Multi-wavelength flare observations of the blazar S5 1803+784
R. Nesci, S. Cutini, C. Stanghellini, F. Martinelli, A. Maselli, V.M. Lipunov, V. Kornilov, R.R. Lopez, A. Siviero, M. Giroletti, M. Orienti
MMNRAS , 1–12 (2021) Preprint 22 February 2021 Compiled using MNRAS L A TEX style file v3.0
Multi–wavelength flare observations of the blazar S5 1803+784
R. Nesci, ★ S. Cutini, C. Stanghellini, F. Martinelli, A. Maselli, , V.M. Lipunov, V. Kornilov, R. R. Lopez, A. Siviero, M. Giroletti, and M. Orienti INAF/IAPS, via Fosso del Cavaliere 100, 00133 Roma, Italy INFN, Via Alessandro Pascoli, s.n.c, 06130, Perugia, Italy INAF/IRA, Via P. Gobetti, 101, 40129, Bologna, Italy Lajatico Astronomical Centre, Italy INAF/OAR, Via Frascati 33, I-00078, Monte Porzio Catone (Roma), Italy ASI/SSDC, Via del Politecnico snc, I-00133, Roma, Italy Lomonosov Moscow State University, Physics Department, Sternberg Astronomical Institute, 119991 Moscow, Vorobievy hills, 1 Instituto de Astrofisica de Canarias Via Lactea, s/n E38205 - La Laguna (Tenerife), Spain Department of Physics and Astronomy, University of Padova, Italy
Accepted 2021-02-17. Received 2021-02-17; in original form 2020-11-12
ABSTRACT
The radio, optical, and 𝛾 -ray light curves of the blazar S5 1803+784, from the beginning of the Fermi
Large Area Telescope(LAT) mission in August 2008 until December 2018, are presented. The aim of this work is to look for correlations amongdifferent wavelengths useful for further theoretical studies. We analyzed all the data collected by
Fermi
LAT for this source, takinginto account the presence of nearby sources, and we collected optical data from our own observations and public archive data tobuild the most complete optical and 𝛾 -ray light curve possible. Several 𝛾 -ray flares (F > . − ph ( E > . ) cm − s − ) withoptical coverage were detected, all but one with corresponding optical enhancement; we also found two optical flares without a 𝛾 -ray counterpart. We obtained two Swift
Target of Opportunity observations during the strong flare of 2015. Radio observationsperformed with VLBA and EVN through our proposals in the years 2016-2020 were analyzed to search for morphologicalchanges after the major flares. The optical/ 𝛾 -ray flux ratio at the flare peak varied for each flare. Very minor optical V-I colorchanges were detected during the flares. The X-ray spectrum was well fitted by a power law with photon spectral index 𝛼 =1.5,nearly independent of the flux level: no clear correlation with the optical or the 𝛾 -ray emission was found. The 𝛾 -ray spectralshape was well fitted by a power law with average photon index 𝛼 = 2.2. These findings support an Inverse Compton origin forthe high-energy emission of the source, nearly co-spatial with the optically emitting region. The radio maps showed two newcomponents originating from the core and moving outwards, with ejection epochs compatible with the dates of the two largest 𝛾 -ray flares. Key words:
BL Lacertae objects: individual; galaxies: jets; (galaxies:) quasars: emission lines; X-rays: galaxies; gamma-rays:galaxies
BL Lacertae objects are a subclass of Active Galactic Nuclei (AGN)characterized by fast and large flux variations. Their emission rangesfrom radio frequencies up to 𝛾 -rays, and their spectral energy distri-bution (SED) in a Log( 𝜈 ) vs Log( 𝜈 · 𝐹 𝜈 ) plane may be described by adouble bell shape: the first peak is located between the far infrared andX-ray frequencies, while the second one is always at much higher en-ergies (from X-ray to 𝛾 -ray energies). A current interpretation of thisdouble bell shape is that synchrotron radiation from relativistic elec-trons in a highly collimated jet is responsible for the low-frequencypeak, while inverse Compton radiation from these electrons impact-ing local photons produces the higher-frequency peak. This modelis likely an oversimplification with respect to reality: the black hole ★ E-mail: [email protected] at the centre of the host galaxy is surrounded by an accretion disk,a hot corona, and the relativistic jet approximately directed alongthe black hole rotation axis is likely structured at least in an innerpart (spine) and an outer envelope (sheath) with different physicalconditions (Chiaberge et al. 2000). Such a differentiated structureimplies that the radiation at different frequencies may come fromphysically different spatial regions: flux variations at different fre-quencies may therefore reach the observer at different times. Thusmulti-wavelength simultaneous observations are an efficient tool toexplore the physical structure of the emitting regions.This paper is an observational contribution to the study of S51803+784, a radio-selected BL Lac object (Biermann et al. 1981) atz=0.683 (Lawrence et al. 1996); a detailed modeling of the emissionprocesses and morphology is beyond the scope of this paper. S51803+784 is characterized by large variations in the optical range(Nesci et al. 2002, 2012) and is well detected at 𝛾 -ray energies by the © a r X i v : . [ a s t r o - ph . H E ] F e b R. Nesci et al.
Fermi
Large Area Telescope (LAT) (Atwood et al. 2009), from thefirst LAT Catalog of AGN (Abdo et al. 2009), up to the most recentone (Ajello et al. 2020). In the taxonomical classification scheme ofBL Lac objects by Padovani & Giommi (1995) the source is LowSynchrotron Peaked (LSP), with the synchrotron emission peakedaround 10 Hz: this is clearly shown by its broadband SED (Nesciet al. 2002).In this paper we present the light curves of the source in the radio(15 GHz), optical (R 𝐶 ) and 𝛾 -ray bands from year 2008 to 2018, withsnapshots in the X-ray and optical bands by the Neil Gehrels SwiftObservatory
X-ray Telescope (XRT) and the UltraViolet and OpticalTelescope (UVOT), to search for correlations among different wave-lengths. Furthermore we present Very Long Baseline Interferometry(VLBI) radio observations performed after two large flares, to lookfor morphological changes in its inner jet structure.
