in the range of βobs≈ 3 − 4  and through knot D indicate
a superluminal speed of βobs≈ 2.5 . Assuming the jet plasma moves at the speed of light
(βint= 1), one obtains a maximum viewing angle of cos(θmax) = [β2
highest optically determined superluminal speed requires a jet viewing angle of θ < 18.9deg.
Ontheotherhand, thehighestradiodeterminedsuperluminalspeedrequiresθ < 28.0−36.9deg
for βobs≈ 4 − 3, respectively. Jet angles of θ = 30 − 45deg were derived  based on radio
observations at 43GHz, assuming the component to the East of the core is a counter-jet, and
with a velocity measurement based on only one pair of observations. In this paper we assume a
likely range of the jet angle of θ = 15 − 25deg.
The most rapid optical proper motions are observed along the jet
obs+ 1]. The
The jet opening angle.
function of the distance to the core: ψobs(< 0.5mas) ≈ 60deg , ψobs(1 − 5mas) ≈ 12deg
, and ψobs(> 10mas) ≈ 6deg . Assuming a jet viewing angle of θ = 20deg the
corresponding intrinsic full opening angles are: ψint(< 0.5mas) ≈ 30deg, ψint(1 − 5mas) ≈
4deg, and ψint(> 10mas) ≈ 2deg. The observed pattern of opening angles suggest that the
radio core corresponds to the formation region of the jet.
The observations indicate the following jet full opening angles as a
cies located 24arcsec away from the nucleus in the direction opposite to the jet resulted in a
The optical identification of an emission feature observed at radio frequen-
2Gebhardt & Thomas (2009) assumed a distance of 17.9Mpc. As those authors point out, that mass is about
70% higher than previous determinations , corrected for the different assumed distances. The reason for such
a large change is not fully explained yet. The use of the new mass  implies a higher resolution of our VLBA
observations in terms of Rs, but does not substantially affect our conclusions.
first indication of a counter-jet in M87 . This feature is also seen in observations at mid-
infrared frequencies . Radio observations at wavelengths of 2cm show clear indications of
a counter-feature extending up to 3mas from the radio core in the direction opposite to the jet
span of more than 10 years), the counter feature seems to move in the opposite direction with
an apparent velocity of (0.010±0.001)c, strengthening the counter-jet interpretation. However,
it cannot be fully excluded that this apparent movement is a result of temporal under-sampling
in the kinematic analysis, although the individual tracked jet features give consistent bright-
ness temperatures across the observation epochs, which would be unlikely in case of incorrect
cross-identifications of the components. The jet to counter-jet brightness ratio (0.5 − 3.1mas)
is 10−15. A counter-feature is also observed at 22 and 43GHz, extending roughly 1mas from
the core in the direction opposite to the jet , see also Fig. 1D. The jet to counter-jet bright-
ness ratio is calculated to be ∼ 14. Marginal indication for an apparent velocity of 0.17c of
the counter-jet away from the core is found from three 43GHz radio images. The observations,
however, suggest a temporal under-sampling of those data.
, and its jet morphology is consistent with a wide-opening angle jet base, as seen at 43GHz,
converging near the point of the black hole. Farther out between about 1 and 10mas the bright
edges of the jet converge toward a point close to the maximum extent of the counter-feature.
Interpreting it as a part of the jet would require that the jet before the radio peak is better
collimated than afterwards, although it appears to be as wide as the jet itself. The astrometric
results relative to M84 show that the root mean square scatter of the position of the radio core
is about 5 × 2Rsalong and across the beam with no clear systematic motion (Davies et al., in
preparation). If the radio core is farther down the jet and the jet power is going up fractionally
like the flux density, this would imply a very stable position of the shock region. The spatial
stability is a reasonable assumption for the radio core being located close to the black hole.
Theoretical modelling also supports the hypothesis that the 43GHz radio peak emission results
from the position of the black hole .
