Content uploaded by Jowena Lyn Contreras
Author content
All content in this area was uploaded by Jowena Lyn Contreras on Mar 06, 2023
Content may be subject to copyright.
arXiv:2111.12926v1 [astro-ph.HE] 25 Nov 2021
MNRAS 000,1–21 (2020) Preprint 29 November 2021 Compiled using MNRAS L
A
T
E
X style file v3.0
Multiwavelength study of the gravitationally lensed blazar QSO
B0218+357 between 2016 and 2020.
V. A. Acciari1, S. Ansoldi2, L. A. Antonelli3, A. Arbet Engels4, M. Artero5, K. Asano6,
D. Baack7, A. Babi´c8, A. Baquero9, U. Barres de Almeida10, J. A. Barrio9, I. Batkovi´c11,
J. Becerra Gonz´alez1, W. Bednarek12, L. Bellizzi13, E. Bernardini14 , M. Bernardos11, A. Berti15,
J. Besenrieder16, W. Bhattacharyya14, C. Bigongiari3, A. Biland4, O. Blanch5, G. Bonnoli13,
ˇ
Z. Boˇsnjak8, G. Busetto11, R. Carosi17, G. Ceribella16, M. Cerruti18, Y. Chai16, A. Chilingarian19,
S. Cikota8, S. M. Colak5, E. Colombo1, J. L. Contreras9, J. Cortina20, S. Covino3, G. D’Amico16,
V. D’Elia3, P. Da Vela17,38, F. Dazzi3, A. De Angelis11 , B. De Lotto2, M. Delfino5,39, J. Delgado5,39,
C. Delgado Mendez20, D. Depaoli15, F. Di Pierro15, L. Di Venere21, E. Do Souto Espi˜neira5, D. Do-
minis Prester22, A. Donini2, D. Dorner23, M. Doro11, D. Elsaesser7, V. Fallah Ramazani24,40,
A. Fattorini7, G. Ferrara3, M. V. Fonseca9, L. Font25, C. Fruck16, S. Fukami6, R. J. Garc´ıa
L´opez1, M. Garczarczyk14 , S. Gasparyan26, M. Gaug25, N. Giglietto21, F. Giordano21, P. Gliwny12,
N. Godinovi´c27, J. G. Green3, D. Green16, D. Hadasch6, A. Hahn16, L. Heckmann16, J. Herrera1,
J. Hoang9, D. Hrupec28 , M. H¨
utten16, T. Inada6, S. Inoue29, K. Ishio16 , Y. Iwamura6, I. Jim´enez20,
J. Jormanainen24, L. Jouvin5, Y. Kajiwara30, M. Karjalainen1, D. Kerszberg5, Y. Kobayashi6,
H. Kubo30, J. Kushida31, A. Lamastra3⋆, D. Lelas27 , F. Leone3, E. Lindfors24⋆, S. Lombardi3,
F. Longo2,41, R. L´opez-Coto11, M. L´opez-Moya9, A. L´opez-Oramas1, S. Loporchio21, B. Machado
de Oliveira Fraga10, C. Maggio25, P. Majumdar32, M. Makariev33, M. Mallamaci11, G. Maneva33,
M. Manganaro22⋆, K. Mannheim23, L. Maraschi3, M. Mariotti11, M. Mart´ınez5, D. Mazin6,42,
S. Menchiari13, S. Mender7, S. Mi´canovi´c22, D. Miceli2, T. Miener9, M. Minev33, J. M. Miranda13,
R. Mirzoyan16, E. Molina18 , A. Moralejo5, D. Morcuende9, V. Moreno25, E. Moretti5, V. Neustroev34 ,
C. Nigro5, K. Nilsson24, K. Nishijima31 , K. Noda6, S. Nozaki30, Y. Ohtani6, T. Oka30, J. Otero-
Santos1, S. Paiano3, M. Palatiello2, D. Paneque16, R. Paoletti13, J. M. Paredes18, L. Pavleti´c22,
P. Pe˜nil9, C. Perennes11, M. Persic2,43, P. G. Prada Moroni17, E. Prandini11, C. Priyadarshi5,
I. Puljak27, W. Rhode7, M. Rib´o18, J. Rico5, C. Righi3, A. Rugliancich17, L. Saha9, N. Sahakyan26,
T. Saito6, S. Sakurai6, K. Satalecka14, F. G. Saturni3, B. Schleicher23, K. Schmidt7, T. Schweizer16,
J. Sitarek12⋆, I. ˇ
Snidari´c35, D. Sobczynska12, A. Spolon11, A. Stamerra3, D. Strom16, M. Strzys6,
Y. Suda16, T. Suri´c35, M. Takahashi6, F. Tavecchio3, P. Temnikov33, T. Terzi´c22, M. Teshima16,44,
L. Tosti36 , S. Truzzi13, A. Tutone3, S. Ubach25, J. van Scherpenberg16, G. Vanzo1, M. Vazquez
Acosta1, S. Ventura13, V. Verguilov33, C. F. Vigorito15, V. Vitale37 , I. Vovk6, M. Will16,
C. Wunderlich13, D. Zari´c27, F. de Palma45,46⋆, F. D’Ammando47, A. Barnacka60,61, D. K. Sahu48,
M. Hodges49, T. Hovatta50,51 , S. Kiehlmann52,53, W. Max-Moerbeck54, A. C. S. Readhead49,
R. Reeves55, T. J. Pearson49, A. L¨
ahteenm¨
aki51,56, I. Bj¨
orklund51,56, M. Tornikoski51, J. Tammi51,
S. Suutarinen51, K. Hada57,58, K. Niinuma59
Affiliations are listed at the end of the paper
Accepted XXX. Received YYY; in original form ZZZ
©2020 The Authors
2V. A. Acciari et. al.
ABSTRACT
We report multiwavelength observations of the gravitationally lensed blazar QSO B0218+357 in 2016-2020. Optical,
X-ray and GeV flares were detected. The contemporaneous MAGIC observations do not show significant very-high-
energy (VHE, &100 GeV) gamma-ray emission. The lack of enhancement in radio emission measured by OVRO
indicates the multi-zone nature of the emission from this object. We constrain the VHE duty cycle of the source to
be <16 2014-like flares per year (95% confidence). For the first time for this source, a broadband low-state SED
is constructed with a deep exposure up to the VHE range. A flux upper limit on the low-state VHE gamma-ray
emission of an order of magnitude below that of the 2014 flare is determined. The X-ray data are used to fit the
column density of (8.10 ±0.93stat)×1021cm−2of the dust in the lensing galaxy. VLBI observations show a clear radio
core and jet components in both lensed images, yet no significant movement of the components is seen. The radio
measurements are used to model the source-lens-observer geometry and determine the magnifications and time delays
for both components. The quiescent emission is modeled with the high-energy bump explained as a combination of
synchrotron-self-Compton and external Compton emission from a region located outside of the broad line region. The
bulk of the low-energy emission is explained as originating from a tens-of-parsecs scale jet.
Key words: Gamma rays: galaxies – Gravitational lensing: strong – Galaxies: jets – Radiation mechanisms: non-
thermal – quasars: individual: QSO B0218+357
1 INTRODUCTION
QSO B0218+357, also known as S3 0218+35, is one of only
a handful of Flat Spectrum Radio Quasars (FSRQs) de-
tected in very-high-energy (VHE, &100 GeV) gamma-ray
emission. It has a redshift of zs=0.944 ±0.002 (Cohen et al.
2003;Paiano et al. 2017). The source showed strong variabil-
ity in the GeV range in 2012 (Cheung et al. 2014) when a
series of flares was observed by Fermi Large Area Telescope
(LAT). Another flare was observed by Fermi-LAT in 2014,
and during the follow-up the source was discovered in VHE
gamma rays by MAGIC telescopes (Buson et al. 2015a,b;
Ahnen et al. 2016). Similarly to QSO B0218+357, GeV emis-
sion from the second gravitationally-lensed source detected
by Fermi -LAT, PKS 1830-211, also shows evidence of a mea-
sured delay between different lens images (Barnacka et al.
2011). Observations of PKS 1830-211 by the H.E.S.S. tele-
scopes following a GeV flare did not show any significant
gamma-ray emission (Abdalla et al. 2019).
QSO B0218+357 is gravitationally lensed by B0218+357 G,
a spiral galaxy seen face-on at a redshift of zl=0.68466 ±
0.00004 (Carilli et al. 1993). Strong gravitational lensing is
observed when the lens is a galaxy or a cluster of galaxies.
Such a massive lens can produce multiple images of the source
separated by ∼arcseconds. Thus, the images of strongly
lensed sources can be well resolved at wavelengths from radio
to X-rays with existing instruments.
Stars can cause further lensing effects within a lensing
galaxy. In such cases, the deflection angle of lensed images
is of the order of microarcseconds. Thus, the effect is called
microlensing. The change in position of microlensed images
cannot be observed with existing instruments. The microlens-
ing effect is observed as changes in the flux of the strongly
lensed image.
The relative flux densities observed for lensed images de-
pend on the geometry of the source-lens-observer system, and
⋆Corresponding authors (contact.magic@mpp.mpg.de):
J. Sitarek, A. Lamastra, E. Lindfors, M. Manganaro, F. de
Palma
can be affected by microlensing. Further, different geometri-
cal paths and gravitational delays cause the emission to ar-
rive at different times in various images. In the case of QSO
B0218+357 the image is composed of two images A and B sep-
arated by only 335 mas and an Einstein ring of a similar size
(O’Dea et al. 1992). The A component (located westwards)
is brighter and this signal precedes that from component B.
Variable radio emission observed in 1992/1993 and
1996/1997 with the Very Large Array at 5, 8.4 and
15 GHz yields time delays between these two com-
ponents of 12 ±3d (Corbett et al. 1996), 10.5±0.4d
(Biggs et al. 1999), 10.1+1.5
−1.6d, (Cohen et al. 2000), 11.8±
2.3d (Eulaers & Magain 2011), 11.3±0.2d (Biggs & Browne
2018). The statistical analysis of the 2012 high state Fermi-
LAT >0.1GeV light curve auto-correlation function led to a
similar value of the time delay (11.46 ±0.16 d, Cheung et al.
2014). These values are consistent with the delay between the
two components of the 2014 Fermi-LAT flare (Ahnen et al.
2016).
The gamma-ray emission of QSO B0218+357 comprises
many flares with timescales as short as a few hours. The
short timescales of gamma-ray flares combined with the abil-
ity of the Fermi-LAT observatory to monitor the sky contin-
uously allow one to search for delayed counterparts of flares
and put constraints on the magnification ratio. For example,
the two-night-long sub-TeV flare was observed contempora-
neously with the detection of the image B flare in Fermi-LAT
(Ahnen et al. 2016)
Unfortunately, since the MAGIC observations in 2014 only
covered the time of the B image of the flare, no measurement
of the magnification ratio or delay could be obtained. Moni-
toring of QSO B0218+357 with Cherenkov telescopes during
flaring events could enable the capture of multiple flares and
constrain models of the magnification ratio and time delays.
At radio frequencies the B component is 3.57-3.73 times
fainter than the A component (Biggs et al. 1999). However,
the observed ratio of magnification varies with the radio fre-
quency (Mittal et al. 2006), possibly due to free-free absorp-
tion in the lens (Mittal et al. 2007). In the optical range the
MNRAS 000,1–21 (2020)
MWL study of QSO B0218+357 3
leading image is strongly absorbed, inverting the brightness
ratio of the two images (Falco et al. 1999). It has been sug-
gested that the optical absorption occurs in the host galaxy
rather than the lens (Falomo et al. 2017). Interestingly, the
magnification ratio observed at GeV energies shows variabil-
ity. The average GeV magnification factor during 2012 high
state was estimated to be ∼1(Cheung et al. 2014), while
during the 2014 flare it was comparable to or even larger
than that measured at radio frequencies (Ahnen et al. 2016).
Changes in the observed GeV magnification ratio can be in-
terpreted as microlensing effects either due to individual stars
(Vovk & Neronov 2016) or due to larger scale structures in
the lens (Sitarek & Bednarek 2016).
Because it takes about 1-2 days for Fermi -LAT data to be
collected, downlinked and processed, it is difficult to trigger
observations for phenomena with similar durations, like the
two-night 2014 flare. Therefore, the shortness of the VHE
gamma-ray emission significantly hinders the possibility of
Target of Opportunity observations of a flare in both im-
ages. In addition observational visibility constraints further
limit the possibility of a follow up of the delayed emission
with ground based instruments. Thus, since 2016, we have
taken advantage of the 11 days delay to trigger MAGIC. Ob-
servational windows which allow visibility under favourable
zenith angle conditions in moon-less nights 11 days after each
slot have been identified. During these time slots MAGIC ob-
servations were performed, and contemporaneous multiwave-
length (MWL) coverage from radio to GeV gamma-rays was
obtained. In this paper the results of these observations are
reported. Additionally, a multiwavelength campaign on QSO
B0218+357, organized in August 2020 in response to a hint
of enhanced activity in the source, is also reported.
In Section 2the instruments that took part in the MWL
campaign, the data taken and the analysis methods are de-
scribed. The results are reported in Section 3. In Section 4,
the broadband emission of the low state of the source is mod-
eled. The paper concludes with a summary of the results in
Section 5.
2 OBSERVATIONS AND DATA ANALYSIS
QSO B0218+357 was observed over a broad energy range:
radio (The Owens Valley Radio Observatory, OVRO), radio
interferometry (a joint VLBI array of KVN (Korean VLBI
Network) and VERA (VLBI Exploration of Radio Astrome-
try), KaVA), optical (Kunliga Vetenskapsakademi, KVA and
Nordic Optical telescope, NOT; Neil Gehrels Swift obser-
vatory (Swift) Ultraviolet/Optical Telescope (Swift-UVOT)
and X-ray Multi-Mirror Optical Monitor (XMM -OM)), X-
ray (X-ray Telescope (Swift -XRT) and XMM -Newton), GeV
gamma rays (Fermi-LAT) and VHE gamma rays (MAGIC).
During the August 2020 MWL campaign dedicated obser-
vations by Himalayan Chandra Telescope (HCT), Joan Or´o
Telescope (TJO) and Mets¨
ahovi were taken.
The historical data obtained via the Space Science
Data Center1from the following catalogs are also used:
CLASS (Myers et al. 2003), JVASPOL (Jackson et al.
2007), KUEHR (Kuehr et al. 1981), NIEPPOCAT
1SSDC, http://www.ssdc.asi.it/
(Nieppola et al. 2007), NVSS (Condon et al. 1998),
Planck (Planck Collaboration et al. 2011,2014,2016),
GB6 (Gregory et al. 1996), GB87CAT (Gregory & Condon
1991), WMAP5 (Wright et al. 2009), WISE (Wright et al.