We have been monitoring this source in the optical band since1996 (Nesci et al. 2002, 2012) despite with some large gaps. It wasserendipitously observed by the MASTER robotic network (Lipunovet al. 2010) since 2010. Due to its very northern position in the skyit was not covered by the Catalina Sky Survey (Drake et al. 2009);since July 2011 this source became a target of the blazar monitoringprogram by the KAIT telescope (Li et al. 2003).In this paper we used our observations taken with several tele-scopes since 2009: the 50cm F/4.5 at Vallinfreda, the 30cm F/4.5 atGreve in Chianti, the 1.5m telescope at the Loiano Observatory, the23cm F/10 of the Department of Physics of La Sapienza University,the 50cm F/4.4 at the Lajatico Astronomical Centre. All these tele-scopes are equipped with CCD cameras and Cousins R filters. Mag-nitudes were derived using aperture photometry with IRAF/apphot ,using always the same comparison sequence (Nesci et al. 2002).In past years (Nesci et al. 2002, 2012), the color indexes (V-R andR-I) of the source showed only small variations, with a mild bluer-when-brighter behaviour. This is also shown by the present data set(see Table 1, from which we derive an average increase of 0.04 magin R-I for a 1 mag increase in R), and makes us confident of thereliability of converting V magnitudes into R ones with a fixed colorindex for the purpose of building a denser historical light curve.Therefore we included in our data set the V magnitudes from the30cm F/10 telescope of the Foligno Observatory and from the 40cmF/8 telescope of the Royal Observatory of Belgium (Lampens & vanCauteren 2015), converting their V magnitudes into R 𝐶 , adopting V-R=0.50 mag. For the KAIT observations, which are unfiltered and usethe USNO-B1 (Monet et al. 2003) R2 magnitudes as reference, weapplied an average correction of -0.20 mag to their psf magnitudes,based on the average offset of our photometric sequence stars withrespect to the USNO-B1 catalog: this value is within the quotedsystematic uncertainty given in the KAIT database. For the MASTERobservations, which are also unfiltered and expressed as a nearly R 𝐶 magnitude, we verified a good consistency with our magnitudes forthe few nights of simultaneous observations. The resulting overallR 𝐶 light curve is shown in the middle panel of Fig. 1.This light curve shows a baseline flux level between 1 and 2 mJy, IRAF - Image Reduction and Analysis Facility, distributed by NOAO,operated by AURA, Inc. under agreement with the US NSF.
Table 1.
Optical colors from la Lajatico (LA) and Loiano (LO) Observatoriesmultifilter observations. Date in Column 1, R 𝐶 in column 2, V-R 𝐶 in column3, R 𝐶 -I 𝐶 in column 4, V-I 𝐶 in column 5, Observatory in column 6.date date R V-R R-I V-I telMJD mag mag mag mag2011-07-21 55763.84 15.85 0.49 0.70 1.19 LO2011-07-22 55764.83 15.88 0.50 0.70 1.20 LO2015-08-21 57255.85 14.19 0.49 0.59 1.08 LA2015-08-26 57260.82 13.91 0.50 0.59 1.09 LA2015-08-28 57262.84 13.76 0.51 0.64 1.15 LA2015-08-31 57265.86 13.97 0.46 0.61 1.07 LA2015-09-05 57270.86 14.66 0.48 0.68 1.16 LA2015-09-09 57274.84 14.55 0.48 0.68 1.16 LA2015-09-19 57284.87 14.30 0.53 0.65 1.18 LA2015-09-21 57286.81 14.58 0.48 0.69 1.17 LA2015-10-11 57306.81 15.02 0.45 0.68 1.13 LA2015-10-22 57317.75 15.75 0.51 0.71 1.22 LA2015-11-24 57350.74 14.23 0.50 0.62 1.12 LA2016-05-21 57729.92 15.97 0.49 0.69 1.18 LA2016-06-26 57565.07 15.29 0.49 0.78 1.27 LA2016-07-01 57570.07 15.48 0.61 0.63 1.24 LA2016-07-09 57578.92 16.33 0.35 0.66 1.01 LA2016-12-15 57737.80 15.13 0.49 0.00 0.00 LA2016-12-29 57751.78 15.77 0.46 0.00 0.00 LA Table 2.
Optical flares data: date of the maximum (column 1), the peak R 𝐶 magnitude (column 2), the flare duration (column 3), the slope of the rising(column 4), and of the dimming branch (column 5).date peak R 𝐶 duration rising falling 𝛾 flareMJD mag days mag/d mag/d55211 15.0 45 0.07 yes55897 14.7 yes56001 14.4 yes56206 14.5 60 0.076 no56639 14.5 yes56859 13.9 45 0.22 yes57099 15.0 yes57262 13.7 0.11 yes57348 14.0 yes57488 14.8 no57728 14.4 30 0.08 0.10 yes with several flares, and some short time intervals with flux below 1mJy.The definition of flare is somewhat arbitrary, since it depends onthe assumed quiescent level of the source. From the optical lightcurve we derived an average quiescent level of 1.5 mJy (R=16.0mag), and we considered a flare as a rise and fall episode in thelight curve with a peak value higher than 3 mJy (R=15.2 mag). Forthe better sampled flares we measured the duration and the slopesof the rising and falling branches. The duration was defined as thetime interval during which the source was above the average level(R 𝐶 =16.0 mag): the uncertainties of these intervals depend on thesampling of the light curve. The slopes (mag/day) were derived witha linear fit to the manually selected relevant data. A total of 11 flareswere identified in this way and are listed in Table 2. The simultaneouspresence of a 𝛾 -ray flare (see Section 2.3 below) is given in the lastcolumn. MNRAS000
Optical flares data: date of the maximum (column 1), the peak R 𝐶 magnitude (column 2), the flare duration (column 3), the slope of the rising(column 4), and of the dimming branch (column 5).date peak R 𝐶 duration rising falling 𝛾 flareMJD mag days mag/d mag/d55211 15.0 45 0.07 yes55897 14.7 yes56001 14.4 yes56206 14.5 60 0.076 no56639 14.5 yes56859 13.9 45 0.22 yes57099 15.0 yes57262 13.7 0.11 yes57348 14.0 yes57488 14.8 no57728 14.4 30 0.08 0.10 yes with several flares, and some short time intervals with flux below 1mJy.The definition of flare is somewhat arbitrary, since it depends onthe assumed quiescent level of the source. From the optical lightcurve we derived an average quiescent level of 1.5 mJy (R=16.0mag), and we considered a flare as a rise and fall episode in thelight curve with a peak value higher than 3 mJy (R=15.2 mag). Forthe better sampled flares we measured the duration and the slopesof the rising and falling branches. The duration was defined as thetime interval during which the source was above the average level(R 𝐶 =16.0 mag): the uncertainties of these intervals depend on thesampling of the light curve. The slopes (mag/day) were derived witha linear fit to the manually selected relevant data. A total of 11 flareswere identified in this way and are listed in Table 2. The simultaneouspresence of a 𝛾 -ray flare (see Section 2.3 below) is given in the lastcolumn. MNRAS000 , 1–12 (2021) MJD . . . . F nu [ m J y ] F nu [ m J y ] F l u x ( E > M e V ) [ − c m − s − ph ] . . . . . . . . . . . . . . . . . . . . Figure 1.