2) The VHE and radio observations and frequency of flares
renkov telescopes (IACTs) located in Namibia, the Canary Islands (Spain) and Arizona (USA),
respectively. The telescopes measure cosmic γ-ray photons (entering the atmosphere of the
Earth) in an energy range of 0.1TeV up to several 10’s of TeV. M87 has been observed at those
energies for the last ten years. Except for the 2008 observation campaign, the observations were
scheduled in advance and did not follow any external or internal triggers, leading to arbitrarily
sampled light curves. During the observations of the past 10 years only two episodes of flaring
activity have been measured: in 2005  and in 2008 (reported in this paper). For the first time,
M87 was observed by H.E.S.S., MAGIC and VERITAS in a joint campaign for more than 120h
in 2008 (more than 95h of data after quality selection). The integral fluxes presented in this
paper (Fig. 2) were calculated3under the assumption that the spectrum of M87 is described by
a power-law function dN/dE ∝ E−2.3. Any correlation between the spectral shape and the
flux level has not yet been established for M87. The relative frequency of flaring activity was
estimated by fitting the night-by-night binned light curves as measured by H.E.S.S., MAGIC
and VERITAS with a constant function (using all available data – partly archival – from 2004
to 2008). Subsequently flux nights with the most significant deviation from the average were
removed until the fit resulted in a reduced χ2per degree of freedom of less than 1; all removed
points corresponded to flux values higher than the average. The light curves are compatible with
constant emission for 49 out of 53 nights (H.E.S.S., 2004-2008), 12 out of 21 nights (MAGIC,
2008) and 50 out of 51 nights (VERITAS, 2007-2008). Combining these numbers one finds
flaring activity in the so far recorded data in 14 out of 125 nights of observations, resulting in
a relative frequency of flares on the order of 10% of all observed nights. Almost all data were
recorded arbitrarily and except for four nights (with a time difference of ∼ 0.5days between
the VERITAS and the H.E.S.S./MAGIC observations) all observations were separated in time
by more than one day. Therefore we assume that this number gives an estimate of the general
chance to measuring a VHE γ-ray flare from M87. However, the relative frequency of flaring
activity is overestimated by the fact that the 2008 observations were intensified for some nights
during the high flux state following the VHE trigger by MAGIC .
surements of the X-ray flux of the nucleus during the last ten years . Three measurements
exceed a flux level higher than 2 times the root mean square (RMS) of the average flux of all
data points (relative occurrence of ∼ 5%). Only one measurement exceeds the level of 3RMS
which was taken during the radio flare with a deviation of ∼ 4.3RMS (Fig. 2 in the main text).
M87 was regularly observed at X-ray energies with Chandra, resulting in 61 mea-
three weeks . The aim of this ’movie project’ was to study morphological changes of the
plasma jet with time. Preliminary analysis of the first 7 months showed a fast evolving structure,
somewhat reminiscent of a smoke plume, with apparent velocities of about twice the speed of
light. These motions were faster than expected so the movie project was extended from January
to April 2008 with a sampling interval of 5 days. A full analysis of these data is in progress
and details will be published elsewhere. The observed radio flux densities reached at the end
of the 2008 observations, roughly 2 months after the VHE flare occurred, are larger than seen
in any previous VLBI observations of M87 at this frequency, including during the preceding
12 months of intensive monitoring, in 6 observations in 2006 and in individual observations in
1999, 2000, 2001, 2002, and 2004 . The MOJAVE project web site4gives 15GHz VLBA
Throughout 2007, M87 was observed with the VLBA on a regular basis roughly every
3the H.E.S.S. flux points – measured with a higher energy threshold – were extrapolated down from ∼ 1TeV.
flux densities at 27 epochs since 1995, with the highest flux value measured on May 1, 20085;
most of the data (except the last two epochs) are published in . Assuming a flare duration
of ∼4 months, a similar flare was not observed during a total period of 5 × 4 = 20 months
based on the 5 observations from 2004 and earlier, which are well separated, 8 months taking
into account overlap based on the 2006 pilot observations spread over 4 months, and 14 months
during the 2007/2008 monitoring including 2 months before the start but not including the time
during the observed flare. That is 42 months total for which a similar flare was not in progress.
By the same accounting, there are 4 months with a flare. So the probability of a radio flare
being in progress at any given time is 4/46 ≈ 10%, suggesting that radio flares of the observed
magnitude are uncommon.