2010), 1SWXRT (D’Elia et al. 2013), 1SXPS (Evans et al.
2014), FGL (Abdo et al. 2010;Nolan et al. 2012;Acero et al.
2015).
2.1 MAGIC
MAGIC is a system of two imaging atmospheric Cherenkov
telescopes with a mirror dish diameter of 17 m each. The
telescopes are located in the Canary Islands, on La Palma
(28.7◦N, 17.9◦W), at a height of 2200 m above sea level
(Aleksi´c et al. 2016a). The data were analyzed using MARS,
the standard analysis package of MAGIC (Zanin et al. 2013;
Aleksi´c et al. 2016b). Wherever applicable, upper limits on
the flux were computed following the approach of Rolke et al.
(2005) using a 95% confidence level and assuming a 30% total
systematic uncertainty on the collection area.
The regular monitoring observations were performed be-
tween MJD 57397 and 58875 in dark night conditions. The
monitoring time slots were scheduled so as to allow for possi-
ble observations in ∼11 days if enhanced emission was seen.
This results in possible observation slots (up to two per moon
period) lasting between 1 and 6 days. Motivated by the 2-day
duration of the 2014 VHE flare, in such slots observations
every second night were scheduled (on some occasions this
scheme was modified due to weather conditions or competing
sources). After the data selection, based mainly on the atmo-
spheric transmission measured with LIDAR (Fruck & Gaug
2015) and on hadronic background rates, the data set consists
of 72.7hr, spread over 73 nights.
Since MJD 58122 the data have been taken with the novel
Sum-Trigger-II (Dazzi et al. 2021). The Sum-Trigger-II part
of the dataset consists of 38.4 hr and was analyzed with ded-
icated low-energy analysis procedures including a special im-
age cleaning procedure (Shayduk 2013;Ceribella et al. 2019).
Additionally, during the August 2020 campaign, 2 hr of
good quality data were taken on MJD 59081 and 59082. Due
to a forest fire observations on MJD 59083 could not be used.
2.2 Fermi -LAT
The LAT is a pair conversion detector on the Fermi Gamma-
ray Space Telescope, which was launched on June 11, 2008.
It observes the whole sky every three hours in the energy
range between a few tens of MeV and few TeV (Atwood et al.
2009). The Fermi-LAT data taken between MJD 56929 and
58876 in the energy range 100 MeV – 2 TeV in a re-
gion of interest of 15◦were selected. The data were pro-
cessed using the Fermitools version 1.2.23 and Fermipy
(Wood, et al. 2017) version 0.19.0, with instrument response
function P8R3 SOURCE V2. The data were binned in 8 en-
ergy bins per decade and in spatial bins of 0.1◦. To reduce
the contamination from the Earth limb, a zenith angle cut
of 90◦was applied to the data. The model used in the like-
lihood analysis is composed of the sources listed in the LAT
8-year Source Catalog (4FGL, Abdollahi, et al. 2020) that are
within 20◦of the QSO B0218+357 location, the latest inter-
stellar emission model (gll iem v07), and an isotropic back-
ground model (iso P8R3 SOURCE V2 v1). At the beginning
MNRAS 000,1–21 (2020)
4V. A. Acciari et. al.
of the analysis, we iteratively optimized our spectral source
models using fermipy’s optimization method. Sources with
a Test Statistic2(TS) lower than 1 were removed from the
fit. Four new point sources with a TS higher than 16 (∼4
σ
significance) within 10◦of QSO B0218+357 were added it-
eratively, in order to account for emission not modeled by
known background sources (RAJ2000,DecJ2000=30.54◦, 39.67◦;
35.55◦, 37.53◦3; 31.72◦, 38.56◦; 43.58◦, 33.63◦). For each of
these sources a power-law spectral model was used and iter-
atively optimized. The closest new source is 1.6◦away from
QSO B0218+357, and has a TS slightly above 40. The effect
of energy dispersion4(reconstructed event energy differing
from the true energy of the incoming photon) is accounted
for by generating a detector response matrix with two addi-
tional energy bins in logEtr ue above and below the considered
energy range (edisp bins = -2). This method is applied to all
the sources in the model except for the isotropic background
which was derived from dispersed data. The normalisation
of both diffuse components in the source model were allowed
to vary during the spectral fitting procedure. In the whole
interval analysis, sources within 7◦from QSO B0218+357
had their normalisation free to vary, sources within 5◦from
QSO B0218+357 had also their spectral index free to vary.
In both cases this selection was applied only to sources with
a TS in the full time interval MJD 56929-58876 higher than
10. The QSO B0218+357 was modeled with a LogParabolic
spectrum:
dN
dE =N0E
Eb−(
α
+
β
log(E/Eb))
(1)
as in the 4FGL. In the overall analysis, the source was ob-
served with a TS of 7678 ( ∼87
σ
) with a flux of (9.70 ±0.31)×
10−8cm−2s−1above 0.1 GeV. The spectral energy distribu-
tion (SED) was also evaluated over the whole time range,
and over only the days in which QSO B0218+357 was ob-
served by MAGIC. In the second case the summed likelihood
technique was used for combining the analysis in the different
time bins.
In order to estimate weekly and daily fluxes of QSO
B0218+357 the number of free spectral parameters is lim-
ited. The sources located within 7◦from QSO B0218+357
had their normalisation set as a free parameter if their vari-
ability index was higher than 18.4835, while all sources within
5◦from QSO B0218+357 had their normalization free to vary
if their TS was higher than 40 integrated over the full time
period. The spectral indices of all the sources with free nor-
malisation were left as a free parameter if the source showed
a TS value higher than 25 over the integration time (weekly
or daily), in all the other cases the indexes where frozen to
the value obtained in the overall fit.
2The Test Statistic is defined as T S =−2ln(Lmax,0/Lmax,1), where
Lmax,0is the maximum likelihood value for a model without an
additional source and Lmax,1is the maximum likelihood value for a
model with the additional source. It is a measure of the detection
significance of a source.
34FGL J0221.8+3730 is a new source in
the LAT 10-year Source Catalog (4FGL-DR2
https://fermi.gsfc.nasa.gov/ssc/data/access/lat/10yr_catalog/)
compatible with this location.
4https://fermi.gsfc.nasa.gov/ssc/data/analysis/documentation/Pass8_edisp_usage.html
5The level was chosen according to the 4FGL catalogue.
The spectral analysis was also performed in two differ-
ent smaller intervals corresponding to the optical/GeV flare
(MJD 57600-57700) and to the X-ray flare (MJD 58860.7 –
58866.7). The source was significantly detected in both time
intervals, with a significance of 45.9
σ
and 6.6
σ
, respectively.
2.3 XMM-Newton
XMM-Newton (Jansen et al. 2001) observed the source four
times between August 2019 and January 2020. The integra-
tion times of the observations were in the range of 11.3 ks
and 19.8 ks. The EPIC pn CCD camera (Str¨
uder et al. 2001)
operated in full-frame mode with the medium filter applied
during all the observations. The data were processed using
the XMM-Newton Science Analysis System (SAS v.18.0.0,
Gabriel et al. 2004) following standard settings and using the
calibration files available at the time of the data reduction.
The EPIC pn Observation Data Files (ODFs) were processed
with the SAS-task epproc in order to generate the event files.
Event files were cleaned of bad pixels, and events spread at
most in two contiguous pixels (PATTERN≤4) were selected.
Periods of high background levels were removed by analysing
the light curves of the count rate at energies higher than 10
keV. The resulting net-exposure times are reported in Table
4. In order to include all of the source counts and simultane-
ously minimise the background contribution, source counts
were extracted from a circular region of radius between 30
and 35 arcsec. The background counts were extracted from
a circular region of radius 50-65 arcsec located on a blank
area of the detector close to the source. Response matrices
for spectral fitting were obtained using the SAS-task rmfgen
and arfgen. All the spectra were binned in order to have
no less than 20 counts in each background-subtracted spec-
tral channel, and the instrumental energy resolution was not
oversampled by a factor larger than 3.
X-ray spectral analyses were carried out with XSPEC
v.12.9.1 (Arnaud 1996). No variability in the spectra of the
XMM-Newton observations performed at MJD 58697, 58721,
and 58724 (obs. ID 0850400301, 0850400401, 0850400501)
was observed, thus all the observations were combined with
the SAS-task epicspeccombine for the spectral modelling of
the source. In contrast, the observation performed at MJD
58863.7 (obs. ID 0850400601) indicated an increase of the
X-ray flux by a factor of ∼1.4 with respect to previous obser-
vations, thus this spectrum was fitted separately.
2.4 Swift-XRT
The X-ray Telescope (XRT, Burrows et al. 2004) on-board
the Neil Gehrels Swift observatory (Swift) observed the source
four times between January 2016 and January 2020. Addi-
tionally seven pointings were taken around the time of the
August 2020 campaign. Due to the source faintness, all of
these observations were performed in photon counting mode.
The event lists for the period of interest were downloaded
from the publicly available SWIFTXRLOG (Swift -XRT In-
strument Log)6. The data were processed using the stan-
dard data analysis procedure (Evans et al. 2009) and the
configuration described by Fallah Ramazani et al. (2017) for
6https://heasarc.gsfc.nasa.gov/W3Browse/swift/swiftxrlog.html
MNRAS 000,1–21 (2020)
MWL study of QSO B0218+357 5
blazars. The spectra of each observation were binned in a way
that each bin contains one count. Therefore, the maximum
likelihood-based statistic for Poisson data (Cash statistics,
Cash 1979) method was used in the spectral fitting proce-
dure and flux measurements of individual observations.
No spectral variability was observed within the Swift-
XRT data. In order to evaluate the average state of
the source during the monitoring, two combined spec-
tra were produced using the observational data taken
during 2016-2017 (OBSIDs 00032533003, 00032533005,
00032533006, and 00032533007) and August 2020 (OB-
SIDs 00032533008, 00032533009, 00032533010, 00032533011,
00032533012, 00032533013, 00032533014, 00032533015).
These spectra are binned in a way that each bin contains
20 counts. Therefore, the maximum likelihood-based statis-
tic for Gaussian data method is the spectral fitting procedure
of these two spectra.
2.5 UV
During the four monitoring Swift pointings, the UVOT in-
strument observed the source in the uoptical photometric
band (Poole et al. 2008;Breeveld et al. 2010). The data were
analysed using the uvotimsum and uvotsource tasks included
in the HEAsoft package (v6.28) with the 20201026 release of
the Swift/UVOTA CALDB. Source counts were extracted
from a circular region of 5 arcsec radius centered on the
source, while background counts were derived from a circular
region of 20 arcsec radius in a nearby source-free region. The
source was not detected with a significance higher than 3
σ
in
the single observations, therefore the four UVOT images were
summed using the uvotimsum task and analyzed the summed
image with the uvotsource task.
The Optical Monitor (OM) observed the source four times
in ufilters in imaging mode. The total exposure times of
the imaging observations were: 16400 s, 17700 s, 11300 s,
and 19800 s. The data were processed using the SAS task
omichain. For the count rate to flux conversion the conversion
factors given in the SAS watchout dedicated page7were used
.
During the August 2020 campaign five observations with
Swift-UVOT were taken in uband. None of these pointings
resulted in the detection of a signal above 2
σ
significance.
Similarly to 2016–2020 monitoring data, a stacked analysis
was performed to evaluate the average emission in this period.
The UVOT and OM flux densities were corrected
for Galactic extinction using a value AU= 0.299 mag
(Schlafly & Finkbeiner 2011).
2.6 Optical
The source was monitored with the NOT and 35cm Celestron
telescope attached to the KVA telescope. Both telescopes
are located at the same site as the MAGIC telescopes and
the NOT observations were carried out quasi-simultaneously
with MAGIC, while KVA performed additional monitoring
more frequently. Starting in July 2020 the source was also
7https://www.cosmos.esa.int/web/xmm-newton/sas-watchout-
uvflux.
monitored with the TJO at the Montsec Astronomical Ob-
servatory8. In addition during the August MWL campaign
the source was observed with the HCT 9.
NOT observations were carried out using ALFOSC in B,
V, R and I-bands, while the KVA and TJO observations op-
erated in R-band only. The data were analyzed using the
semi-automatic pipeline and standard procedures of differ-
ential photometry (Nilsson et al. 2018). The same compari-
son and control stars were used as in Ahnen et al. (2016)10 .
The r-,g- and i-band magnitudes of the stars were avail-
able in the PANSTARRS database11 . The magnitudes in
B, V and I-band were calculated from i-, r- and g- band
magnitudes using the formulae of Lupton (2005) 12 . Using
those formula, consistency of the R-band values derived in
Ahnen et al. (2016) was checked. The I-band filter used at
the NOT differs from the standard I-band filter enough for
the color correction to become significant for a very red input
spectrum as in this case (Falomo et al. 2017). The spectrum
obtained by Falomo et al. (2017) was downloaded from the
ZBLLLAC repository13 and synthetic photometry was per-
formed through the standard I-band and the NOT I-band.
This showed that the NOT I-band magnitudes needed to be
corrected by +0.08 mag to transform to the standard system.
The aperture used was 3 arcsec, slightly smaller than the 4
arcsec used for KVA and TJO data.
The source was observed with HCT in four epochs (MJD
59081–59085). The observations were carried out in the
Bessell U, B, V, R and I bands available with HFOSC. The
data were reduced in a standard manner using various tasks
available in IRAF. Aperture photometry was performed on
the source and nearby stars. The standard magnitude of the
source was obtained using differential photometry with the
same comparison and control stars as used by Ahnen et al.
(2016).
The observed magnitudes were corrected for the galac-
tic extinction using values obtained from the NED14
(Schlafly & Finkbeiner 2011) and listed in Table 1. The mag-
nitudes were converted to flux densities using formula F=
F0·10−mag/2.5with F0=4260 Jy in B, F0=3640 Jy in V,
F0=3080 Jy in R and F0=2550 Jy in I.
The flux densities needed to be corrected for the
contribution from the host galaxy of QSO B0218+357
at (zs=0.944) and the lens galaxy at zl=0.684.
Jackson, Xanthopoulos & Browne (2000) imaged this target
with the HST through the F555W,F814Wand F160Wfilters
and measured the flux density from the face-on spiral galaxy
to be (6±2), (13±2) and (15±2) (10−18 erg s−1cm −2˚
A−1),
respectively. Since QSO B0218+357 is classified as a FSRQ
(Abdollahi, et al. 2020;Paiano et al. 2017), the host galaxy is
likely to be a luminous (MK∼ −26.5) bulge-dominated galaxy
(e.g. Olgu´ın-Iglesias et al. 2016). The R-band magnitude of
such a galaxy at the redshift of 0.944 would be ∼22.5mag. An
early-type galaxy template was taken from Mannucci et al.