The combined light curve of S5 1803+784 in linear scale: lower scale in MJD, upper scale in calendar years. Upper panel, 𝛾 -ray, binned at 10-dayintervals: blue points are detections, grey triangles are upper limits; middle panel, optical: orange points are our own data, red points are KAIT public data;lower panel, radio 15 GHz from OVRO monitoring. Vertical lines mark the flares discussed in the text. Only three spectra of this source have been published so far: one takenin 1987 (Lawrence et al. 1996) showing the MgII 2790Å emissionline with equivalent width (E.W.)=8.8Å; one taken in 1996 (Rector& Stocke 2001) that reports for the MgII line an E.W=2.8 Å; the lasttaken in 2018 (Paiano et al. 2020) with E.W.=8.3 Å, very similar tothe value of 1987. The magnitude of the source in 1987 is unknown,while from our monitoring we know that it was faint in 1996 (R=16.2mag), as well as in 2018 (R=15.9 mag).We obtained a spectrum of the source on 27 August 2015 (MJD57261), when it was near the maximum level (R=13.8 mag), with the1.22m telescope of the Asiago Observatory equipped with an AndoriDus DU440 camera and a Boller & Chivens spectrograph with a300 gr/mm grating: the dispersion was 2.3 Å/pixel and the spectralresolution 9.5 Å, derived from the FWHM of the night-sky emissionlines. Wavelength calibration was made with a He-Fe-Ar lamp andinstrumental response calibration with the star HR 7596. The blazarspectrum in arbitrary flux units is shown in Fig.2.The S/N ratio was about 20, allowing the detection of an emissionline of E.W= 3 Å. No such line was present at 4710 Å, the expectedposition of the MgII line, and the spectrum showed no features be- sides the telluric ones. The spectral slope was fully consistent withthat derived from our VRI photometry.
Fermi -LAT data analysis
The data analysis was performed following the
Fermi -LAT collab-oration recommendations for point-source analysis and is brieflydescribed as follows. We built the historical light curve collecting al-most 10 years of data available from the Fermi -LAT archive, startingfrom 4 August 2008 up to 31 December 2018. We analyzed the
Fermi -LAT data using the FermiTools package version 1.2.1 available fromthe Fermi Science Support Center (FSSC) through the Conda pack-age manager hosted on GitHub and the P8R3_SOURCE_V2 instru-ment response functions (Atwood et al. 2013). We selected a Regionof Interest (ROI) with a radius of 15 degrees centered at the positionof S5 1803+784 and we used only events belonging to the “source” https://fermi.gsfc.nasa.gov/ssc/data/analysis/documentation/Pass8_usage.html https://fermi.gsfc.nasa.gov/ssc/ https://github.com/fermi-lat/Fermitools-conda/ MNRAS , 1–12 (2021) R. Nesci et al.
Figure 2.
The spectrum of S5 1803+784 from the Asiago Observatory, cor-rected for instrumental response and smoothed with a running mean of 7pixels, in arbitrary flux units. Telluric Oxygen (O ) and water absorptionbands (H O) are well evident; the expected position of the MgII emissionis also marked. The apparent emission around 4525 Å is a residual fromnight-sky emission lines. class. The binned maximum likelihood was performed using as back-ground model the 4FGL sources lying in a ROI with a radius of 20degrees (4FGL; Abdollahi et al. 2020) and the isotropic and Galacticdiffuse emission components (iso_P8R3_SOURCE_V2_v1.txt andgll_iem_v07.fits , respectively). We fixed the spectral indexes ofbackground sources to the 4FGL values, but the fluxes of sourceswithin a radius of 10 degrees from the source of interest were leftfree to vary. Light curves were derived by dividing the data in binsof 10-day duration for the historical and 1-day and 3-day bins forthe flaring epochs. We defined as flaring state a flux above 100 MeVlarger than 2 . × − ph cm − s − in a 10-day bin, three times theaverage value of fluxes in the 10-day light curve. Each bin of the lightcurves and of the spectra was obtained using a power-law model forthe source of interest. For each bin we extracted the Test Statistic (TS;Mattox et al. 1996) defined as: 𝑇 𝑆 = − ( 𝑙𝑛𝐿 − 𝑙𝑛𝐿 ) where 𝐿 isthe value of the likelihood calculated in the null hypothesis (no pointsource in ROI) and 𝐿 is the value of the likelihood when the sourceof interest is included (after parameter optimization). We calculatedthe upper limit (UL) at 2 𝜎 confidence level using the Bayesian com-putation (Helene 1983) and integrated UL when the TS < < <
3, or the per-centage error on the flux greater than 50%. The light curve in the0.1-300 GeV range, binned at 10-day intervals, is shown in the upperpanel of Fig. 1: the light curve shows a flat, oscillating behaviourwith a typical flux level of 0 . × − ph cm − s − and several largeflares, irregularly spaced, listed in Table 3. The flux in the largestflares was nearly 10 times higher than the quiescent state. Thanksto the regular monitoring of the satellite, for all the 𝛾 -ray flares itwas possible to measure a duration, which ranges between 20 and 90 https://fermi.gsfc.nasa.gov/ssc/data/access/lat/BackgroundModels.html days, with an uncertainty of 3 days. In Table 3 we report the 𝛾 -rayflare details. We used X-ray data from the X-Ray Telescope (XRT) on board the
Neil Gehrels Swift Observatory (Gehrels et al. 2004), retrieved fromthe
Swift archive and acquired from 2009 to 2015; these include datacorresponding to the August 2015 flare, respectively 5 days beforeand 11 days after the the optical maximum. All data were processedwith the XRTDAS software package (v.3.6.0), developed at the SpaceScience Data Center (SSDC) of the Italian Space Agency (ASI)and distributed within the HEASoft package (v.6.28) by the NASAHigh Energy Astrophysics Archive Research Center (HEASARC).All the XRT observations were carried out in the most sensitivephoton counting (PC) readout mode. Event files were calibrated andcleaned applying standard filtering criteria with the xrtpipeline taskand using the latest calibration files available in the
Swift
CALDBdistributed by HEASARC. Events in the energy range 0.3–10 keVwith grades 0–12 were used in the analysis, and the exposure mapswere also created with xrtpipeline. Source detection was carried outusing the detection algorithm detect within ximage, also providingthe count rate corrected for psf, vignetting, and exposure.For each observation, source events were extracted from a cir-cle with a radius of 20 pixels centered at the source coordinates;a concentric annulus with radii of 50 and 80 pixels was used forthe background. These region files were then used as input to thexrtproducts task to obtain high-level scientific products: amongthese, the spectrum and ancillary files were used to perform a spec-tral analysis with xspec (v.12.11.1). We fitted these spectra using apower-law model, fixing the absorption at the Galactic value (n 𝐻 =3.4 10 cm − ), and we computed the flux in the 0.3–10 keV bandwith its 1 𝜎 error.To compare the X-ray and optical luminosity we looked at theSwift-UVOT simultaneous observations. Aperture photometry of theUVOT images was made with the UVOT on-line analysis tool avail-able from the SSDC (uvotimgqlprocver v1.14). The source radiuswas 5 arcsec and the sky level was evaluated within a concentriccorona between 27 and 35 arcsec distant from the source. Unfor-tunately only in 4 pointings were all 6 UVOT filters used, but wefound that the optical spectral slope was substantially constant (-0.57 ± 𝜈 ) vs log( 𝜈 F( 𝜈 ) plane), and therefore we couldcompute by extrapolation the corresponding R 𝐶 magnitudes for allthe UVOT pointings. In some cases we also had nearly simultaneousR 𝐶 observations from the ground, confirming the reliability of ourextrapolation procedure within 0.2 mag.Our results from the analysis of the Gehrels et al. 2004 data arereported in Table 4. The last column is the corresponding R 𝐶 mag-nitude. Observations with relatively poor exposure are also reported.Overall the XRT fluxes changed by a factor ∼