The observed radio/X-ray/VHE-pattern.
ceeding a flux baseline (for which the chance probability is p) is: p(n,k) =
The probability of observing k or more flare nights is P(n,≥ k) =?n
scales of flux changes are different and the data are not sampled equally, or are partly sam-
pled with higher frequencies as the characteristic time scale of the flux changes (over-sampling)
or much less frequent (under-sampling). However, if defining a time window covering the
whole increase of the radio flux between end of January to mid of April one finds 8 out of 40
VHE measurements (PVHE) and 1 out of 2 X-ray measurements (PX−ray) exceeded their base-
lines. The chance probability of observing this pattern during the radio flare is on the order of
P = PVHE· PX−ray= 0.026 · 0.095 < 0.5%. The fluxes in the radio, X-ray, and VHE bands
reached their highest archival level during the defined time window.
The probability of observing k out of n nights ex-
j=kp(n,j). A joint esti-
mated of the chance probability of the observed pattern is difficult, since the characteristic time
from a slow outer sheath
the 43GHz radio emission from the nucleus steadily increases over a time period of two months
(Fig. 2). A model calculation was performed to test if the slow variations of the radio flux can
be explained with a self-absorbed synchrotron model of electrons injected into a ”slow outer
sheath” of jet plasma. The slow outer sheath has the geometry of a hollow cone, an assumption
which is supported by the edge-brightened structure of the jet observed at radio frequencies
(Fig. 1D). As the plasma travels down the jet, it expands, leading to a decline of the frozen-in
magnetic field and to adiabatic cooling of the electrons. In the model, a γ-ray flare leads to the
injection of radio-emitting plasma at the base of the jet and at the base of the slow outer sheath.
The model assumes that the VHE γ-rays are produced very close to the black hole, e.g. in the
Whereas the VHE γ-ray emission from M87 varies on time scales of a few days,
5This is only a few weeks later than the maximum of the 43GHz flux reported in this paper. However, no
further 15GHz data for 2008 are listed on the MOJAVE web page.
43 GHz flux ?Jy?
curve resulting from a single injection
of radio-emitting plasma at time t = 0
(Γ = 1.01, βjet = 0.14, B = 0.5G).
Initially, the plasma is opaque owing
to synchrotron self-absorption. As the
plasma travels away from the point of
injection, it expands and becomes trans-
parent. The expansion leads to a de-
crease of the magnetic field and to adi-
abatic cooling of the electrons, and thus
to the decline of the radio emission.
The small deviations during the injec-
tion stage are due to numerical effects.
Simulated 43GHz light
uses the observed γ-ray fluxes to normalize the energy spectrum of the electrons responsible
for the radio emission. In the beginning, the radio-emitting plasma is optically thick, and the
synchrotron emission cannot escape. Owing to the adiabatic expansion, the plasma eventually
becomes optically thin leading to a radio flare. The radio flare dies down owing to the decline
of the magnetic field and the adiabatic cooling of the electrons.
Following the injection of radio-emitting plasma at time t0, a ring of plasma with radius
R = Rs+ βjetcsinα(t − t0) (with the thickness of the radio bright sheath being 1/5th of
the cone radius) travels down the jet. The emission of the ring is computed in the frame of
the moving plasma, assuming that the magnetic field scales as B ∝ 1/R, and the electrons
cool adiabatically. The calculation uses the standard equations for Lorentz transforming the
emission of different sections of the ring, see for example . Taking into account light travel
time effects, the received radio flux at 43GHz is computed. The overall normalization of the
electron energy spectrum is adjusted to reproduce the observed radio flux. The results of the
model strongly depend on the choice of the minimal radius Rmin= Rs, and weakly depend on
the choices of the magnetic field B, the jet opening angle α, and and the thickness of the sheath.
andaratherlowmagneticfieldofB = 0.5Gatthebaseofthejetforwhichradiativecoolingcan
be neglected. The simulated radio light curve produced by a single injection of radio-emitting
plasma is shown in Fig. 4 for an assumed bulk Lorentz factor of Γ = 1.01 (βjet= 0.14, giving
the best fit result). The radio flux needs approximately 20 days to reach its maximum. Figure 5
shows the corresponding radio light curve obtained when choosing a time-dependent electron
injection function proportional to the measured VHE γ-ray fluxes, starting with the VHE data
taken in January 2008. The spatial extent of the predicted radio source after 50 days is ∼ 3 light
days and therefore still within the central resolution element of the VLBA observations (Fig. 3).