8http://www.ieec.cat/en/content/210/telescope-and- dome
9https://www.iiap.res.in/iao/2mtel.html
10 The stars are marked in finding chart available at:
http://users.utu.fi/kani/1m/finding_charts/B2_0218+35_map.html
11 https://catalogs.mast.stsci.edu/panstarrs/
12 http://classic.sdss.org/dr4/algorithms/sdssUBVRITransform.html#Lupton2005
13 https://web.oapd.inaf.it/zbllac/
14 https://ned.ipac.caltech.edu
MNRAS 000,1–21 (2020)
6V. A. Acciari et. al.
Filter AXgalaxy flux density [mJy]
B 0.25 1.4
V 0.189 4.4
R 0.15 12
I 0.104 31
Table 1. Galactic absorption values and contribution of the galaxy
within the aperture in each filter.
(2001), redshifted to z=0.944 and integrated over the R-band
filter transmission. Then the template was scaled to match
the integrated flux density to R = 22.5. The scaled spec-
trum corresponds to ∼40% of the flux densities observed
by Jackson, Xanthopoulos & Browne (2000), i.e. a signifi-
cant part of the “spiral galaxy” surrounding component B
could actually be the host galaxy. This is what Falomo et al.
(2017) propose based on a high signal-to-noise ratio spec-
trum of B0218+357. Their spectrum shows gaseous absorp-
tion lines at the lens redshift, but no stellar photospheric
lines are detected, which led them to propose that the spi-
ral structure belongs to the host galaxy, not the lens. It is
impossible to determine the relative contributions of the lens
galaxy and the host galaxy from the present data, especially
since the latter may also be lensed and absorbed by the for-
mer. A simple assumption, that 100% of the flux densities de-
termined by Jackson, Xanthopoulos & Browne (2000) arise
from the lens is used. Thus a fit of a late type (Sa) galaxy
template from Mannucci et al. (2001), redshifted to 0.684, to
the Jackson, Xanthopoulos & Browne (2000) flux densities
was carried out. Then synthetic photometry was performed
through the BVRI bands to the fitted template to obtain flux
densities within the aperture for each filter, and these values
are reported in Table 1. These values were then subtracted
from total flux densities.
The August 2020 observations with NOT were interrupted
by the forest fire on MJD 59084. The data from MJD 59083
have lower signal to noise ratio and gradients in the back-
ground, resulting in larger than usual reported uncertainties
in our analysis.
2.7 Radio
Between January 2017 and January 2019, QSO B0218+357
was frequently observed with KaVA at 22 and 43 GHz. A to-
tal of 16 sessions were performed during this period. In most
cases each session lasted 2 consecutive nights, with a 5–8-
hour track at 22 GHz on the first day and a similar track
at 43 GHz on the following day. By default 7 stations (3
from KVN and 4 from VERA) joined each session. How-
ever, occasionally VERA-Mizusawa or VERA-Ishigaki was
missing due to local issues. In addition, triggered by the
August 2020 campaign, KaVA performed follow-up observa-
tions at 43 GHz for a total of 9 sessions between 2020 Au-
gust 22 and October 8. The observing time of each follow-
up session was 3.5–4 hours and on average 5–6 stations
joined. All the data were recorded at 1 Gbps (a total band-
width of 256 MHz with eight 32-MHz subbands) with left-
hand circular polarization and correlated by the Daejeon
hardware correlator (Lee et al. 2015). The initial data cali-
bration (amplitude, phase, bandpass) was performed using
the National Radio Astronomy Observatory (NRAO) As-
tronomical Image Processing System (AIPS; Greisen 2003)
based on the standard KaVA/VLBI data reduction proce-
dures (Niinuma et al. 2014;Hada et al. 2017). Imaging was
performed using Difmap software (Shepherd et al. 1994) with
the standard CLEAN and self-calibration procedures.
During the 4 KaVA 43 GHz sessions made on January
15th 2017 (MJD 57768), October 17th 2017 (MJD 58043),
November 12th 2017 (MJD 58069) and January 5th 2018
(MJD 58123), additional simultaneous observations of the
source with the KVN-only array with a 43 GHz/86 GHz
dual-frequency recording mode were carried out. A wide-
band 4 Gbps mode was used where each frequency band was
recorded at 2 Gbps (a bandwidth of 512 MHz for each band).
86 GHz fringes were detected by transferring the solutions
derived at 43 GHz using the frequency-phase transfer (FPT)
technique (e.g., Algaba et al. 2015;Zhao et al. 2019). Imag-
ing was carried out in Difmap.
Some of the KaVA/KVN results are presented in
Hada et al. (2020) together with detailed radio images and
analyses at each frequency. See Hada et al. (2020) for full
details of the KaVA/KVN data reduction and imaging pro-
cedures. The typical angular resolution of KaVA (a maximum
baseline length D=2300 km) is 1.2 mas (22 GHz) and 0.6 mas
(43 GHz), that of KVN (D=560 km) is 1 mas at 86 GHz. Here
we report on the whole data set, and investigate the kinemat-
ics of the jet.
The OVRO 40-Meter Telescope uses off-axis dual-beam op-
tics and a cryogenic receiver with 2 GHz equivalent noise
bandwidth centered at 15 GHz. The double switching tech-
nique (Readhead et al. 1989), where the observations are con-
ducted in an ON-ON fashion so that one of the beams is al-
ways pointed on the source, was used to remove gain fluc-
tuations and atmospheric and ground contributions. Until
May 2014 a Dicke switch was used to alternate rapidly be-
tween the two beams. Since May 2014 a 180 degree phase
switch has been used, with a new pseudo-correlation receiver.
Gain drifts were compensated with a calibration relative to
a temperature-stabilized noise diode. The primary flux den-
sity calibrator was 3C 286 with an assumed value of 3.44 Jy
(Baars et al. 1977). DR21 was used as secondary calibrator.
Richards et al. (2011) describe the observations and data re-
ductions in detail. Since the telescope is a single dish, it mea-
sures total flux densities integrated over the whole lensed
structure: A, B and the Einstein ring.
The 37 GHz observations taken during the August 2020
campaign were made with the 13.7 m diameter Mets¨
ahovi
radio telescope. The observations were ON–ON observations,
alternating the source and the sky in each feed horn. A typ-
ical integration time to obtain one flux density data point
was between 1200 and 1400 s. The detection limit of the tele-
scope at 37 GHz was on the order of 0.2 Jy under optimal
conditions. Data points with a signal-to-noise ratio <4were
treated as non-detections. The flux density scale was set by
observations of DR 21. Sources NGC 7027, 3C 274 and 3C 84
were used as secondary calibrators. A detailed description of
the data reduction and analysis is given in Ter¨
asranta et al.
(1998). The error estimate in the flux density includes the
contribution from the measurement rms and the uncertainty
of the absolute calibration.
MNRAS 000,1–21 (2020)
MWL study of QSO B0218+357 7
3 RESULTS
The MWL light curves measured during the monitoring cam-
paign are presented in Fig. 1.
3.1 Search for VHE emission
No significant VHE gamma-ray emission was found in the to-
tal data set of MAGIC monitoring data (see the left panel of
Fig. 2). Due to expected variability of the emission an addi-
tional analysis separating the data set into individual nights
was performed. The distribution of the significances of the
measured excess is shown in the right panel of Fig. 2, and the
upper limits on the flux above 100 GeV are reported in the
top panel of Fig. 1. As the source is a known VHE gamma-
ray emitter, we also report the nominal flux values on each
observation night, however none of them is significant, com-
paring to the corresponding uncertainty bar. The distribution
is consistent with the lack of a measurable gamma-ray excess.
By using the Fermi-LAT data, an additional study was per-
formed to evaluate the expected VHE gamma-ray flux on
individual nights (see Appendix A), however no clear hard
GeV states could be identified.
The SED upper limits were computed from the total moni-
toring sample of the MAGIC observations and compared with
the extrapolation of the Fermi-LAT SED (see Fig. 3). The
Fermi-LAT data for this comparison are quasi-simultaneous,
i.e. 24 hr-long time windows centered on each MAGIC obser-
vations are stacked together. The MAGIC upper limits are
an order of magnitude below the flux observed during the
flare in 2014 (Ahnen et al. 2016). However, within the uncer-
tainties of the Fermi-LAT extrapolation the upper limits are
in agreement with a power-law SED from GeV to sub-TeV
range.
In order to constrain the VHE gamma-ray duty cycle of the
source the nights with optimal exposure were selected. The
data set contains 37 nights with exposure >100 GeV of at
least 1.4×1012 cm2s(corresponding to about 1 hr of observa-
tion with a typical effective area of 4×108cm2). All but one
of those nights provide 95% C.L. flux upper limits stronger
than the VHE gamma-ray flux observed during the 2014 flare
(5.8×10−11 cm−2s−1). Following the 2014 event, a duration
of an individual flare of at least 2 days was assumed. Us-
ing Monte Carlo simulations, we related the assumed rate of
flares with the corresponding probability of at least one of
them being caught in the observation slots of MAGIC. We
found that the VHE duty cycle is consistent with less than
16 flares per year at 95% C.L. with a flux >100 GeV of at
least 5.8×10−11 cm−2s−1.
3.2 Enhanced emission periods
Flux variations were detected across different energy bands
during the 4-years-long multi-instrument observations of
QSO B0218+357 (see Fig. 1and Table 2). Enhanced GeV
emission was observed by Fermi -LAT around MJD ∼57650.
The rise of the GeV emission was gradual. In our study, the
period from MJD 57600 to MJD 57700 (dubbed as F1) was
selected, which covers a time interval where the GeV flux in-
creases and decreases (see Fig 1). Based on the obtained light
curve, the resulting spectrum reported in this manuscript is
not expected to depend strongly on the exact definition of
Tag MJD description
F1 57600 – 57700 optical and GeV flare
F2 58863.7 X-ray flare
Aug 2020 59071.5 & 59069.6 Fermi-LAT >10 GeV
Table 2. List of discussed enhanced emission periods observed dur-
ing the monitoring (F1 and F2). The campaign in August 2020
was organized after the regular MWL monitoring of the source.
the start and end of this time interval, and that similar re-
sults would have been obtained by modifying this time in-
terval by a few days. During this time interval, three optical
measurements (MJD 57627.2, 57639.2 and 57640.2) yielded a
flux nearly an order of magnitude larger than that of the low
state of the source. Comparing to the 2014 flare discussed in
Ahnen et al. (2016), the GeV emission is at a similar level
(however with a softer spectrum), but the optical emission is
nearly an order of magnitude higher. Interestingly the first
two points are separated by 12 days, similar to the time de-
lay between the two lensed images of the source. The opti-
cal flux density between those two measurements returned to
the low state level. However, due to poor optical sampling
of the source, the hypothesis that the two optical flares are
indeed the two images of the same flare cannot be validated.
Two of the nights of the enhanced optical activity had simul-
taneous MAGIC data. No significant emission was observed
(significances of −0.21
σ
and −0.54
σ
). The flux upper limit
above 100 GeV is ∼3×10−11 cm−2s−1, i.e. constrained to be
at least two times below the level observed during the 2014
flare. Previously, gravitational lensing was used to predict
possible sites of gamma-ray flares detected in 2012 and 2014
(Barnacka et al. 2016). They entertained a hypothesis that if
both flares were caused by the same plasmoid created in the
vicinity of SMBH and traveling toward the radio core then
the interaction of the plasmoid with the radio core should
be observed around July 2016. While OVRO monitoring at
15 GHz do es not show a significant increase in flux density,
the predicted interaction coincides with the beginning of F1
in gamma rays, and available observations in R-band show
a significant increase in flux density during the predicted in-
teraction of the plasmoid with the radio core. This example
illustrates the complexity of studying emissions from these
sources but also points to a unique potential and the impor-
tance of long-term monitoring of QSO B0218+357 to eluci-
date the multiwavelength origin of emission.
A hint of enhanced activity was observed in the X-ray band
by XMM-Newton on MJD 58863.7 (dubbed as F2). The flux
density increased by (44±19)% with respect to the previous
XMM-Newton measurements. The contemporaneous MAGIC
observations did not yield any significant detection (excess at
the significance of 2.1
σ
). These observations were used to de-
rive 95% C.L. upper limit on the flux of ∼2.8×10−11cm−2s−1
above 100 GeV, which is two times below the VHE flux mea-
sured during the flare in 2014. No significant excess is ob-
served in the MAGIC observations in the two neighbouring
nights further suggesting that the marginal excess in MAGIC
data during the X-ray flare is a background fluctuation. No
excess of GeV flux was observed during the X-ray flare. In-
terestingly, while the optical flux density did not change con-
MNRAS 000,1–21 (2020)
8V. A. Acciari et. al.
MAGIC >100 GeV
57400 57600 57800 58000 58200 58400 58600 58800
0
0.5
1
1.5
]
-1
s
-2
cm
-10
F [10
F1
F2
MAGIC >100 GeV
Fermi-LAT flux
57400 57600 57800 58000 58200 58400 58600 58800
0
2
4
6
8
]
-1
s
-2
cm
-7
F [10
Fermi-LAT flux
Fermi-LAT index
57400 57600 57800 58000 58200 58400 58600 58800
1.5
2
2.5
3
3.5
Index
Fermi-LAT index
Swift-XRT/XMM (0.3-10keV)
57400 57600 57800 58000 58200 58400 58600 58800
0
1
2
]
-1
s
-2
erg cm
-12
F[10
Swift-XRT
XMM
Swift-XRT/XMM (0.3-10keV)
U-band
57400 57600 57800 58000 58200 58400 58600 58800
0
5
10
]
-1
s
-2
erg cm
-14
F [10
UVOT (U.L.) XMM-OM XMM-OM (U.L.)
U-band
B-band (NOT)
57400 57600 57800 58000 58200 58400 58600 58800
0
0.01
0.02
0.03
F [mJy]
B-band (NOT)
R-band (KVA, NOT)
57400 57600 57800 58000 58200 58400 58600 58800
0.01
0.1
1
F [mJy]
R-band (KVA, NOT)
KaVA
57400 57600 57800 58000 58200 58400 58600 58800
0
100
200
300
F [mJy]
A (22 GHz) B (22 GHz)
A+B (86 GHz)
KaVA
OVRO (15 GHz)
57400 57600 57800 58000 58200 58400 58600 58800
600
800
1000
F [mJy]
OVRO (15 GHz)
Figure 1. MWL light curve of QSO B0218+357 between January 2016 and January 2020. From top to bottom: MAGIC flux above 100GeV,
Fermi-LAT flux above 0.1 GeV, Fermi -LAT spectral index, X-ray flux in 0.3 – 10 keV range (corrected for Galactic absorption) measured
with Swift-XRT and XMM-Newton, U-band observations from Swift-XRT-UVOT and XMM-OM, B-band observation from NOT, optical
observations in R-band with KVA and NOT. KaVA V LBI observations at 22 GHz (filled symbols show A image, empty ones B image)
and 86 GHz (sum of A and B images shown with stars). OVRO monitoring results at 15 GHz. Flux upper limits are shown with downward
triangles. Optical data are corrected for the host/lens galaxy contribution and galactic absorption. The points in red are contemporaneous
(within 24 hr slot) with MAGIC observations. The gray filled regions mark the enhanced emission p eriods F1 and F2.