2, and the spectral slopehad an average value of 1.51 ± To explore if the strong flares of the years 2014 and 2015 pro-duced detectable changes in the structure of the inner part of thesource, we obtained radio images of the core of S5 1803+784 at sub-milliarcsecond resolution. Our observations were performed with theVery Long Baseline Array (VLBA) and the European VLBI Network(EVN) at 15, 22, and 43 GHz in six epochs, spanning from June 2016
MNRAS000
MNRAS000 , 1–12 (2021) Table 3. 𝛾 -ray flare data: the MJD of the peak emission (column 1), the peak flux value (column 2), the flare duration (column 3), the corresponding opticalflux (column 4), the optical/ 𝛾 -ray flux ratio (column 5, × ). date peak-flux duration R 𝐶 Opt/gamma-rayMJD ph cm − s − days mJy55067 2.6x10 − − − −
60 3.6 0.9055707 3.5x10 −
80 2.7 0.7755897 3.8x10 −
90 4.5 1.1855987 3.6x10 −
40 6.0 1.6656647 2.3x10 −
20 5.7 2.4856857 3.6x10 −
60 12.3 3.4157257 5.0x10 −
40 11.6 2.3257347 3.7x10 −
30 8.5 2.3057717 2.5x10 −
20 6.0 2.40
Table 4. data from Gehrels et al. 2004 observations: the date (column 1), the MJD (column 2), the last digits of the observation identifier (column 3), theexposure time (column 4), the total counts (column 5), the net count rate (column 6), the photon index (column 7), the flux in the 0.3-10 kev range assuming apower-law spectrum and the Galactic absorption (column 8).date date obs exp counts count rate Photon Index Flux 10 − R 𝐶 MJD id s cts/s erg cm − s − mag2009-06-07 54989.93 04 9337 710 0.075 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± to March 2018, as listed in Table 5. We also collected radio data at15 GHz from the 40m Telescope of the Owens Valley Radio Obser-vatory (OVRO; Richards et al. 2011) public archive to explore thepossibility of activity simultaneous to the optical and 𝛾 -ray flares.The radio data are shown in the bottom panel of Fig. 1. We did notfind any obvious evidence of correlations. Amplitude and phase calibration of the VLBA data were carried outwith the AIPS data reduction package in a standard way followingthe guidelines given in appendix C of the AIPS cookbook: the typicalflux error was 5%. Data from the third VLBA epoch were of lowerquality especially at 43 GHz, maybe for bad weather conditionsat some antennas. A-priori calibration of antenna gains proved tobe infeasible for EVN data, and we could not obtain reliable fluxdensity values. We relied on several iterations of imaging and self-calibration (amplitudes and phases) to obtain the relative positionof the components. Model fitting of the visibilities was performedwith the task MODELFIT available in the Caltech software packageDIFMAP.The morphological radio structure of S5 1803+784 at milliarc-second (mas) scale can be described as a diffuse emission from anopening jet, directed East-West, with knots and brightness enhance- ments in several locations along the jet (see e.g. Roland et al. 2008and references therein). Evidence in support of a helical jet has beenpresented by Britzen et al. (2010) and Kun et al. (2018) from a longterm monitoring at different frequencies.This complex structure makes it difficult to compare observationsat different epochs observed with different baseline distributions, thusdifferent angular resolution and different sensitivity to diffuse emis-sion. In particular, our 2016 observations were made with the fullVLBA array (10 antennas) while in the 2017 sessions two antennaswere missing (Los Alamos and Pie Town in March, St. Croix and PieTown in November). To have consistency in the components foundwith MODELFIT, it is important to minimize the effect of the differ-ent baseline distributions, especially at the lowest frequency wherethe detection of the diffuse emission is more relevant. Therefore at15 GHz, solely for the MODELFIT procedure, the UV planes of thedata were matched as much as possible, excluding 2 antennas fromthe 2016 data sets and cutting as needed the inner baselines.VLBA additional data at 15 GHz (7 epochs) from the MOJAVEproject (Lister et al. 2018) on 2019 and 2020 were added to the anal-ysis. The typical UV coverage of the MOJAVE data was significantlydifferent from ours, and generally better. Calibrated visibilities weredownloaded from the MOJAVE archive, imaged and model fittedagain in AIPS and DIFMAP.
MNRAS , 1–12 (2021)
R. Nesci et al.
Figure 3.
The radio map at 15 GHz on 4 September 2016 at 1.03 x 0.38 mas resolution from VLBA. Contours increase by a factor of 2. Figure 4.
The radio map at 43 GHz on 4 September 2016 at 0.50 X 0.23 mas resolution from VLBA. Contours increase by a factor of 2. In Fig. 3 we show our VLBA map at 15 GHz taken on 4 September2016, while in Fig. 4 we show our map at 43 GHz, which has ahigher resolution. Fig. 5 was built by us reprocessing all the publicMOJAVE data at 15 GHz of the years 2019-2020, stacking the sevenimages together, and shows the source morphology at a larger scale.Given the morphology of this radio source, the attempt to de-scribe the parsec scale emission with a limited number of discretecomponents gives a rough approximation of the real structure. Fur-thermore, the choice of the number of components to describe thesource is not trivial, especially when trying to compare the struc-ture at various epochs. Additional caution is needed to compare themorphology at different frequencies, as different angular resolution,different sensitivity to diffuse emission, and different intrinsic opac-ity may introduce shifts in the positions and delays in the estimatedproper motion.We decided to describe the radio source with a minimum numberof Gaussian components. Besides the core, we identified a componentA, very close to the core, which is detected only at 22 and 43 GHzin our data, but is well detected in the 15 GHz MOJAVE data taken4 years later; a component B, detected at all frequencies, definitely
Figure 5.
The radio map at 15 GHz at 0.77x0.65 mas resolution recomputedby us using MOJAVE data of the years 2019-2020. Contours increase by afactor of 2. Figure 6.