The model was chosen to minimize the number of free parameters and assumptions. Other de-
pendencies could affect the results as follows: (i) If the emitting plasma is more compact (and
We assume an intrinsic cone opening angle of α = 5deg, a jet angle of θ = 20deg
and modelled (line) 43GHz radio flux.
The modelled curve was obtained by us-
ing the measured VHE γ-ray light curve
as source function for injecting radio-
emitting plasma into the slow sheath of
the jet. The Bulk Lorentz factor used
is Γ = 1.01 (βjet = 0.14). No VHE
γ-ray observations were possible during
phases of full moon, leaving some un-
certainties for the injection function.
Observed (data points)
the volume filling factor is < 1), it stays synchrotron self-absorbed for a longer time, increasing
the time lag between the γ-ray flares and the rise of the radio flux. (ii) The emission volume may
expand slower than proportional to t2, as assumed in the model. Slower expansion would slow
down the time scale for the rise and decay of the radio flux. (iii) A higher value of βjetwould
result in a faster radio flare. (iv) All non-thermal particles are injected into the jet right at the
base of the slow sheath. This model assumption may not be accurate. Additional non-thermal
particles may be accelerated further downstream which would lead to a longer duration of the
radio flare. (v) If the magnetic field at the base of the jet is stronger, radiative cooling is not neg-
ligible any longer and fitting the data would require an assumption of continued acceleration as
the outer shell flows down the jet. (vi) In a turbulent jet flow, efficient stochastic re-acceleration
may occur, which could change the picture. However, the current model calculations show that
synchrotron self-absorption may play a role in explaining the observed slower turn-on of the
4) Investigation of VHE γ-ray models for the M87 VHE/radio flare
The key questions for the understanding of the VHE γ-ray emission measured from the radio
galaxy M87 and from the more than 20 known VHE γ-ray blazars are: (i) What is the under-
lying particle distribution which is accelerated, (ii) what are the mechanisms to generate the
γ-rays, and (iii) where is the region of the emission located. In the following paragraphs a se-
lection of models discussed for M87 in the literature is investigated with a focus on the question
whether they can explain the observed VHE/radio light curves. Note, however, that some of the
models have difficulties in explaining the observed hard VHE γ-ray spectra [2, 5], which will
however not be discussed in more detail.
4.1) Black hole magnetosphere models
Models of VHE emission from the black hole magnetosphere.
anism of extraction of the rotational energy of black holes by the Blandford & Znajek scenario
The electromagnetic mech-
 seems to be a viable mechanism for powering the relativistic jets of AGN. Particles can
be accelerated by the electric field of vacuum gaps in the black hole magnetosphere  (the
electric field component parallel to the ordered magnetic field is not screened out) or due to
centrifugal acceleration in an active plasma-rich environment, where the parallel electric field is
screened . Synchrotron and curvature radiation of the charged particles, and inverse Comp-
ton scattering of thermal photons can produce VHE γ-ray photons . An important question
in this scenario is if the γ-rays can escape the central region or if they are absorbed through pair
creation processes with either photons from the accretion disk  or infrared photons emitted
by a potential dust torus for which no clear observational evidence is found so far [30, 44]. If
M87 harbours a non-standard (advection-dominated) accretion disk, γ-rays could escape with-
out being absorbed . An alternative scenario could be that the primary photons create a pair
cascade whose leakage produces the observed γ-ray emission . The delayed radio emission
could be explained by the effect of synchrotron self-absorption (see Sec. 3) or the time needed
to cool the electrons before they dominantly emit synchrotron radiation in the radio regime.
4.2) Hadronic jet models
The model of Reimer et al. (2004).
injected together with high-energy protons into a highly magnetized emission region . The
VHE emission is dominated by either µ±/π±synchrotron radiation or by proton synchrotron
radiation. The low-energy component is explained by the synchrotron emission of the electron
population. However, the radio flux is underestimated by the (steady-state) model (explanations
are discussed in the paper) so that a discussion of the observed radio/VHE flare is beyond the
scope of this particular model.