MNRAS 000,1–21 (2020)
MWL study of QSO B0218+357 9
]
2
[ deg
2
θ
0 0.1 0.2 0.3 0.4
events
N
0
2000
4000
6000
8000
10000
12000
14000
16000 Time = 72.38 h
97± = 28144
off
= 28025; N
on
N
193± = -119
ex
N
σSignificance (Li&Ma) = -0.6
]σSignificance [
3−2−1−0 1 2 3
Number of days
0
1
2
3
4
5
6
7
Figure 2. Left panel: Distribution of the squared angular distance between the nominal and reconstructed source position of the total
dataset of MAGIC data (points) and corresponding background estimation (shaded region). Right panel: distribution of significances of
individual nights of MAGIC observations, the red line shows the expected distribution for lack of significant emission, i.e. a Gaussian
distribution with a mean of 0 and standard deviation of 1.
1−
10 1 10 2
10 E [GeV]
13−
10
12−
10
11−
10
10−
10
]
-1
s
-2
dN/dE [TeV cm
2
E
MAGIC monitoring U.L.
Fermi-LAT contemporaneous
MAGIC, 2014 flare
Fermi-LAT, 2014 flare
Figure 3. Comparison of the MAGIC SED upper limits (black
empty squares) with the extrapolation (black line and shaded re-
gion) of the Fermi-LAT spectrum (black filled circles). The extrap-
olation assumes a power-law behaviour and an extragalactic back-
ground light (EBL) absorption following Dom´ınguez et al. (2011)
model. For comparison, the Fermi -LAT and MAGIC SED during
the 2014 flare (Ahnen et al. 2016) are shown with grey markers.
siderably during F2, a hint of increase in the UV flux density
by (70 ±41)% was found.
3.3 August 2020 campaign
Besides the multi-instrument observations described above,
QSO B0218+357 is one of the sources that are regularly
checked for GeV flares in the Fermi-LAT data, and addi-
tional observations are organized if flares or hints of flares are
found. On MJD 59071.502 a photon with estimated energy
of 59.4 GeV was observed from the vicinity of the source. Ad-
ditionally, quasi-simultaneous Swift-XRT observations per-
formed on MJD 59069.566 as well as TJO on MJD 59070.993
showed hints (at ∼2.5
σ
level) of enhanced flux density. Imme-
diate follow up with the MAGIC telescopes was not feasible,
due to the presence of bright moonlight, which would have
59066 59068 59070 59072 59074 59076 59078 59080 59082 59084 59086
0
2
4
]
-1
s
-2
cm
-7
F [10
Fermi-LAT F >0.1GeV
59066 59068 59070 59072 59074 59076 59078 59080 59082 59084 59086
0
1
2
3
4
]
-1
s
-2
erg cm
-12
F[10
XRT, 0.3-10 keV
59066 59068 59070 59072 59074 59076 59078 59080 59082 59084 59086
0
0.05
0.1
F [mJy]
R-band
TJO NOT HCT
59066 59068 59070 59072 59074 59076 59078 59080 59082 59084 59086
MJD
0
0.01
0.02
0.03
F [mJy]
B-band
NOT HCT
Figure 4. MWL light curve of QSO B0218+357 during the August
2020 multiwavelength campaign. Vertical lines: MAGIC observa-
tion nights (red) and Fermi-LAT >10 GeV photons(blue).
substantially increased the energy threshold of the observa-
tions. Instead, similarly to the 2014 event, a multiwavelength
campaign was organized at the expected arrival of the B im-
age of the flare, at the assumption that the observed one was
the A image.
The observations are summarized in Fig. 4. A second
HE photon with energy of 20 GeV was observed on MJD
59082.826. The association probability of each photon above
10 GeV was assessed with QSO B0218+357 using the stan-
dard Fermi-LAT tool (gtsrcprob). The whole data set was
MNRAS 000,1–21 (2020)
10 V. A. Acciari et. al.
divided in the four Point Spread Functions (PSF) classes15.
This is needed since there are relevant PSFs differences be-
tween the four classes and those differences have an impor-
tant effect on the association probability. During the August
observations, only these two photons above 10 GeV were de-
tected with an association probability higher than 80%. The
probability of the two photons being associated to the source
is 99.91% and 86.88% for the first and the second photon re-
spectively. The time difference of these two photons is 11.324
days, which is curiously consistent with the previously mea-
sured delay between the two images.
In order to evaluate statistical chance probability of occur-
rence of HE photons close in time with the emission of the
source in a broader time scale is investigated. 120 such pho-
tons spanning the total observations of the source by Fermi-
LAT (MJD=54683 – 59299) are obtained. Conservatively as-
suming the time window for the second photon of ±1day
(motivated by the spread of radio delays of ∼10 −12 days)
a chance probability of 5.2% is obtained. The time delay of
11.324 days is also within the 1
σ
uncertainty of that mea-
sured from 2012 Fermi-LAT high state (11.46 ±0.16 days,
Cheung et al. 2014). For such a narrow window the corre-
sponding chance probability is 0.83%.
In the optical range a hint of increase of the R-band emis-
sion (2.5
σ
difference to the previous point and 4.2
σ
to the
next) occurred close to the arrival time of the first Fermi-
LAT photon. The observations during the planned moni-
toring at MJD=59081 – 59085 were performed with higher
cadence with additional instruments (NOT and HCT). The
NOT measurement at MJD=59083.1is: 2.5
σ
above the pre-
vious HCT observation, 2.4
σ
above the following HCT mea-
surement. Similarly, comparing the enhanced NOT point to
the previous NOT measurement at MJD=59082.2, the dif-
ference shows a similar weak hint of 2.4
σ
. The B-band flux
density increase was not accompanied by a similar R-band
increase at the expected time of arrival of the second compo-
nent of the flare. This would favour the interpretation of the
B-band increase as a statistical fluctuation.
While some small hint of variability (2.3
σ
difference) can
be seen between the first two Swift-XRT points, no variability
can be seen during the expected time of arrival of the delayed
component. This is understandable since the delayed compo-
nent is expected to have a lower flux density at the peak,
hence any small variability present in the leading emission
can be easily missed in the trailing one.
In Fig. 5radio follow-up light curves of the August 2020
campaign obtained with OVRO at 15 GHz and KaVA at
43 GHz is shown. For KaVA data in which A and B are
spatially separated, the radio flux density for the core (the
brightest component) was measured in each of A and B im-
ages. The core of A in August 2020 is found to be significantly
brighter than the average flux density level in 2017–2018, and
subsequently it shows continuous decrease in flux density at
least until the end of our follow-up period (October 8), where
the 43 GHz core flux density reaches a lowest level since the
start of our KaVA monitoring from 2017. In contrast, we cau-
tion that the observed light curve of the weak component B is
less defined because the observing conditions of KaVA follow-
up sessions in 2020 were generally severe compared to the
15 https://fermi.gsfc.nasa.gov/ssc/data/analysis/documentation/Cicerone/Cicerone_Data/LAT_DP.html
59040 59060 59080 59100 59120 59140
MJD
10
2
10
3
10
Flux[mJy]
OVRO 15GHz
hovi 37GHzaMets
hovi 37GHz (SNR<4)aMets
KaVA 43GHz core A
KaVA 43GHz core B
Figure 5. Radio follow up light curve of the August 2020 campaign
of QSO B0218+357 with OVRO (black points) and KaVA (red
filled and blue empty circles for core image A and B respectively).
Black squares show the significant 37 GHz Mets¨
ahovi flux densities
(empty squares shows the times of observations that resulted in
signal-to-noise ratio below 4). The dotted lines show the average
flux density from the monitoring period between January 2017 and
December 2018. The vertical blue lines show the times of arrival of
the two HE Fermi -LAT photons during the August 2020 campaign.
regular sessions in 2017-2018 (shorter integration time and
smaller number of stations). In Fig. 5there may be a signifi-
cant amount of missing flux density in the light curve for B.
This prevents us from cross-correlating the light curves of A
and B. The 15 GHz OVRO monitoring did not cover the pe-
riod in which the 43 GHz flux density decreased. The OVRO
data 3 weeks before and 2 weeks after the KaVA flux den-
sity minimum show consistent flux densities. Mets¨
ahovi data
cover the period in which the flux density measured by KaVA
starts to decay, however higher uncertainties and a larger in-
tegration region prevents sensitive probing of variability.
3.4 Radio jet image
In Fig. 6the evolution of the radio images of the source be-
tween January 2017 and December 2018 is presented. Here
KaVA 43 GHz images are shown since the highest angular
resolution is available. In addition to the core, a strong jet
component is clearly detected (at ∼1.5/1.3 mas from the core
in A/B respectively, which corresponds to the projected dis-
tance between the two components of the order of 10 pc) in all
the images. No additional knots have been observed through-
out the observations. The jet direction is different in Image
A and Image B, however this is a geometrical effect caused
by the lensing.
In order to examine the kinematics of the jet component,
the core and jet in the KaVA 43 GHz images were fitted by
two 2D elliptical Gaussians, and the core-jet separation was
measured as function of time. As can be seen in Fig. 6, the
core-jet angular separations for both A and B images are
quite stable over ∼2 years of our observing period. Assuming
that the same jet component is traced over 2 years, a simple
linear fit to the data is performed. Best-fit values of 0.06 ±
0.03 mas/yr and 0.04±0.04 mas/yr are obtained for Jet-A and
Jet-B, respectively (corresponding to 3.1±1.5cand 2±2c).
In addition to the bright core and jet, the KaVA images
MNRAS 000,1–21 (2020)
MWL study of QSO B0218+357 11
Figure 6. Selected KaVA images at 43 GHz spanning 2 year p eriod. Top panel and bottom panels are images A and B of the source. For all
images, contours start from −1(dotted lines), 1, 2, ... times 1.8 mJy beam−1(approximately 3
σ
) and increase by factors of 21/2. Blue and
red dots represent the average (over all the images) position of the core and jet components. Two cyan points in image A represent the
average positions of the sideway components, which were obtained based on 5 epochs where the sideway extension was clearly detected. For
all the points, the position uncertainties were estimated based on the scatter from multiple epochs. Gray circle represents the smoothing
kernel of the image.
of A also reveal diffuse extension in the direction perpendic-
ular to the jet (see also Biggs et al. 2003;Hada et al. 2020).
Two Gaussian models were additionally fitted to each KaVA
43 GHz image of A to characterize the positions of these ex-
tended structures. Since the sideways structures are generally
diffuse and weak, reasonable fitting results were obtained on
these features only for 5 epochs (MJD=58069, 58123, 58434,
58450, 58476; 2017 November 12, 2018 January 5, November
12, November 28 and December 24) where the image quality
was relatively high and SNR∼4–10 were obtained for the fit-
ted sideways components. In Fig. 6, the positions of the these
additional features (averaged over the 5 epochs) are plotted
in cyan. With respect to the core, the apparent ‘opening an-
gle’ of this perpendicular extension (the angle between two
vectors from the red point to each cyan point in Fig. 6) is
estimated to be ∼62◦(if only the extension of the main jet is
considered, the opening angle is roughly a half of this). Pos-
sible proper motions in these components were also searched
for, but both of the components are essentially stationary,
similar to the main jet component.
Regarding the mas-scale jet morphology during the August
2020 campaign, higher noise levels in the KaVA images (due
to shorter integration time, smaller number of stations, higher
humidity in the summer season) than in the 2017–2018 ses-
sions made our image analysis more challenging (especially
for the image B). Nevertheless, the overall radio morphology
was quite similar to that in 2017–2018 (Fig. 6). While the
core of A in August 2020 was significantly brighter than the
average flux density level in 2017–2018 (see Fig. 5), no clear
ejection of new components from the core during our KaVA
follow-up period was found.
3.5 Lens geometry model
The observations of QSO B0218+357 with the Hubble
Space Telescope (HST) show that the lensing galaxy is
isolated pointing to a simple gravitational lens poten-
tial. The mass distribution of the lens has been shown
to be well represented by a Singular Isothermal Sphere
(SIS) model (Wucknitz et al. 2004;Larchenkova et al. 2011;
York et al. 2005;Barnacka et al. 2016). The SIS model of
QSO B0218+357 predicts a time delay of ∼10 days and
a magnification ratio of ∼3.6. The predicted magnification
ratio fits the observed ratio between radio images. How-
ever, the time delay is ∼1 day shorter as compared to the
time delay measured at gamma rays (Cheung et al. 2014;
Barnacka et al. 2016).
Hada et al. (2020) used a Singular Elliptical Power-law
(SEP) model and parameters fixed to the values obtained
by Wucknitz et al. (2004). The SEP model provides a higher
rate of change in the lens potential. As a result, the model
with the same image positions predicts longer time delays
and a larger magnification ratio between the images. The
SEP model adapted by Hada et al. (2020) predicts a time
delay of 11.6 days for the core, which matches better the
observed time delay at gamma rays. However, the predicted
magnification ratio is ∼4.8, which significantly deviates from
a reported average value of 3.5 from a broad range of radio
observations (Patnaik, Porcas, & Browne 1995).
At first parameters of the lens model presented in
Barnacka et al. (2016) are adopted, which reconstructed the
positions of the lensed images with an accuracy of 1mas.
The previous model was based on 15 GHz radio observations
Patnaik, Porcas, & Browne (1995). Here, the reconstruction
of the lens model is further improved by using high resolution
KaVA observations of the radio core and jet listed in Table 3.
MNRAS 000,1–21 (2020)
12 V. A. Acciari et. al.
Component Image RA DEC ∆MODE L Source Time Delay
µ
Ratio
[mas] [mas] [mas] [mas] [days]
Core A 0 ±0 0 ±0 0.067 (90.0,37.1) 10.36 2.72 3.81
B 309.144±0.015 127.450±0.029 0 -0.71
Jet A 0.681±0.031 1.331±0.045 0.036 (89.06,36.21) 10.30 2.74 3.67
B 310.444±0.014 127.253±0.038 0.260 -0.75
Table 3. Positions of radio images observed by KaVA and the lens model predictions. The positions of the images are referenced to the
lensed image A (0,0). Position errors are estimated based on the scatter of fitted positions based on the assumption that all of the
components are stationary. The positions (RA,DEC) are shown for the observed lensed images A and B for the core and jet components.