The separation between the core and the A (bottom), B (middle),and C (top) components as a function of time. X-axis in years, y-axis in mas;filled circles 15 GHz, open circles 43 GHz, open hexagons 22 GHz. 2015and 2016 data are from our VLBA and EVN observations, 2019-2020 datafrom the MOJAVE archive. Asterisks mark the epochs of the two large 𝛾 -rayflares. moving outwards, and a component C, apparently steady. Both Band C components were required to fit the MOJAVE 2020 data: ourC component corresponds to the Ca component in the Kun et al.(2018) and Roland et al. (2008) nomenclature. The uncertainty onour component positions is ∼ MNRAS000
The separation between the core and the A (bottom), B (middle),and C (top) components as a function of time. X-axis in years, y-axis in mas;filled circles 15 GHz, open circles 43 GHz, open hexagons 22 GHz. 2015and 2016 data are from our VLBA and EVN observations, 2019-2020 datafrom the MOJAVE archive. Asterisks mark the epochs of the two large 𝛾 -rayflares. moving outwards, and a component C, apparently steady. Both Band C components were required to fit the MOJAVE 2020 data: ourC component corresponds to the Ca component in the Kun et al.(2018) and Roland et al. (2008) nomenclature. The uncertainty onour component positions is ∼ MNRAS000 , 1–12 (2021) Figure 7.
The position in the sky plane of the B component at different timeswith respect to the core at coordinates 0,0. North is up and East to the left.Units are mas for both axis, with 0,0 indicating the core component. Filledcircles 15 GHz, open circles 43 GHz, open hexagons 22 GHz. those between 2015 and 2016, while those in the 2019-2020 rangecome from the MOJAVE archive. The epochs of the two large 𝛾 -rayflares are marked with asterisks located at the core position. In Fig.7 we report the positions in the plane of the sky, measured in masfrom the core, of component B at different epochs and frequencies:this plot allows to check that the position angle of the motion isconsistent with the jet axis, as apparent in Fig. 5. The presence ofa quasi stationary, more or less oscillating component at 1.4 masfrom the core (our component C), is in agreement with previousstudies on this radio source (see e.g. Roland et al. 2008), whilecomponents A and B can be regarded as new. Despite the difficultyin relating the components at different epochs with differing UVcoverage and observing frequencies, our overall results are consistentwith a scenario of a component B with a clear outward motion,with respect to the core component, along the jet direction, with adecreasing flux, and an additional component A separating from thecore more slowly and with less reliably estimated positions.We derived the proper motion of these components by means of alinear fit (see Fig. 8). We found that component A was moving with anapparent velocity v_app = (0.088 ± ± ± ± ± ± 𝛾 -ray flareon 19 July 2014.The OVRO light curve (see Fig. 1) shows a local maximum aroundthe beginning of November 2014 (MJD 56966), that may be relatedto the ejection of component B. However, the data points around thisperiod are poorly sampled, preventing us from estimating the peakin a more accurate way. The same argument applies to the poorlysampled data points in 2015. Year − . − . . . . . . . . S e p a t a t i o n ( m a s ) Figure 8.
The separation between the core and the A (bottom), and B (top)components as a function of time. X-axis in years, y-axis in mas; the regressionlines used to evaluate the ejection epochs of the two components are shown.For both components, extrapolation to y=0 are compatible with the times ofthe two large 𝛾 -ray flares. Filled circles 15 GHz, open circles 43 GHz, openhexagons 22 GHz. Table 5.
VLBI observations epochs. The array (column 1), the observationproject name (column 2), the date (column 3), the frequencies used (column4). array Project date freq.GHzEVN EC057A 14 Jun 2016 22VLBA BC224A 04 Sep 2016 15, 43VLBA BC224B 04 Mar 2017 15, 43EVN EC057B 08 Mar 2017 22VLBA BC224C 07 Nov 2017 15, 43EVN EC057C 14 Mar 2018 22VLBA MOJAVE 15 Aug 2019 15VLBA MOJAVE 08 Oct 2019 15VLBA MOJAVE 14 Nov 2019 15VLBA MOJAVE 15 dec 2019 15VLBA MOJAVE 04 Jan 2020 15VLBA MOJAVE 28 Feb 2020 15
Given the nature of the emission processes producing the spectralenergy distribution of a LSP blazar such as S5 1803+784, somecorrelations are expected between the low and the high frequencycomponents (see e.g. Röken et al. 2018 and references therein). Webriefly discuss here the results of our monitoring.
The historic optical light curve of S5 1803+784 since 1996 (see Fig.9) shows several flares, but without a well-defined periodicity: thetime interval between consecutive flares is not constant, nor similar,as can be seen from Fig.1 and Table 2. A formal test using the DiscreteFourier Transform (DFT) for unevenly spaced data (Deeming 1975)also gave no dominant frequency.In the 2008-2018 time interval the source was monitored also by
MNRAS , 1–12 (2021)
R. Nesci et al.
Figure 9.
The historic R 𝐶 light curve of S5 1803+784 from 1996 to 2018. the Fermi -LAT and had two major flares with a large (factor ∼ 𝛾 -ray light curve also does not show a clearperiodicity. The main parameters of optical flares are given in Table 2;the 𝛾 -ray ones in Table 3. We briefly describe below the main flares.The first flare was followed by us a few days after the 𝛾 -ray flare of11 January 2010 (MJD 55207) (Donato 2010), so the actual opticalpeak value is unknown and only the falling slope was measured.A remarkably long high state occurred between MJD 55892 and56040, peaking at R=14.4 mag; the final part is under-sampled, sothe dimming slope is not well constrained. A shorter flare occurredaround MJD 56206: the source was bright for about 60 days, butagain the dimming phase is not well sampled: apparently no 𝛾 -rayincrease was detected. A single bright optical point (R=14.5 mag)was detected by MASTER on 56639, corresponding to a bright 𝛾 -raylevel. A very bright flare occurred between MJD 56851 and 56899,apparently with a small precursor, and a brightening slope of 0.22mag/day. The source remained in a high activity state for about 45days, with some oscillations. The actual end was not covered by themonitoring, so the flare was likely much longer. Around MJD 57099we have just one photometric point, moderately bright (R=15.0 mag),from the Belgium Royal Observatory, corresponding to a high LATflux. Another very bright flare started at MJD 57248, apparently againwith a precursor, with a brightening slope of 0.11 mag/day, followedby an oscillating behaviour. The total duration was about 60 days.Two other minor optical flares occurred on 57348 and 57488, thislast one without a 𝛾 -ray counterpart. A long bright state was detectedbetween MJD 57719 and MJD 57747 peaking around R=14.4 mag,with a corresponding high 𝛾 -ray flux. Overall the bright states aretypically long-lasting (tens of days) while the rising slopes show arange of values from 0.07 up to 0.22 mag/day. The dimming phaseswere not well recorded but apparently have flatter slopes. A fairqualitative correlation between the optical and 𝛾 -ray light curves isapparent from Fig. 1, although the lack of a 𝛾 -ray counterpart for the"moderate" optical flares of October 2012 (MJD ∼ ∼ 𝛾 -ray peak with the optical one. In the second flare, the delay of the optical peak with respect to the 𝛾 -ray one is just 3 days, corresponding to our time bin, so it is notsignificant.No clear correlation is present between the 15 GHz total emission,as seen from the OVRO radio telescope, and the optical (and 𝛾 -ray)emission. This is in agreement with the previous finding by Nesci etal. (2002) based on the Medicina radio telescope at 8.4 GHz in theyears 1996-2002.Regarding the X-ray and optical emissions (see Table 4), the overallX-ray flux variation in the Swift pointings was a factor ∼
2, while thecorresponding optical luminosities changed by a factor ∼
6. Overall,no clear correlation emerges from the data, suggesting that the X-ray emission is not directly linked to the optical photons. This isexpected in a SSC scenario for a LSP BL Lac object, where theX-ray emission is considered to be produced by inverse Comptonof relativistic electrons over photons of the radio/submm range: theavailable X-ray data are however too sparse to look for a correlationbetween radio and X-ray emission.We also searched for a correlation between X-ray and 𝛾 -ray fluxesfor the available XRT pointings. To this purpose we re-binned theLAT data on 10-day windows centered on the XRT epochs, andsummed the XRT data taken at a few days distance: the results areshown in Fig.12. The range of flux variation in the 𝛾 -ray band wasabout a factor of 10, against a factor of only ∼ 𝛾 -ray frequencies.One may therefore expect that the X-ray flux is larger for larger 𝛾 -rayflux. Unfortunately, we have simultaneous XRT data only during the2015 𝛾 -ray flare, while no data are available for the 2014 one. Allthe other observations were performed with the source in a nearlyquiescent state. Our data in Fig.12 suggest that, in a quiescent state,there is, if any, an anti-correlation between X-ray and 𝛾 -ray fluxes,i.e. larger X-ray flux for lower 𝛾 -ray flux.For S5 1803+784 the position of the 𝛾 -ray peak is around 10 Hz, somewhat below the energy band where the LAT instrument ismost sensitive, and therefore not well measured. The LAT data tracethe falling branch of the emission, which is expected to be muchmore variable than the X-ray branch. Therefore the detection of acorrelation might require a higher accuracy in the flux measurementsfor both energy ranges.