In this model a primary relativistic electron population is
4.3) Leptonic jet models
The model of Lenain et al. (2008).
to VHE) consists of small blobs (∼ 1014cm) travelling through the extended jet and radiating at
distances just beyond the Alfven surface . The emission takes place in the broadened jet for-
mation region, in the innermost part of the jet (corresponding to the central resolution element
in the 43GHz radio map). This multi-blob model is a two-flow model, where the fast, compact
blobs contribute to X-rays and γ-rays through the synchrotron self-Compton mechanism, and
are embedded in an extended, diluted and slower jet emitting synchrotron radiation from radio
to optical frequencies. Even though this model describes only steady state emission, the ob-
served radio/VHE variability can be discussed qualitatively. For instance, a sudden rise of the
density of the underlying leptonic population at the stationary shock (i.e. in the blobs) translates
into a flare of X-rays and γ-rays, but no immediate rise of radio emission is expected, because
the emission volume is synchrotron self-absorbed at radio frequencies, see Sec. 3. However, as
the flare propagates into the extended, less magnetized, neighbouring jet, the leptons in the jet
are energized and could cool by emitting at radio frequencies with some delay, creating a di-
In this model the high-energy emission region (X-rays up
luted radio flare in response to the VHE flare. An alternative scenario by Giannios et al. (2009)
explains the fast variability of VHE γ-ray radiation in blazars as a possible result from large
Lorentz factor (100) filaments within a more slowly moving jet flow .
The model of Tavecchio and Ghisellini (2008).
of a fast spine and slow sheath layer. The photons from the fast spine are external-Compton
boosted to VHE by the slower sheath. In this framework one may assume that the VHE flare
comes from near to the core and the time lag to the radio maximum is entirely a result of
propagation of some disturbance down the jet and the associated reduction in synchrotron self-
absorption. The X-rays should be synchrotron emission that is not self-absorbed and flare at
about the same time as the VHE γ-rays. In the radio data we see significant edge brightening
suggestive of a de-boosted spine. Note, however, that edge brightening like that observed can
also be produced by enhanced surface emissivity. In the model the radio emission originates
from a region different from that producing the VHE emission, so that a strict flux correlation
is not required. Detailed modeling would be needed to explain the observed light curves in the
framework of this model.
According to this model  the jet consists
The model of Georganopoulos et al. (2005).
of 0.1pc = 3 · 1017cm . The VHE emission is assumed to come from the fast moving
part near the jet base by inverse-Compton scattering of low-frequency photons from the slower
moving part of the jet. The model calculations are steady state so that they are difficult to apply
to an ejection event. However, one can assume that the VHE flare is directly associated with the
ejection event. As the disturbance propagates down the jet and decelerates we see the radio rise
later as a result of synchrotron self-absorption effects, similar to the model described in Sec. 3.
Here the X-ray flux can still rise and fall with the VHE if it comes directly from the disturbance;
the majority of the observed power comes from the slower part of the flow, but this is a more or
less steady state result. A similar model by Levinson (2007) of radiative decelerating blobs in
the jets of VHE γ-ray emitting blazars is described in .
In this model the jet decelerates over a length
4.4) Jet base / standing shock models
The model of Marscher et al. (2008).
flare in BLLac . The first flare is seen at X-ray/optical energies accompanied by polariza-
tion measurements at a date of 2005.82. It indicates an injection event into the jet acceleration
and collimation region near to the black hole. The flare was followed by the appearance of
a new radio component in VLBA images that approached and passed through the radio core,
accompanied by a second X-ray/optical flare at 2005.92, where the 14.5GHz radio flux begins
to increase and peaks at about 2006.0. This is interpreted as the passage of the disturbance
through a standing shock in the jet at a distance of ∼ 104Rs, see Fig. 3 in . VHE γ-ray emis-
sion has been detected in 2005, however, no evidence for flux variability has been found in the
data. In this picture the radio core is located at the standing shock and the radio emission in the
The model is based on the observation of a double