The image A was modeled by 4 Gaussians (core, main jet, left wing, right wing). The lensed image B was modeled by 2 Gaussians (core,
jet). Table reports averaged positions over 5 epochs (2017Nov12, 2018Jan05, 2018Nov12, 2018Nov28, 2018Dec24). The ∆MODEL represents
a difference between the predicted and observed positions of the lensed images, as well as reconstructed positions of the core and jet in
the source plane in respect to the lens center. Table also shows time delays, magnifications, and magnification rations predicted using the
best SIS lens model.
Figure 7. Compilation of observations and lens model predictions.
The color map shows the Fermat potential of the lens model for
the best reconstructed position of the radio core. The images form
at the extreme of the Fermat surface (blue). The gray contours
show the lensed images of the core and the Einstein ring observed
at 1.687 GHz Wucknitz et al. (2004). The image is centered at the
reconstructed position of the lens indicated as the blue asterisk.
The red open circles correspond to the reconstructed position of
the images of the jet(A on the right, and B on the left). The red
dashed-dotted line connects the images. For the SIS lens model, the
source (green asterisk) is located at half distance between images.
The observed and reconstructed images of the radio core are also
shown, however they are superimposed with the red open circles.
A softened power-law potential (Keeton 2001;
Barnacka et al. 2016) is used, which includes an Ein-
stein radius, a scale radius of a flat core ( sin mas), a
projected axis ratio (q), and a power-law exponent (
α
).
Similarly, like in Barnacka et al. (2016), the best lens model
of softened power-law potential resulted in s∼0,q∼1, and
α
∼1. Thus, the model reduced to the SIS potential.
Figure 7shows the radio observations with the prediction
of the SIS model including the Fermat potential, predicted
positions of the lensed images, as well as reconstructed po-
sition of the the jet and core. The image is centered at the
position of the lens located at (244.55,100.85) with respect to
image A.
Only positions of the radio images were used to find the
Relative RA. (mas) -2-10123
Relative Dec. (mas)
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
Relative RA. (mas) -2-10123
Relative Dec. (mas)
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
Figure 8. Comparison of the KaVA 43 GHz image taken on
MJD=54470 (2018 January 5), A in left panel, B in right panel
compared with the reconstructed positions of the lens model im-
ages (stars)
best fit, as the observed time delays and magnification ratios
are a subject of discussion. Figure 8shows the position of re-
constructed images as red and green open circles. The model
reconstructs the positions of the lensed images with average
accuracy of 0.03mas for the radio core and 0.15 mas for the
jet.
The projected distance between reconstructed positions of
the core and jet is 1.2±0.1mas, which correspond to 9.8pc
in the source plane, consistent with the result obtained by
Hada et al. (2020). The SIS model predicts a time delay of
10.36 days for the core and 10.30 days for the jet. The shorter
time delay for the jet indicates that the radio jet is positioned
toward the lens center as argued by Barnacka et al. (2016)
based on observations of the Einstein radius at 1.687 GHz
Wucknitz et al. (2004). The Einstein ring forms from emis-
sion of the kpc-scale jet aligned with the lens center. The
Einstein ring observed at 1.687 GHz (Wucknitz et al. 2004)
is also added to Figure 7.
The predicted time delays reported here are calculated us-
ing H0=67.3±1.2km s−1Mpc−1(Planck Collaboration et al.
2014). Note that the time delay is inversely proportional to
H0. Thus, larger values of H0=73.3+1.7
−1.8kms−1Mpc−1as re-
ported by Wong et al. (2020) would result in an even shorter
time delay, and as such would further increase the discrep-
ancy between the lens models and observations. The Hub-
ble parameter estimation based on gravitationally-induced
time delays of variable sources with relativistic jets is in
particular prone to biases as the emission can originate
MNRAS 000,1–21 (2020)
MWL study of QSO B0218+357 13
from multiple sites, which can introduce systematical errors
(Barnacka et al. 2015).
Both the SEP (Hada et al. 2020) and improved SIS recon-
struction based on KaVA observations, provide accurate re-
construction of the positions of the lensed images and consis-
tent distance separation of the radio core and jet components.
However, the predictions of these two models differ in terms
of time delay and magnification ratios. The SIS scenario pre-
dicts magnification ratios of 3.81 and 3.67 for the core and
jet, respectively. The flux densities of the lensed images at 43
GHz reported in Table 3 (Hada et al. 2020) indicate magnifi-
cation ratios of 3.7 and 3.2 for the core and jet, respectively.
For comparison, the SEP model predicts magnification ratios
of 5 and 4.84 for the core and jet, respectively.
In principle, magnification ratio could be affected by mi-
crolensing or “substructure lensing” such as compact clumps
in the lens galaxy (Sitarek & Bednarek 2016). To match
the prediction of the magnification ratio of ∼5of the SEP
model with the observed ∼3.5, either the brighter image A
would have to be magnified by a factor of 1.35, or image
B would have to be de-magnified by the same factor. More-
over, the lensed images of both jet and core would have to
be magnified/de-magnified simultaneously by a similar fac-
tor. As a result, the diameter of the perturbing mass needs
to be ≫10 pc, precluding microlensing. In principle, possible
clumps of tens of pc of a giant molecular cloud (GMC) (see
e.g. Chevance et al. 2020) could result in additional moder-
ate amplifications by a factor of 1.5 (see Sitarek & Bednarek
2016).
Interestingly, the observed magnification ratio fits the pre-
diction of the SIS model well. The SIS model predicts cor-
rectly not only the values but also the degree to which the
magnification ratio of the core is greater than the magnifi-
cation ratio of the jet. However, the time delay of 10.3 days
predicted by the SIS model is one day shorter that the time
delay measured at gamma rays. Such a discrepancy in the
time delay could be explained if there is an offset of ∼50 pc
between sites of radio and gamma-ray emission (for a review
see Barnacka 2018).
The degeneracy between the SIS and SEP lens models
could be broken by precise measurement of the radio time
delay. The time delay obtained in the SIS model depends
only on H0and the image angular separation. Thus, the SIS
model can be ruled out if the radio time delay is not 10.3
days.
However, measuring time delay at radio with an accuracy
of hours is difficult as radio observation of B2 0218+35 shows
almost no variability, in addition to gaps between the observa-
tions. Further high resolution KaVA observations combined
with detailed modeling of the lens could elucidate the true
potential of the lens and test the clump scenario. KaVA ob-
servations provide well-resolved images of both the jet and
core, thus providing multiple tests of the lens potential. One
of the predictions of the SIS model is that the diameter of the
Einstein ring is equal to the distance between the lensed im-
ages of the source. The current KaVA observations show that
the distance between the lensed images of both the core and
jet are equally within the range of uncertainty of the observa-
tions - as such, consistent with the SIS model. More precise
observations, or detailed predictions of the SEP model on the
expected difference between the lensed images of the core and
jet could help exclude the SEP model. Moreover, elliptical
lens models should result in formation of the odd number of
images (Gottlieb 1994;Zhang et al. 2007;Petters & Werner
2010). Thus, detection of the third image in the vicinity of
the predicted lens center could provide further constraints on
the model of the lens.
Here, we focused on the two most general lens models,
namely SIS and SEP, and on the possibility to break the
degeneracy between them. However, a potentially more gen-
eral class of models might be required to reconstruct both the
time delay and the magnification ratio.
The in-depth observations of the object B2 0218+35, com-
bined with a precise model of the lens, have the potential to
provide unique insights on the origin and site of the gamma-
ray emission, the Hubble constant, or substructures in the
lensing galaxy.
3.6 Modelling of dust in the lens
A Galactic absorption of NH=5.56×1020 cm−2was
adopted from the Leiden/Argentine/Bonn (LAB) survey
(Kalberla et al. 2005). The X-ray flux density, corrected
for the Galaxy absorption, in the (0.3–10) keV band is
f=(1.53±0.11)×10−12 erg cm−2s−1for the low flux density
state, and f=(2.25±0.24)×10−12 erg cm−2s−1for the high-
flux density state from XMM-Newton observations.
In order to evaluate and correct the effect of additional
absorption in the host or lens galaxies an approach similar
to Ahnen et al. (2016) is applied. The higher sensitivity of
XMM-Newton compared to the Swift -XRT telescope allows
us to study a few alternative models of absorption and in-
trinsic source spectrum and select between them. For the in-
vestigations of the low state three cases are considered: (a)
no additional absorption, (b) absorption at the host, (c) ab-
sorption at the lens. In the case (b) the absorption will affect
the total emission observed from the source (in both images).
In the case (c), since the two images cross different parts of
the lens galaxy, the absorption would be different for them.
There are reasons to believe that in such a situation the ab-
sorption would mainly affect the brighter, A, image (see the
discussion in Ahnen et al. 2016). Therefore in the case (c) the
observed emission is assumed to be comp osed of two“virtual”
sources located at the nominal location of QSO B0218+357
in the sky. The first component is affected only by Galactic
absorption, and the second one is additionally absorbed by
a hydrogen column density (NH,z) at the redshift of the lens
(z=0.68).
Two spectral models are investigated: a simple power
law and a log parabola (defined as F(E) = k(E/E0)−Γand
F(E) = k(E/E0)−(
α
+
β
log(E/E0)), where E0= 1 keV in all the
models). For the absorption, the Tuebingen-Boulder ISM ab-
sorption model (Wilms et al. 2000) available in XSPEC was
adopted. The column density NH,zand the parameters of the
log-parabola or power-law components were fitted by impos-
ing that the values of
α
,
β
, or Γare the same for the two
components, and that the normalization of the component
with only Galactic absorption is a factor 0.71/2.72 =0.261
(corresponding to magnification ratio, see Section 3.5) lower
than the normalization of the component with also internal
absorption.
The results of the fits are summarized in Table 4. The sim-
ple log-parabola model, without an additional absorption is
not sufficient to explain the XMM-Newton data (
χ
2/Ndo f =
MNRAS 000,1–21 (2020)
14 V. A. Acciari et. al.
Table 4. Best-fit parameters of the XMM-Newton and Swift -XRT analysis.
obs. ID Exp. time Model
α β
Γk1k2zNH,z
χ
2/Ndo f
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
low state 27.6 LP 1.96±0.05 0.17±0.05 – 0.71 2.73±0.11 0.68 8.10±0.93 375.0/373
low state 27.6 PL – – 2.06±0.03 0.72 2.78±0.10 0.68 8.83±0.82 386.6/374
low state 27.6 LP 1.37±0.08 0.72±0.10 – 2.43±0.09 – 0.94 1.88±0.49 398.9/373
low state 27.6 LP 1.05±0.03 1.09±0.05 – 2.14±0.03 – – – 413.4/374
low state 27.6 PL – – 1.95±0.03 2.98±0.07 – 0.94 5.16±0.27 442.8/374
0850400601 10.3 LP 1.58±0.16 0.53±0.15 – 0.96 3.68±0.36 0.68 5.4±1.8 90.5/94
0850400601 10.3 LP+low 1.89±0.40* 0.17±0.42* – – 2.14±0.42* 0.68 7.0±2.3* 89.7/94
Swift-XRT (2016-2017) 9.4 PL – – 1.62±0.13 0.47 1.79±0.18 0.68 8.10 7.7/9
Swift-XRT (2020) 20.4 PL – – 1.83±0.06 0.80 3.07±0.15 0.68 8.10 33.2/34
Notes. Columns: (1) observation identifier (“low state” corresponds to combined 0850400301, 0850400401 and 0850400501 epochs of
XMM-Newton observations); (2) exposure time filtered for good time intervals in ks; (3) log-parabola (LP)/power-law (PL) spectral
model; (4) and (5) LP spectral index and curvature; (6) PL spectral index; (7) and (8) normalization of the LP/PL model at 1 keV
in units of 10−4cm−2s−1keV−1of B and A image respectively; (9) redshift of the absorber; (10) column density of the absorber in
units of 1021 cm−2; (11) reduced
χ
2Final selected model for each dataset is marked with bold face. * - only the additional flaring
component
413.4/374). Including an additional absorption at the lens for
the brighter image the
χ
2improves by 38.4at the cost of
one additional degree of freedom (hydrogen column density
at the lensed image A). The model is compared with observed
XMM-Newton rates in Fig. 9.
Since not all the investigated models are nested, to compare
them Akaike Information Criterion (Akaike 1974) is used. For
the case of
χ
2statistics the relative difference of AIC param-
eter of two models can be computed as: ∆AIC =2∆np+∆
χ
2.
The relative likelihood of the models can be computed (see
e.g. Burnham et al. 2011) as: p=exp(∆AIC/2). Including the
absorption at z=0.68 the intrinsic curvature of the emission
is preferred. The difference of
χ
2values is 11.6for one ad-
ditional parameter (describing the curvature) corresponds to
p=0.008 relative likelihood of the power law model to the log
parabola model. On the other hand, comparing models with
absorption at z=0.68 and z=0.94, with the same number of
free parameters in both models, the former has a lower
χ
2
value by 23.9. Therefore the model with absorption at the
source is only 1.8×10−5as likely as the model with the ab-
sorption at the lens. Summarizing, for the low state of the
source, the preferred model of the emission involves an in-
trinsic log-parabola spectrum and absorption by a column
density of (8.10 ±0.93stat)×1021 cm−2at the lens.
Comparing to the result obtained in Ahnen et al. (2016)
24 ±5stat ×1021cm−2, the absorption obtained in this work
is more precise, but also lower, and both values are consis-
tent in the broad range derived by Menten & Reid (1996)
(5−50 ×1021cm−2). The actual absorption in the lens might
have changed if the region emitting X-rays has moved along
the jet, or changed its size compared to the observations in
2014. However it is equally likely that additional systematics
(assumption of the intrinsic spectral model, flux density and
spectral variability, absorption in the other image of the lens
and in the host galaxy) affected one or the other measure-
ment.
The proper correction for the absorption and lensing dur-
ing the MJD 58863.7 high X-ray state is more uncertain.
Since the observations are separated by 140 days from the
previous X-ray flux measurement, is not clear if the X-ray
Figure 9. Differential energy flux of QSO B0218+357 folded with
the response of XMM-Newton observed from low-state pointings
(top panel) and from the X-ray flare (bottom panel). Points show
the observed rate, while lines show the LP model for image A
(blue dashed), image B (green dot-dashed) and total emission (red
solid).