MNRAS000
MNRAS000 , 1–12 (2021) Table 6.
Position of radio components with respect to the core: the componentnomenclature (column 1), year (column 2) distance from core (column 3),the position angle (column 4), the size (column 5), the flux density for VLBA(column 6) and frequency (column 7).comp year dist. PA size S freq.mas deg mas mJy GHzcore 2016.67 0.13 1740 15core 2017.17 0.12 1610 15core 2017.83 0.11 1720 15core 2019.62 0.08 1660 15core 2019.77 0.09 1530 15core 2019.87 0.10 1470 15core 2019.96 0.10 1370 15core 2020.01 0.10 1410 15core 2020.16 0.10 1430 15core 2020.33 0.12 1520 15A 2019.62 0.36 -79 0.27 181 15A 2019.77 0.39 -78 0.29 159 15A 2019.87 0.36 -76 0.41 206 15A 2019.96 0.40 -80 0.28 147 15A 2020.01 0.39 -79 0.32 176 15A 2020.16 0.39 -76 0.39 158 15A 2020.33 0.41 -80 0.39 146 15B 2016.67 0.36 -87 0.32 254 15B 2017.17 0.46 -79 0.41 264 15B 2017.83 0.57 -77 0.51 183 15B 2019.62 0.94 -76 0.31 148 15B 2019.77 0.98 -77 0.33 150 15B 2019.87 0.97 -76 0.33 147 15B 2019.96 0.98 -76 0.33 131 15B 2020.01 0.99 -77 0.35 139 15B 2020.16 1.01 -76 0.32 126 15B 2020.33 1.03 -77 0.31 102 15C 2016.67 1.32 -92 0.52 184 15C 2017.17 1.37 -93 0.62 213 15C 2017.83 1.35 -91 0.77 251 15C 2019.62 1.44 -88 0.31 130 15C 2019.77 1.46 -88 0.36 138 15C 2019.87 1.44 -87 0.36 157 15C 2019.96 1.44 -88 0.39 167 15C 2020.01 1.45 -88 0.38 171 15C 2020.16 1.45 -88 0.42 190 15C 2020.33 1.46 -88 0.45 197 15core 2016.45 0.09 - 22core 2017.19 0.09 - 22core 2018.21 0.08 - 22A 2016.45 0.16 -90 < .05 - 22A 2017.19 0.22 -66 0.14 - 22A 2018.21 0.27 -66 0.25 - 22B 2016.45 0.31 -76 0.38 - 22B 2017.19 0.51 -81 0.28 - 22B 2018.21 0.88 -74 0.14 - 22C 2016.45 1.27 -93 0.65 - 22C 2017.19 1.43 -91 0.25 - 22C 2018.21 1.29 -91 0.51 - 22core 2016.67 0.06 1110 43core 2017.17 0.05 992 43core 2017.83 0.06 1110 43A 2016.67 0.13 -46 0.09 144 43A 2017.17 0.13 -69 0.17 115 43A 2017.83 0.10 -126 0.14 80 43B 2016.67 0.36 -85 0.29 64 43B 2017.17 0.57 -78 0.36 61 43B 2017.83 0.72 -79 0.43 83 43C 2016.67 1.34 -92 0.37 83 43C 2017.17 1.39 -93 0.21 55 43C 2017.17 1.44 -92 0.45 90 43 MJD F l u x ( E > M e V ) [ − c m − s − ph ] F nu [ m J y ] (days) Figure 10.
Above: the 𝛾 -ray light curve in the 0.1-300 GeV energy range,binned at 1 day and 3 days intervals shown in blue open circle and black starrespectively, for the flare around MJD 56857. The upper limit data points arenot displayed. The optical light curve is over-plotted for comparison, withthe same colour codes of Fig.1. Below: the Discrete Correlation Functionbetween 𝛾 -ray and optical LC during the first flaring activity: x-axis is indays. The optical and UV spectral slopes showed very small variationsduring our monitoring. The near constancy of the optical and UVcolor indexes allowed us to more densely build our light curve, asdiscussed in Sections 2.1 and 2.4.The 𝛾 -ray spectrum of the source can be well fitted by a simplepower law 𝑁 𝜈 = 𝐴𝜈 − 𝑎 , where N is the number of photons for a givenfrequency: the photon index 𝑎 has an average value 2.26 ± Fermi -LAT (Ajello et al. 2020).We looked for a possible correlation of the photon index withthe source luminosity. Indeed we found a monotonic increase of thespectral index with flux, considering only observations with TS ≥ MNRAS , 1–12 (2021) R. Nesci et al.
MJD F l u x ( E > M e V ) [ − c m − s − ph ] F nu [ m J y ] ( d ays) Figure 11.
Above: the 𝛾 -ray light curve in the 0.1-300 GeV energy range,binned at 1 day and 3 days intervals shown in blue and open circle and blackstar respectively, for the flare around MJD 57257. The upper limits data pointsare not displayed. The optical light curve is over-plotted for comparison, withthe same color codes of Fig.1. Below: the Discrete Correlation Functionbetween 𝛾 -ray and optical LC during the second flaring activity: x-axis is indays. Positive values indicate 𝛾 rays leading. . . . . . Flux ( . − )[ erg cm − s − ] × − . . . . . E n e r g y F l u x [ − M e V c m − s − ] Figure 12.