MNRAS 000,1–21 (2020)
MWL study of QSO B0218+357 15
high state was a short duration flare, or a longer time scale
high state. In a case of a short flare the observations might
have happened when the corresponding image A or image B
reached the observer, resulting in a different absorption. On
the other hand, if the enhanced state was significantly longer
than the ∼11 days delay between the two images the observed
emission should be the average from both images, similar to
the case of the low state. This is further supported by the
fact that the data collected by Swift -XRT for MJD 59069
– 59108 show also higher X-ray fluxes, similar to the last
XMM-Newton measurement. To analyze MJD 58863.7 data
of XMM-Newton the assumption that the observed increase
in the X-ray emission was over a longer time scale is applied,
therefore the log-parabola spectral model was used with ab-
sorption of the brighter image and fixed flux ratio between
both components. Such a model describes sufficiently well
the observations (
χ
2/Ndo f =90.5/94), and provides a some-
what harder and more curved X-ray spectrum than during
the low state. The derived NH,zis roughly consistent (at 1.3
σ
level) with the values obtained from the low state fit. An
alternative model has been tested in which the high state
emission is a sum of the low state emission, with the spec-
tral shape and flux fixed to the low-state-fitted values and an
additional flaring component with an absorption at z=0.68
(i.e. the flare originating from the brighter image A). The fit
results for such an additional flaring component are reported
in ”LP+low” row of Table 4. The two investigated models for
the flaring state are not nested, however they have the same
number of free parameters and result in nearly the same
χ
2,
therefore neither can be rejected. For the sake of simplicity
the same absorption model is used for the flaring state as for
the low state.
As discussed in Section 2.4, two combined spectra are used
for the spectral study using Swift-XRT data during 2016-
2017 and August 2020. Due to the restriction aroused by
Swift-XRT sensitivity, only two spectral models within the
scenario of case (c) are investigated. In addition to the as-
sumption implemented for XMM-Newton observations, NH,z
is also assumed to be equal to 8.10 ×1021 cm−2. The log-
parabola model cannot describe the observed spectrum bet-
ter than the power law model. The results are presented in
Table 4.
4 LOW STATE BROADBAND SED MODELLING
The multiwavelength behaviour of the source in different
states is summarized in Fig. 10. During the low state the
GeV emission was significantly lower than during the 2014
flare described in Ahnen et al. (2016), and even during the
F1 enhanced state the GeV spectrum was softer than in 2014.
Except for the strong optical flares in F1, the rest of the MWL
SED does not vary strongly with respect to the 2014 flare
and historical measurements. The difference in reported ra-
dio measurements to the historical ones is most likely caused
by much smaller integration region (only the inner jet of the
source) achieved in radio interferometry with KaVA.
For the average state a detailed modelling is performed.
The emission is modeled in the framework of an External
Compton (EC) model, which is a common scenario for FS-
RQs. There is growing evidence that the main target for FS-
RQs EC process is the reprocessed Dust Torus (DT, see e.g.
9
10 11
10 13
10 15
10 17
10 19
10 21
10 23
10 25
10 27
10
[Hz]ν
15−
10
14−
10
13−
10
12−
10
11−
10
10−
10
]
-1
s
-2
dN/dE [erg cm
2
E
2014 flare
2016-2020 low/average state
opt. & GeV flare (F1)
X-ray flare (F2)
Figure 10. Multiwavelength SED of QSO B0218+357 in different
periods: average state of the monitoring (excluding optical flare
data) in red (for clarity Swift-XRT and MAGIC data are shown
with empty symbols, while XMM-Newton and Fermi-LAT and
the rest of MWL data are shown with full symbols), X-ray flare
in cyan, and optical/GeV flare in green. For the case of optical
flare and minimum and maximum optical flux density during in-
vestigated period are plotted. For comparison historical data (ob-
tained from SSDC service) are plotted in gray, while the 2014 flare
(Ahnen et al. 2016) is in black.
Costamante et al. 2018;van den Berg et al. 2019). Moreover,
even while no VHE gamma rays has been detected from the
QSO B0218+357 during the monitoring period, the source
is a known emitter in this band (Ahnen et al. 2016), again
supporting EC-DT scenario.
The recent measurement of the accretion disk luminosity of
Ld=4.3×1044erg s−1(Paliya et al. 2021) is used. The value is
close to the Ld=6×1044 ergs−1estimated by Ghisellini et al.
(2010) and applied in Ahnen et al. (2016). Using the updated
value of Ldthe sizes of the BLR and DT are computed using
the scaling laws of Ghisellini & Tavecchio (2009), resulting in
RBLR =6.6×1016 cm, and RDT =1.6×1018 cm. The tempera-
ture of the DT is set to 1000 K and its luminosity to 0.6Ld. A
conical jet geometry is used, with half-opening angle of 1/Γ,
where Γis the Lorentz factor of the jet.
For the modeling the Doppler factor of the jet D=Γ=15
is assumed. The electron energy distribution (EED) is as-
sumed to follow a power law with an index of p1up to
γ
b,
where
γ
bis the Lorentz factor of the electrons for which the
time scale for the dominating energy loss process is equal to
the dynamic scale (see Acciari et al. 2020 for details). Above
such a cooling break the EED steepens by 1 up to
γ
max, which
is determined from balancing the acceleration gain with the
dominating energy loss process. The radiation processes are
calculated using the agnpy16 code (Nigro et al. 2020), which
implements the synchrotron and Compton processes follow-
ing the prescriptions described in Dermer & Menon (2009);
Finke (2016). While the
γ
band
γ
max are calculated assuming
Thompson regime of the inverse Compton scattering, the ac-
tual spectra are computed using the full Klein-Nishina cross-
section formula.
The emission region (hereafter “Close” region) responsible
16 https://github.com/cosimoNigro/agnpy/
MNRAS 000,1–21 (2020)
16 V. A. Acciari et. al.
for the high energy bump is assumed to be located at the dis-
tance of d1=2×1017 cm, i.e. a factor of a few more distant
than the size of the BLR, but deep in the DT radiation field.
The model is confronted with the observations in Fig. 11, tak-
ing into account the magnification induced by the lensing (us-
ing the strong lensing magnifications derived in Section 3.5),
and the absorption of emission from one of the images in the
lens. The possible effect of microlensing is not corrected for,
however we expect it to have a minor influence on the long-
term average spectrum. The free and derived parameters are
summarized in Table 5.
The gamma-ray emission is explained as EC process on
DT photons (which is also the dominating energy loss pro-
cess of the electrons). On the other hand, according to the
model, the X-ray emission is mostly caused by SSC process.
The synchrotron emission corresponding to the “Close”region
can (largely) explain the optical and the rapidly falling UV
emission.
However, the region is too compact for explaining the
low-frequency radio emission which is heavily absorbed in
the “Close” region by synchrotron-self-absorption. Such low-
energy emission is expected to originate from a larger scale
jet. A commonly applied solution is the assumption of two
emission regions (see e.g. MAGIC Collaboration et al. 2020
and references therein). Therefore, motivated also by the ra-
dio knot observed by KaVA, a second region (hereafter “Far”)
is added, located at the distance of 100 p c. The distance of
the emission region is motivated by (deprojected) distance of
the jet component in the KaVA image. The low-energy slope
in this region is set to 2, and equipartition (i.e. ue=B2/(8
π
))
is applied. Then the magnetic field strength and the acceler-
ation coefficient are fixed to the values explaining the broad-
band synchrotron emission. The two emission regions are as-
sumed not to be co-spatial (“Far” region is more distant in
the jet) and thus contrary to e.g. MAGIC Collaboration et al.
(2020) are not interacting with each other. The “Far” re-
gion is distant and large-enough such that the dominating
energy loss process is the synchrotron cooling, which, again
due to size of the region and low values of the acceleration
efficiency, does not introduce a cooling break up to the maxi-
mum reached energies. Inverse Compton emission of the “Far”
region is negligible in comparison to the “Close” region (the
region is beyond the DT radiation field for EC process to play
any role, and the energy density of the electrons is too low
for SSC to be effective).
Combination of both emission regions can describe re-
markably well the whole broadband emission of the source.
In fact, the previous modelling of the source, presented in
Ahnen et al. (2016), also suggested two-zone model for this
source. That modeling was however used to explain the flaring
episode of the source, and neglected the radio and microwave
emission from the large-scale jet. It is therefore likely that
the three regions contribute to the time-variable, broad-band
emission of the source: large scale jet responsible for the radio
and microwave emission, emission region within DT respon-
sible for the broadband, high-energy, low state emission of
the source and a third region (or a sub-region of the second
one) in which VHE and HE gamma ray flares occur. As the
low state modelling attributes most of the radio emission to
the “Far” region, it is curious to observe radio variability in
August 2020 campaign KaVA data over time scales of tens
of days (see Fig. 4). Such variability might not be connected
with the source itself, but rather with the absorption and
milli-lensing effects of large scale structures in the lensing
galaxy. In Appendix Ba possible scenario is discussed for
explaining the MWL flares seen from the source during the
monitoring.
5 CONCLUSIONS
Broadband (radio, optical-UV, X-ray, gamma ray) monitor-
ing of QSO B0218+357 has been performed. The monitoring
was aimed at the detection of a VHE gamma-ray flare of
the source in time periods selected to allow additional fol-
low up at the expected time of arrival of the second image.
The deep exposure of 72 hrs of data did not reveal low-state
VHE gamma-ray emission and constrained it to be less than
about of an order magnitude below the level observed dur-
ing 2014 flare. VLBI radio images obtained with KaVA show
clear core-jet structure in both lensed images. No significant
movement of the VLBI radio features was seen. No signif-
icant variability has been seen in KaVA images during the
2016-2019 monitoring, however the follow up of August 2020
campaign showed a clear decay of core flux density in image
A. The radio data have been used to improve the lens mod-
elling to evaluate image magnifications and time delays for
the core and jet component of the source. Precise measure-
ments of the X-ray spectrum with XMM-Newton instrument
was used to evaluate the absorption in the lens, and fit the
hydrogen column density of the lens in the image A to a value
of (8.10 ±0.93)×1021cm−2.
The low-state broadband emission of the source can be
described with a two-zone model, in which the electron en-
ergy distribution shape is determined self-consistently from
the cooling, acceleration and dynamical time scales. Most of
the radio and FIR emission is explained to originate from
a large region with size/location motivated by the radio jet
component. UV (and partially optical) data are explained as
the synchrotron emission of the smaller region, that is also
responsible for the gamma-ray emission (produced in EC sce-
nario) and X-ray emission (generated via SSC process).
No short VHE gamma-ray flares have been observed in
the night-by-night analysis. Comparison with the Fermi-LAT
state of the source shows that it is unlikely that the source
has reached a comparable flare to the one of 2014 during the
monitoring. The MAGIC data were used to place a 95% C.L.
limit on the VHE gamma-ray duty cycle of the source: below
16 flares per year.
Monitoring data have revealed however a few flares/hints
of enhanced states in optical, X-ray and gamma-rays, during
which no VHE gamma-ray emission was detected. While the
limited MWL data and variability during enhanced periods
do not allow us to properly model the enhanced states, a
plausible scenario explaining qualitatively the change of be-
haviour of the source during those states by change of the
basic parameters of the model is presented.
Additional MWL campaign triggered by hints of enhanced
emission in gamma-rays, X-rays and optical has been also dis-
cussed. Unfortunately lack of MAGIC data on the predicted
night of the flare prevents us from drawing a firm conclu-
sion on possible hardening of the electron energy distribution
during the campaign.
While the primary goal of the MWL monitoring of the
MNRAS 000,1–21 (2020)
MWL study of QSO B0218+357 17
1081011 1014 1017 1020 1023 1026
ν
/
Hz
10−15
10−14
10−13
10−12
10−11
10−10
10−9
νFν
/(erg cm−2 s−1)
Synchr.
(Close)
SSC (Close)
EC DT (Close)
EC BLR (Close)
Synchr. (Far)
Total
SSDC
2014 flare
2016-2020
Figure 11. Multiwavelength SED of QSO B0218+357 composed of contemporaneous data to the MAGIC observations (red points), historical
data from SSDC (gray) and data from the 2014 flare (Ahnen et al. 2016, in black) compared with the broad-band model derived from a
two-zone SSC+EC scenario (with parameters reported in Table 5). Optical, UV and X-ray data are corrected for the Galactic absorption,
optical data are in addition corrected for the host/lens galaxy contribution. The lensing magnification, absorption in the lens galaxy and
EBL attenuation are corrected for in the model curves. For the closer region, dotted curve is the synchrotron emission, dashed the SSC,
dot-dashed EC on DT, dot-dot-dashed EC on BLR. For the farther region, long-dotted is the synchrotron emission. The total emission is
shown with an orange line.
Region
δ
d[cm]
ξ
B[G] ue[ergcm−3]p1
γ
min |p2
γ
break
γ
max rb[cm]
Close 15 2×1017 5×10−70.11 0.7 2.450 |3.41500 26 000 1.3×1016
Far 15 3×1020 6×10−10 3.2×10−34×10−72.42|- - 51 000 2×1019
Table 5. Parameters used for the modelling: Doppler factor
δ
, distance of the emission region d, acceleration efficiency
ξ
, magnetic field B,
electron energy density ue, EED: slope before the break: p1, minimum Lorentz factor
γ
min, slope after the break p2, the Lorentz factor of the
break
γ
break, maximum Lorentz factor
γ
max, co-moving size of the emission region rb. Free parameters of the model and derived parameters
are put on the left and right side of the vertical line respectively. For the case of“Far” region Band ueare tied with equipartition condition.
source has not been achieved due to in general low gamma-
ray activity of the source in the last years, the campaign
resulted in multiple interesting results, and observations of a
few interesting events. Since the achieved constraints on the
low-state VHE gamma-ray emission approach the extrapo-
lation of the GeV emission, it is expected that the future
Cherenkov Telescope Array (Acharya et al. 2013) will allow
us to study it in detail.
ACKNOWLEDGEMENTS
We would like to thank the Instituto de Astrof´ısica de
Canarias for the excellent working conditions at the Ob-
servatorio del Roque de los Muchachos in La Palma.
The financial support of the German BMBF, MPG and
HGF; the Italian INFN and INAF; the Swiss National
Fund SNF; the ERDF under the Spanish Ministerio
de Ciencia e Innovaci´on (MICINN) (FPA2017-87859-P,
FPA2017-85668-P, FPA2017-82729-C6-5-R, FPA2017-90566-
REDC, PID2019-104114RB-C31, PID2019-104114RB-C32,
PID2019-105510GB-C31,PID2019-107847RB-C41, PID2019-
107847RB-C42, PID2019-107988GB-C22); the Indian De-
partment of Atomic Energy; the Japanese ICRR, the Uni-
versity of Tokyo, JSPS, and MEXT; the Bulgarian Ministry
of Education and Science, National RI Roadmap Project
DO1-268/16.12.2019 and the Academy of Finland grant
nr. 320045 is gratefully acknowledged. This work was also
supported by the Spanish Centro de Excelencia “Severo
Ochoa” SEV-2016-0588 and CEX2019-000920-S, and “Mar´ıa
de Maeztu” CEX2019-000918-M, the Unidad de Excelencia
“Mar´ıa de Maeztu” MDM-2015-0509-18-2 and the “la Caixa”
Foundation (fellowship LCF/BQ/PI18/11630012) and by the
CERCA program of the Generalitat de Catalunya; by the
Croatian Science Foundation (HrZZ) Project IP-2016-06-
9782 and the University of Rijeka Project uniri-prirod-18-
48; by the DFG Collaborative Research Centers SFB823/C4
and SFB876/C3; the Polish National Research Centre grant
UMO-2016/22/M/ST9/00382; and by the Brazilian MCTIC,
MNRAS 000,1–21 (2020)
18 V. A. Acciari et. al.
CNPq and FAPERJ. The Fermi LAT Collaboration acknowl-
edges generous ongoing support from a number of agencies
and institutes that have supported both the development and
the operation of the LAT as well as scientific data analysis.
These include the National Aeronautics and Space Adminis-
tration and the Department of Energy in the United States,
the Commissariat `a l’Energie Atomique and the Centre Na-
tional de la Recherche Scientifique / Institut National de
Physique Nucl´eaire et de Physique des Particules in France,
the Agenzia Spaziale Italiana and the Istituto Nazionale di
Fisica Nucleare in Italy, the Ministry of Education, Cul-
ture, Sports, Science and Technology (MEXT), High En-
ergy Accelerator Research Organization (KEK) and Japan
Aerospace Exploration Agency (JAXA) in Japan, and the
K. A. Wallenberg Foundation, the Swedish Research Coun-
cil and the Swedish National Space Board in Sweden. Ad-
ditional support for science analysis during the operations
phase is gratefully acknowledged from the Istituto Nazionale
di Astrofisica in Italy and the Centre National d’´
Etudes Spa-
tiales in France. This work performed in part under DOE
Contract DE-AC02-76SF00515. We thank Director of Indian
Institute of Astrophysics for allotting us observing time with
HCT under DDT. We also thank the staff of IAO, Hanle
and CREST, Hosakote, that made HCT observations possi-
ble. The facilities at IAO and CREST are operated by the
Indian Institute of Astrophysics, Bangalore. The Joan Or´o
Telescope (TJO) of the Montsec Astronomical Observatory
(OAdM) is owned by the Catalan Government and operated
by the Institute for Space Studies of Catalonia (IEEC). This
research has made use of data from the OVRO 40-m moni-
toring program which was supported in part by NASA grants
NNX08AW31G, NNX11A043G and NNX14AQ89G, and NSF
grants AST-0808050 and AST-1109911, and private fund-
ing from Caltech and the MPIfR. S.K. acknowledges sup-
port from the European Research Council (ERC) under the
European Unions Horizon 2020 research and innovation pro-
gramme under grant agreement No. 771282. This research has
made use of the NASA/IPAC Extragalactic Database (NED),
which is funded by the National Aeronautics and Space Ad-
ministration and operated by the California Institute of Tech-
nology. This publication makes use of data obtained at the
Mets¨
ahovi Radio Observatory, operated by Aalto University
in Finland. Part of this work is based on archival data, soft-
ware or online services provided by the Space Science Data
Center - ASI. We would like to thank the anonymous journal
reviewer for his/her comments that helped to improve the
manuscript.
DATA AVAILABILITY
The data used in this article were accessed from the MAGIC
telescope http://vobs.magic.pic.es/fits/. The derived
data generated in this research will be shared on reasonable
request to the corresponding author.
REFERENCES
H. E. S. S. Collaboration, Abdalla, H., Aharonian, F., et al. 2019,
MNRAS, 486, 3886
Abdo, A. A., Ackermann, M., Ajello, M., et al. 2010, ApJS, 188,
405. doi:10.1088/0067-0049/188/2/405
Abdollahi S., et al., 2020, ApJS, 247, 33
Acciari, V. A., Ansoldi, S., Antonelli, L. A., et al. 2020,
arXiv:2012.11380
Acero, F., Ackermann, M., Ajello, M., et al. 2015, ApJS, 218, 23.
doi:10.1088/0067-0049/218/2/23
Acharya, B. S., Actis, M., Aghajani, T., et al. 2013, Astroparticle
Physics, 43, 3. doi:10.1016/j.astropartphys.2013.01.007
Ahnen, M. L., Ansoldi, S., Antonelli, L. A., et al. 2016, A&A, 595,
A98
Akaike, H., 1974, IEEE Trans. Autom. Control, AC-19, 716
Aleksi´c, J., Ansoldi, S., Antonelli, L. A., et al. 2016, Astroparticle
Physics, 72, 61
Aleksi´c, J., Ansoldi, S., Antonelli, L. A., et al. 2016, Astroparticle
Physics, 72, 76
Algaba, J.-C., Zhao, G.-Y., Lee, S.-S., et al. 2015, JKAS, 48, 237
Arnaud, K. A. 1996, Astronomical Data Analysis Software and
Systems V, 17
Atwood W. B., Abdo A. A., Ackermann M., Althouse W., Ander-
son B., Axelsson M., Baldini L., et al., 2009, ApJ, 697, 1071
Baars, J. W. M., Genzel, R., Pauliny-Toth, I. I. K., et al. 1977,
A&A, 500, 135
Barnacka, A., Glicenstein, J.-F., & Moudden, Y. 2011, A&A, 528,
L3. doi:10.1051/0004-6361/201016175
Barnacka, A., Geller, M. J., Dell’Antonio, I. P., et al. 2016, ApJ,
821, 58. doi:10.3847/0004-637X/821/1/58
Barnacka A., Geller M. J., Dell’Antonio I. P., Benbow W., 2015,
ApJ, 799, 48. doi:10.1088/0004-637X/799/1/48
Barnacka A., 2018, PhR, 778, 1. doi:10.1016/j.physrep.2018.10.001
van den Berg, J. P., B¨
ottcher, M., Dom´ınguez, A., et al. 2019, ApJ,
874, 47
Biggs, A. D., Browne, I. W. A., Helbig, P., et al. 1999, MNRAS,
304, 349
Biggs, A. D., Wucknitz, O., Porcas, R. W., et al. 2003, MNRAS,
338, 599. doi:10.1046/j.1365-8711.2003.06050.x
Biggs, A. D., & Browne, I. W. A. 2018, MNRAS, 476, 5393
Breeveld, A. A., et al. 2010, MNRAS, 406, 1687
Burnham, K.P., Anderson, D.R., & Huyvaert, K.P., 2011, Behav-
ioral Ecology and Sociobiology, 65, 23
Burrows, D. N., Hill, J. E., Nousek, J. A., et al. 2004, in Proc. SPIE,
Vol. 5165, X-Ray and Gamma-Ray Instrumentation for Astron-
omy XIII, ed. K. A. Flanagan & O. H. W. Siegmund, 201–216
Buson, S., Cheung, C. C., Larsson, S., et al. 2015, arXiv:1502.03134
Buson, S., Cheung, C. T., Larsson, S., et al. 2015, 34th Interna-
tional Cosmic Ray Conference (ICRC2015), 34, 877
Carilli, C. L., Rupen, M. P., & Yanny, B. 1993, ApJ, 412, L59
Cash W., 1979, ApJ, 228, 939
Ceribella, G., D’Amico, G., Dazzi, F., et al. 2019, 36th Interna-
tional Cosmic Ray Conference (ICRC2019), 645
Cheung, C. C., Larsson, S., Scargle, J. D., et al. 2014, ApJ, 782,
L14
Cohen, A. S., Hewitt, J. N., Moore, C. B., & Haarsma, D. B. 2000,
ApJ, 545, 578
Cohen, J. G., Lawrence, C. R., & Blandford, R. D. 2003, ApJ, 583,
67
Condon, J. J., Cotton, W. D., Greisen, E. W., et al. 1998, AJ, 115,
1693. doi:10.1086/300337
Corbett, E. A., Browne, I. W. A., Wilkinson, P. N., & Patnaik,
A. 1996, Astrophysical Applications of Gravitational Lensing,
173, 37
Costamante, L., Cutini, S., Tosti, G., et al. 2018, MNRAS, 477,
4749
Dazzi., F. et al., IEEE Transactions on Nuclear Science, doi:
10.1109/TNS.2021.3079262.
Dermer, C. D. & Menon, G. 2009, High Energy Radiation from
Black Holes: Gamma Rays, Cosmic Rays, and Neutrinos by
Charles D. Dermer and Govind Menon. Princeton Univerisity
Press, November 2009.
D’Elia, V., Perri, M., Puccetti, S., et al. 2013, A&A, 551, A142.
MNRAS 000,1–21 (2020)
MWL study of QSO B0218+357 19
doi:10.1051/0004-6361/201220863
Dom´ınguez, A., Primack, J. R., Rosario, D. J., et al. 2011, MNRAS,
410, 2556
Eulaers, E., & Magain, P. 2011, A&A, 536, A44
Evans, P. A., Beardmore, A. P., Page, K. L., et al. 2009, MNRAS,
397, 1177
Evans, P. A.; Osborne, J. P.; Beardmore, A. P. et al., Astrophys.
J. Suppl. Ser., 210, 8
Falco, E. E., Impey, C. D., Kochanek, C. S., et al. 1999, ApJ, 523,
617
Fallah Ramazani, V., Lindfors, E., & Nilsson, K. 2017, A&A, 608,
A68
Falomo, R., Treves, A., Scarpa, R., et al. 2017, MNRAS, 470, 2814
Finke, J. D. 2016, ApJ, 830, 94
Fruck, C., & Gaug, M. 2015, European Physical Journal Web of
Conferences, 89, 02003
Gabriel, C., Denby, M., Fyfe, D. J., et al. 2004, Astronomical Data
Analysis Software and Systems (ADASS) XIII, 759
Ghisellini, G. & Tavecchio, F. 2009, MNRAS, 397, 985.
doi:10.1111/j.1365-2966.2009.15007.x
Ghisellini, G., Tavecchio, F., Foschini, L., et al. 2010, MNRAS,
402, 497
Gregory, P. C., Scott, W. K., Douglas, K., et al. 1996, ApJS, 103,
427. doi:10.1086/192282
Gregory, P. C. & Condon, J. J. 1991, ApJS, 75, 1011.
doi:10.1086/191559
Greisen, E. W. 2003, ASSL, 28, 71
Gottlieb D. H., 1994, JMP, 35, 5507. doi:10.1063/1.530762
Hada, K., Park, J., Kino, M., et al. 2017, PASJ, 69, 71
Hada, K., Niinuma, K., Sitarek, J., et al. 2020, ApJ, 901, 2.
doi:10.3847/1538-4357/abaab1
Jackson, N., Xanthopoulos, E., Browne, I.W.A., 2000, MNRAS,
311, 389
Jackson, N., Battye, R. A., Browne, I. W. A., et al. 2007, MNRAS,
376, 371. doi:10.1111/j.1365-2966.2007.11442.x
Jansen, F., Lumb, D., Altieri, B., et al. 2001, A&A, 365, L1
Kalberla, P. M. W., Burton, W. B., Hartmann, D., et al. 2005,
A&A, 440, 775
Kuehr, H., Witzel, A., Pauliny-Toth, I. I. K., et al. 1981, A&AS,
45, 367
Keeton C. R., 2001, arXiv, astro-ph/0102341
Larchenkova, T. I., Lutovinov, A. A., & Lyskova, N. S. 2011, As-
tronomy Letters, 37, 233. doi:10.1134/S1063773711040050
Lee, S.-S., Oh, C.-S., Roh, D.-G., et al. 2015, JKAS, 48, 125
MAGIC Collaboration, Acciari, V. A., Ansoldi, S., et al. 2020,
A&A, 640, A132. doi:10.1051/0004-6361/202037811
Mannucci F., Basile, F., Poggianti, B. M., Cimatti, A., Daddi, E.,
Pozzetti, L., Vanzi, L., 2001, MNRAS, 326, 745
Menten, K. M., & Reid, M. J. 1996, ApJ, 465, L99
Mittal, R., Porcas, R., Wucknitz, O., Biggs, A., & Browne, I. 2006,
A&A, 447, 515
Mittal, R., Porcas, R., & Wucknitz, O. 2007, A& A, 465, 405
Myers, S. T., Jackson, N. J., Browne, I. W. A., et al. 2003, MNRAS,
341, 1. doi:10.1046/j.1365-8711.2003.06256.x
Nieppola, E., Tornikoski, M., L¨
ahteenm¨
aki, A., et al. 2007, AJ,
133, 1947. doi:10.1086/512609
Nigro, C., Sitarek, J., Craig M. & Gliwny, P. (2020, September 28).
agnpy: modelling Active Galactic Nuclei radiative processes
with python. Zenodo. http://doi.org/10.5281/zenodo.4055176
Niinuma, K., Lee, S.-S., Kino, M., et al. 2014, PASJ, 66, 103
Nilsson, K., Lindfors, E., Takalo, L.O., et al. 2018, A&A, 620, 185
Nolan, P. L., Abdo, A. A., Ackermann, M., et al. 2012, ApJS, 199,
31. doi:10.1088/0067-0049/199/2/31
O’Dea, C. P., Baum, S. A., Stanghellini, C., et al. 1992, AJ, 104,
1320
Olgu´ın-Iglesias, A., Le´on-Tavares, J., Kotilainen, J. K., et al. 2016,
MNRAS, 460, 3202
Paiano, S., Landoni, M., Falomo, R., et al. 2017, ApJ, 837, 144
Paliya, V. S., Dom´ınguez, A., Ajello, M., et al. 2021, ApJS, 253,
46. doi:10.3847/1538-4365/abe135
Patnaik A. R., Porcas R. W., Browne I. W. A., 1995, MNRAS,
274, L5.