The X-ray vs 𝛾 -ray flux for S5 1803+784 from nearly simultaneousSwift-XRT and Fermi -LAT data. input to the
Fermi -LAT software synthetic spectra with power lawindices of 1.5, 2.0, and 2.5, and total fluxes varying as in our source,with the same background as in our real data. Indeed we found thatbelow a flux level corresponding to about 1.0 x 10 − ph cm − s − ,the simulated spectra pointed out a systematic underestimation ofpower law index value. Therefore the apparent trend found in ourobservations of S5 1803+784 is likely spurious.Finally we built the average spectrum of the quiescent state, sum-ming all the observations with flux below 2.3 10 − ph cm − s − (ourflaring level defined in Section 2.3), and the average spectrum ofthe active state, summing all the observations above that level. Bothspectra were well described by a power law with the same spectralindex (2.22 ± ± Besides being useful to measure the redshift, and then the distanceof the source, the emission lines in a blazar can help us to understandwhere they are formed: close to the central engine, before the jet isfully developed, or beyond the jet, illuminated by its radiation. Inthe first case, if they are excited by the photons coming from theaccretion disk, their luminosity should remain unchanged and theirE.W. would drop significantly during a flare, which increases onlythe continuum emission. In the second case, the emission lines areexcited by the jet and their luminosity should follow the jet intensity,so their E.W. should remain somewhat constant during a flare.Our spectrum was taken near a flare peak, with the continuumemission a factor ∼ No definite periodicity was detected in our optical and 𝛾 -ray lightcurves of S5 1803+784. In recent studies Kun et al. (2018) a periodof about 8 years for the precession of the relativistic jet has beenclaimed, based on the analysis of VLBI images. From the data thatwe collected, there is no clear optical or 𝛾 -ray counterpart of thisphenomenon.Regarding the spectral variability, the optical (V,R,I) spectral indexwas rather constant during our monitoring, regardless of the sourceoptical luminosity, and very similar to that shown by the source inthe previous twelve years (Nesci et al. 2002). The X-ray spectralindex showed no evidence of hardening at higher fluxes. Also the 𝛾 -ray photon spectral index showed no significant variations correlatedwith the luminosity, in contrast to "softer when brighter" behaviourobserved in some cases (e.g PKS 2155-304, Foschini 2010 and PKS0219-164, Bhatta 2017), or a "harder when brighter" behaviour foundby Foschini et al. 2013 in the case of PKS 1510-089. The absence ofspectral evolution as a function of the luminosity could be related tothe location of the 𝛾 -ray emission region in the flaring state beyondthe emission line region. However, being a BL Lacertae object, S51803+78 is characterized by extremely weak emission lines, with asmall size of the emission line region which does not induce a strongabsorption of higher energy 𝛾 -ray photons and a steepening of thespectral index in flaring phase (Costamante et al. 2018).An optical spectrum taken during the maximum of the 2015 flaredid not show the emission line of MgII 2790Å, present when the MNRAS000
Fermi -LAT software synthetic spectra with power lawindices of 1.5, 2.0, and 2.5, and total fluxes varying as in our source,with the same background as in our real data. Indeed we found thatbelow a flux level corresponding to about 1.0 x 10 − ph cm − s − ,the simulated spectra pointed out a systematic underestimation ofpower law index value. Therefore the apparent trend found in ourobservations of S5 1803+784 is likely spurious.Finally we built the average spectrum of the quiescent state, sum-ming all the observations with flux below 2.3 10 − ph cm − s − (ourflaring level defined in Section 2.3), and the average spectrum ofthe active state, summing all the observations above that level. Bothspectra were well described by a power law with the same spectralindex (2.22 ± ± Besides being useful to measure the redshift, and then the distanceof the source, the emission lines in a blazar can help us to understandwhere they are formed: close to the central engine, before the jet isfully developed, or beyond the jet, illuminated by its radiation. Inthe first case, if they are excited by the photons coming from theaccretion disk, their luminosity should remain unchanged and theirE.W. would drop significantly during a flare, which increases onlythe continuum emission. In the second case, the emission lines areexcited by the jet and their luminosity should follow the jet intensity,so their E.W. should remain somewhat constant during a flare.Our spectrum was taken near a flare peak, with the continuumemission a factor ∼ No definite periodicity was detected in our optical and 𝛾 -ray lightcurves of S5 1803+784. In recent studies Kun et al. (2018) a periodof about 8 years for the precession of the relativistic jet has beenclaimed, based on the analysis of VLBI images. From the data thatwe collected, there is no clear optical or 𝛾 -ray counterpart of thisphenomenon.Regarding the spectral variability, the optical (V,R,I) spectral indexwas rather constant during our monitoring, regardless of the sourceoptical luminosity, and very similar to that shown by the source inthe previous twelve years (Nesci et al. 2002). The X-ray spectralindex showed no evidence of hardening at higher fluxes. Also the 𝛾 -ray photon spectral index showed no significant variations correlatedwith the luminosity, in contrast to "softer when brighter" behaviourobserved in some cases (e.g PKS 2155-304, Foschini 2010 and PKS0219-164, Bhatta 2017), or a "harder when brighter" behaviour foundby Foschini et al. 2013 in the case of PKS 1510-089. The absence ofspectral evolution as a function of the luminosity could be related tothe location of the 𝛾 -ray emission region in the flaring state beyondthe emission line region. However, being a BL Lacertae object, S51803+78 is characterized by extremely weak emission lines, with asmall size of the emission line region which does not induce a strongabsorption of higher energy 𝛾 -ray photons and a steepening of thespectral index in flaring phase (Costamante et al. 2018).An optical spectrum taken during the maximum of the 2015 flaredid not show the emission line of MgII 2790Å, present when the MNRAS000 , 1–12 (2021) source was in a much fainter state: this is expected if the emission lineis generated in a volume closer to the core than to the jet emitting thesynchrotron radiation, and therefore is not excited by the jet radiation.We found no correlation between X-ray and optical fluxes: thisis expected if the X-ray flux comes from the low-energy tail ofan Inverse-Compton emission process, while the