Petters A. O., Werner M. C., 2010, GReGr, 42, 2011.
doi:10.1007/s10714-010-0968-6
Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2011,
A&A, 536, A7. doi:10.1051/0004-6361/201116474
Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2014,
A&A, 571, A28. doi:10.1051/0004-6361/201321524
Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2016,
A&A, 594, A26. doi:10.1051/0004-6361/201526914
Poole, T. S., et al. 2008, MNRAS, 383, 627
Readhead, A. C. S., Lawrence, C. R., Myers, S. T., et al. 1989,
ApJ, 346, 566. doi:10.1086/168039
Richards, J. L., Max-Moerbeck, W., Pavlidou, V., et al. 2011,
ApJS, 194, 29. doi:10.1088/0067-0049/194/2/29
Rolke, W. A., L´opez, A. M., & Conrad, J. 2005, NIM A, 551, 493
Schlafly, E. F. & Finkbeiner, D. P. 2011, ApJ, 737, 103.
doi:10.1088/0004-637X/737/2/103
Shayduk, M., 2013, 33rd International Cosmic Ray Conference,
3000, astro-ph.IM/1307.4939
Shepherd, M. C., Pearson, T. J., & Taylor, G. B. 1994, BAAS,
26, 987
Sitarek, J., & Bednarek, W. 2016, MNRAS, 459, 1959
Str¨
uder, L., Briel, U., Dennerl, K., et al. 2001, A&A, 365, L18
Ter¨
asranta, H., Tornikoski, M., Mujunen, A. et al. 1998, A&AS,
132, 305
Vovk, I., & Neronov, A. 2016, A&A, 586, A150
Wilms, J., Allen, A., & McCray, R. 2000, ApJ, 542, 914
Wood M., Caputo R., Charles E., Di Mauro M., Magill J.,
Perkins J. S., Fermi-LAT Collaboration, 2017, ICRC, 301, 824,
ICRC...35
Wright, E. L., Chen, X., Odegard, N., et al. 2009, ApJS, 180, 283.
doi:10.1088/0067-0049/180/2/283
Wright, E. L., Eisenhardt, P. R. M., Mainzer, A. K., et al. 2010,
AJ, 140, 1868. doi:10.1088/0004-6256/140/6/1868
Wong K. C., Suyu S. H., Chen G. C.-F., Rusu C. E., Millon M.,
Sluse D., Bonvin V., et al., 2020, MNRAS, 498, 1420.
Wucknitz, O., Biggs, A. D., & Browne, I. W. A. 2004, MNRAS,
349, 14. doi:10.1111/j.1365-2966.2004.07514.x
York, T., Jackson, N., Browne, I. W. A., et al. 2005, MNRAS, 357,
124. doi:10.1111/j.1365-2966.2004.08618.x
Zanin, R., Carmona, E., Sitarek, J., et al., 2013, Proc of 33rd
ICRC, Rio de Janeiro, Brazil, Id. 773
Zhang M., Jackson N., Porcas R. W., Browne I. W. A., 2007, MN-
RAS, 377, 1623. doi:10.1111/j.1365-2966.2007.11718.x
Zhao, G.-Y., Jung, T., Sohn, B. W., et al. 2019, JKAS, 52, 23
AFFILIATIONS
1Instituto de Astrof´ısica de Canarias and Dpto. de As-
trof´ısica, Universidad de La Laguna, 38200, La Laguna,
Tenerife, Spain
2Universit`a di Udine and INFN Trieste, I-33100 Udine, Italy
3National Institute for Astrophysics (INAF), I-00136 Rome,
Italy
4ETH Z¨
urich, CH-8093 Z¨
urich, Switzerland
5Institut de F´ısica d’Altes Energies (IFAE), The Barcelona
Institute of Science and Technology (BIST), E-08193 Bel-
laterra (Barcelona), Spain
6Japanese MAGIC Group: Institute for Cosmic Ray Re-
search (ICRR), The University of Tokyo, Kashiwa, 277-8582
Chiba, Japan
7Technische Universit¨
at Dortmund, D-44221 Dortmund,
MNRAS 000,1–21 (2020)
20 V. A. Acciari et. al.
Germany
8Croatian MAGIC Group: University of Zagreb, Faculty of
Electrical Engineering and Computing (FER), 10000 Zagreb,
Croatia
9IPARCOS Institute and EMFTEL Department, Universi-
dad Complutense de Madrid, E-28040 Madrid, Spain
10 Centro Brasileiro de Pesquisas F´ısicas (CBPF), 22290-180
URCA, Rio de Janeiro (RJ), Brazil
11 Universit`a di Padova and INFN, I-35131 Padova, Italy
12 University of Lodz, Faculty of Physics and Applied Infor-
matics, Department of Astrophysics, 90-236 Lodz, Poland
13 Universit`a di Siena and INFN Pisa, I-53100 Siena, Italy
14 Deutsches Elektronen-Synchrotron (DESY), D-15738
Zeuthen, Germany
15 INFN MAGIC Group: INFN Sezione di Torino and Uni-
versit`a degli Studi di Torino, 10125 Torino, Italy
16 Max-Planck-Institut f¨
ur Physik, D-80805 M¨
unchen, Ger-
many
17 Universit`a di Pisa and INFN Pisa, I-56126 Pisa, Italy
18 Universitat de Barcelona, ICCUB, IEEC-UB, E-08028
Barcelona, Spain
19 Armenian MAGIC Group: A. Alikhanyan National Science
Laboratory
20 Centro de Investigaciones Energ´eticas, Medioambientales
y Tecnol´ogicas, E-28040 Madrid, Spain
21 INFN MAGIC Group: INFN Sezione di Bari and Diparti-
mento Interateneo di Fisica dell’Universit`a e del Politecnico
di Bari, 70125 Bari, Italy
22 Croatian MAGIC Group: University of Rijeka, Department
of Physics, 51000 Rijeka, Croatia
23 Universit¨
at W¨
urzburg, D-97074 W¨
urzburg, Germany
24 Finnish MAGIC Group: Finnish Centre for Astronomy
with ESO, University of Turku, FI-20014 Turku, Finland
25 Departament de F´ısica, and CERES-IEEC, Universitat
Aut`onoma de Barcelona, E-08193 Bellaterra, Spain
26 Armenian MAGIC Group: ICRANet-Armenia at NAS RA
27 Croatian MAGIC Group: University of Split, Faculty of
Electrical Engineering, Mechanical Engineering and Naval
Architecture (FESB), 21000 Split, Croatia
28 Croatian MAGIC Group: Josip Juraj Strossmayer Univer-
sity of Osijek, Department of Physics, 31000 Osijek, Croatia
29 Japanese MAGIC Group: RIKEN, Wako, Saitama 351-
0198, Japan
30 Japanese MAGIC Group: Department of Physics, Kyoto
University, 606-8502 Kyoto, Japan
31 Japanese MAGIC Group: Department of Physics, Tokai
University, Hiratsuka, 259-1292 Kanagawa, Japan
32 Saha Institute of Nuclear Physics, HBNI, 1/AF Bidhanna-
gar, Salt Lake, Sector-1, Kolkata 700064, India
33 Inst. for Nucl. Research and Nucl. Energy, Bulgarian
Academy of Sciences, BG-1784 Sofia, Bulgaria
34 Finnish MAGIC Group: Astronomy Research Unit, Uni-
versity of Oulu, FI-90014 Oulu, Finland
35 Croatian MAGIC Group: Ruđer Boˇskovi´c Institute, 10000
Zagreb, Croatia
36 INFN MAGIC Group: INFN Sezione di Perugia, 06123 Pe-
rugia, Italy
37 INFN MAGIC Group: INFN Roma Tor Vergata, 00133
Roma, Italy
38 now at University of Innsbruck
39 also at Port d’Informaci´o Cient´ıfica (PIC) E-08193 Bel-
laterra (Barcelona) Spain
40 now at Ruhr-Universit¨
at Bochum, Fakult¨
at f¨
ur Physik
und Astronomie, Astronomisches Institut (AIRUB), 44801
Bochum, Germany
41 also at Dipartimento di Fisica, Universit`a di Trieste, I-
34127 Trieste, Italy
42 Max-Planck-Institut f¨
ur Physik, D-80805 M¨
unchen, Ger-
many
43 also at INAF Trieste and Dept. of Physics and Astronomy,
University of Bologna
44 Japanese MAGIC Group: Institute for Cosmic Ray Re-
search (ICRR), The University of Tokyo, Kashiwa, 277-8582
Chiba, Japan
45 Dipartimento di Matematica e Fisica “E. De Giorgi”, Uni-
versit`a del Salento, Lecce, Italy
46 Istituto Nazionale di Fisica Nucleare, Sezione di Lecce, I-
73100 Lecce, Italy
47 INAF-IRA Bologna, I-40129 Bologna, Italy
48 Indian Institute of Astrophysics, Bangalore 560034, India
49 Owens Valley Radio Observatory, California Institute of
Technology, Pasadena, CA 91125, USA
50 Finnish Center for Astronomy with ESO (FINCA), Uni-
versity of Turku, FI-20014, Turku, Finland
51 Aalto University Mets¨
ahovi Radio Observatory, Mets¨
ahov-
intie 114, 02540 Kylm¨
al¨
a, Finland
52 Institute of Astrophysics, Foundation for Research and
Technology-Hellas, GR-71110 Heraklion, Greece
53 Department of Physics, Univ. of Crete, GR-70013 Herak-
lion, Greece
54 Departamento de Astronom´ıa, Universidad de Chile,
Camino El Observatorio 1515, Las Condes, Santiago, Chile
55 CePIA, Departamento de Astronom´ıa, Universidad de
Concepti´on, Concepci´on, Chile
56 Aalto University Department of Electronics and Nanoengi-
neering, P.O. BOX 15500, FI-00076 AALTO, Finland.
57 Mizusawa VLBI Observatory, National Astronomical Ob-
servatory of Japan, 2-12 Hoshigaoka, Mizusawa, Oshu, Iwate
023-0861, Japan
58 Department of Astronomical Science, The Graduate Uni-
versity for Advanced Studies (SOKENDAI), 2-21-1 Osawa,
Mitaka, Tokyo 181-8588, Japan
59 Graduate School of Sciences and Technology for Innova-
tion, Yamaguchi University, Yoshida 1677-1, Yamaguchi, Ya-
maguchi 753-8512, Japan
60 Smithsonian Astrophysical Observatory, Cambridge, MA
02138, USA
61 Astronomical Observatory, Jagiellonian University, ul.
Orla 171, 30-244 Cracow, Poland
APPENDIX A: SEARCH FOR HARD GEV STATES
The Fermi-LAT flux and photon index information are used
to evaluate during which nights a detection with MAGIC
would be most likely. The expected integral fluxes above 100
GeV were calculated and compared with the obtained daily
upper limits (see Fig. A1). For each 24 hr bin of Fermi-LAT
data that overlaps with MAGIC observations the Fermi-LAT
power-law spectrum were extrapolated to sub-TeV energies
and convolved the flux with EBL absorption using the model
of Dom´ınguez et al. (2011). Bins with Fermi-LAT TS <9
and those in which the uncertainty of the flux above 0.1 GeV
exceeds the flux value were removed from the analysis. The
MNRAS 000,1–21 (2020)
MWL study of QSO B0218+357 21
14−
10 13−
10 12−
10 11−
10 10−
10 ]
-1
s
-2
Extrapolated integral flux > 100 GeV [cm
11−
10
10−
10
]
-1
s
-2
Integral flux upper limit > 100 GeV [cm
Figure A1. Integral upper limit on the flux >100 GeV obtained with
MAGIC telescopes as a function of the expected flux using contem-
poraneous Fermi-LAT data (downward triangles). For comparison
a measurement of the same quantities from the 2014 flare is shown
in gray. Thick oblique lines show the proportionality of the two
fluxes for the case when the true flux is equal to extrapolated one
(black) or when the true flux scales with the expected one like in
2014 flare (gray)
VHE flux were integrated and computed its uncertainty tak-
ing into account the uncertainty of the flux above 0.1 GeV
and the spectral index. It should be noted that in case of an
intrinsic break or a cut-off in the spectrum the true flux will
be lower than such extrapolated values. In fact, applying the
same procedure to the 2014 flare data the measured flux is a
factor of ∼5below the extrapolated one.
In none of the bins with contemporaneous MAGIC obser-
vations the extrapolated flux value reached the flux of the
2014 flare, except for the case of MJD=58779 when such flux
is consistent within the uncertainty bars.
APPENDIX B: SCENARIO FOR FLARES
In addition to the average emission, a few other interesting
events occurred during the monitoring. The MWL broadband
SED during the X-ray flare (F2) shows a very similar shape
in the synchrotron peak as during the average state. Despite
higher X-ray flux, the GeV spectrum is consistent with the
one obtained from the average state of the source. Limited
MWL data, unknown duration of X-ray flare, uncertain lens-
ing and absorption (i.e. if the emission seen in different bands
is dominantly from A or B image, that would affect magnifi-
cation and absorption) and low statistics in gamma-ray data
make modelling of this events difficult. In the framework of
the model used for the explanation of the average emission the
X-ray emission of the source is mainly of SSC origin. There-
fore, the event can be naturally explained by compression of
the emission region, which would enhance the efficiency of
this process. Such a compression (if it does not change the
ambient magnetic field) would not modify synchrotron and
EC components. Note also that a possible enhancement of
the magnetic field during compression of the emission region
does not have to increase considerably the synchrotron peak,
as it is mainly explained in the average state modelling as
the emission from large scale jet component. Alternative ex-
planation of the X-ray flare would involve enhanced emission
from image A, which would show up in the hard part of the
X-ray spectrum, while due to the strong absorption would
not increase considerably the optical flux density. Unfortu-
nately the X-ray data are not precise enough to allow us to
distinguish between the two scenarios based on the X-ray ab-
sorption.
The second interesting episode involves short optical flare
during a longer GeV flare (F1). Hardening of the GeV spec-
trum and increase of the optical flux density points to hard-
ening of the electron distribution and thus shifting of both
peaks to higher energies. The combination of different vari-
ability time scales in those two ranges makes the associa-
tion of both events uncertain and complicates modelling of
the emission. A possible scenario that would explain differ-
ent time scales of optical and GeV emission would involve a
blob travelling along the jet with a ramping up GeV emis-
sion. Since according to the low state modelling, most of the
synchrotron emission is explained by “Far”emission zone (see
Fig. 11) and thus such newly emerging blob would not show
up as immediately enhanced optical emission. However if the
new blob encounters a stationary feature in the jet, or an in-
ternal shock, it can cause enhancement of the magnetic field
and shift of the synchrotron peak to higher energies. Since
the SED of QSO B0218+357 in optical range is very steep it
would cause a strong optical flare, such as seen during period
F1.
The third investigated period, August 2020 MWL cam-
paign cannot be firmly claimed as an enhanced flux state.
Nevertheless the detection of two >10 GeV photons without
accompanying clear increase of the flux at GeV energies, is
consistent with a very hard electron energy distribution. Un-
fortunately the VHE could be probed only in neighbouring
nights. Curiously, a small hint of enhancement of the B-band
flux is also consistent with hardening of the electron spec-
trum, as according to the low state model, the UV data probe
the high energy part of the electron distribution. Short term
wavelength-dependent variability in optical-UV range could
be then the effect of variability of the electron energy distri-
bution convoluted with absorption in the lens galaxy. While
no X-ray variability is present during August 2020, the av-
erage X-ray flux during this period is enhanced with respect
to the low state, and is similar to the flux level of the F2
period. Within the framework of the modelling this could be
explained if the compression of the emission region persisted
between the MJD 58863.7 flare and August 2020.
This paper has been typeset from a T
EX/L
A
T
EX file prepared by
the author.
MNRAS 000,1–21 (2020)