optical emissioncomes from the high-frequency tail of the synchrotron radiation. Onthe contrary, the 𝛾 -ray emission showed a fair correlation with theoptical one, except in the case of some minor optical flares when no 𝛾 -ray enhancements were detected. In the two major flares, whichhave a better time sampling, the 𝛾 -ray flux rises simultaneously withthe optical one but falls a bit earlier, producing a narrower peak.The optical/ 𝛾 -ray flux ratio is expected to be rather constant in aSSC model: actually, the flux ratios listed in column 5 of Table 3show a mildly increasing trend with time, from less than 1 in 2009 toabout 3 in 2015, with correlation coefficient 0.84. The morphology ofthe source, traced at mas resolution by the VLBI radio observationscovering a span of several years after the large 𝛾 -ray-optical flares,shows a clearly detected component moving outwards after the firstflare, slowly decreasing in flux. Another similar component appearsto have been generated by the second flare, although with a smallerapparent velocity in the plane of the sky. This finding strongly sup-ports a causal connection between the mechanisms producing thehigh-energy and the radio-band radiation in blazars. ACKNOWLEDGEMENTS
First of all we want to thank S. Garrappa, S. Buson and D. Thomp-son for their insightful comments, which improved the manuscriptin its final form. We thank the
Neil Gehrels Swift Observatory
TimeAllocation Committee for the ToO observations of the source duringthe August 2015 flare.The European VLBI Network is a joint facility of independent Eu-ropean, African, Asian, and North American radio astronomy in-stitutes. Scientific results from data presented in this publicationare derived from the following EVN project code: EC057 and fromVLBA project code: BC224. The research leading to these resultshas received funding from the European Commission Horizon 2020Research and Innovation Programme under grant agreement No.730562 (RadioNet).The
Fermi -LAT Collaboration acknowledges generous ongoing sup-port from a number of agencies and institutes that have supportedboth the development and the operation of the LAT, as well as scien-tific data analysis. These include the National Aeronautics and SpaceAdministration and the Department of Energy in the United States;the Commissariat a l’Energie Atomique and the Centre National de laRecherche Scientifique/Institut National de Physique Nucléaire et dePhysique des Particules in France; the Agenzia Spaziale Italiana andthe Istituto Nazionale di Fisica Nucleare in Italy; the Centre Nationald’Études Spatiales in France; the Ministry of Education, Culture,Sports, Science and Technology (MEXT), High Energy Accelera-tor Research Organization (KEK), and Japan Aerospace ExplorationAgency (JAXA) in Japan; and the K. A. Wallenberg Foundation, theSwedish Research Council, and the Swedish National Space Boardin Sweden.Additional support for science analysis during the operations phaseis gratefully acknowledged from the Istituto Nazionale di Astrofisicain Italy and the Centre National d’Études Spatiales in France. Thiswork performed in part under DOE Contract DE-AC02-76SF00515.MASTER equipment is supported in part by Lomonosov MSU De-velopment Program. VL is supported by RFBR BRICS grant 17-52- 80133. VK is supported by RFBR grant 19-29-11011. RRL supportsMASTER-IAC observations.This research has made use of public data from the followingsources: the OVRO 40-m monitoring program (Richards, J. L. etal. 2011, ApJS, 194, 29) which is supported in part by NASAgrants NNX08AW31G, NNX11A043G, and NNX14AQ89G andNSF grants AST-0808050 and AST-1109911; the SIMBAD database,operated at CDS, Strasbourg, France; the NASA/IPAC ExtragalacticDatabase (NED) which is operated by the Jet Propulsion Laboratory(JPL), California Institute of Technology, under contract with theNational Aeronautics and Space Administration. This research hasmade use of data from the MOJAVE database that is maintained bythe MOJAVE team (Lister et al. 2018). DATA AVAILABILITY
All the data not already included in the Tables are available from theauthors on request.
REFERENCES
Abdo A. A., Ackermann M., Ajello M., Atwood W. B., Axelsson M.,Baldini L., Ballet J., et al., 2009, ApJ, 700, 597. doi:10.1088/0004-637X/700/1/597Abdollahi S., Acero F., Ackermann M., Ajello M., Atwood W. B., Ax-elsson M., Baldini L., et al., 2020, ApJS, 247, 33. doi:10.3847/1538-4365/ab6bcbAtwood W., Albert A., Baldini L., et al., 2013, arXiv, arXiv:1303.3514Ajello M., Angioni R., Axelsson M., Ballet J., Barbiellini G., Bastieri D.,Becerra Gonzalez J., et al., 2020, ApJ, 892, 105. doi:10.3847/1538-4357/ab791eAtwood W. B., Abdo A. A., Ackermann M., Althouse W., Anderson B.,Axelsson M., Baldini L., et al., 2009, ApJ, 697, 1071. doi:10.1088/0004-637X/697/2/1071Bhatta G., 2017, ApJ, 847, 7. doi:10.3847/1538-4357/aa86edBiermann P., Duerbeck H., Eckart A., Fricke K., Johnston K. J., Kuhr H.,Liebert J., et al., 1981, ApJL, 247, L53. doi:10.1086/183588Britzen S., Kudryavtseva N. A., Witzel A., Campbell R. M., Ros E.,Karouzos M., Mehta A., et al., 2010, A&A, 511, A57. doi:10.1051/0004-6361/20079267Chiaberge M., Celotti A., Capetti A., Ghisellini G., 2000, A&A, 358, 104Deeming T. J., 1975, Ap&SS, 36, 137. doi:10.1007/BF00681947Costamante L., Cutini S., Tosti G., Antolini E., Tramacere A., 2018, MNRAS,477, 4749. doi:10.1093/mnras/sty887Drake A. J., Djorgovski S. G., Mahabal A., Beshore E., Larson S., Gra-ham M. J., Williams R., et al., 2009, ApJ, 696, 870. doi:10.1088/0004-637X/696/1/870Donato D., 2010, ATel, 2386Foschini L., 2010, ATel, 2947Foschini L., Bonnoli G., Ghisellini G., Tagliaferri G., Tavecchio F., StamerraA., 2013, A&A, 555, A138 doi:10.1051/0004-6361/201321675Gehrels N., Chincarini G., Giommi P., Mason K. O., Nousek J. A., WellsA. A., White N. E., et al., 2004, ApJ, 611, 1005. doi:10.1086/422091Helene O., 1983, NIMPR, 212, 319. doi:10.1016/0167-5087(83)90709-3Lampens P., van Cauteren P., 2015, ATel, 7988Lawrence C. R., Zucker J. R., Readhead A. C. S., Unwin S. C., Pearson T. J.,Xu W., 1996, ApJS, 107, 541. doi:10.1086/192375Lipunov V., Kornilov V., Gorbovskoy E., Shatskij N., Kuvshinov D.,Tyurina N., Belinski A., et al., 2010, AdAst, 2010, 349171.doi:10.1155/2010/349171 , 1–12 (2021) R. Nesci et al.
Lister M. L., Aller M. F., Aller H. D., Hodge M. A., Homan D. C., KovalevY. Y., Pushkarev A. B., et al., 2018, ApJS, 234, 12. doi:10.3847/1538-4365/aa9c44Kun E., Karouzos M., Gabányi K. É., Britzen S., Kurtanidze O. M., GergelyL. Á., 2018, MNRAS, 478, 359. doi:10.1093/mnras/sty981Mattox J. R., Bertsch D. L., Chiang J., Dingus B. L., Digel S. W., EspositoJ. A., Fierro J. M., et al., 1996, ApJ, 461, 396. doi:10.1086/177068Monet D. G., Levine S. E., Canzian B., Ables H. D., Bird A. R., Dahn C. C.,Guetter H. H., et al., 2003, AJ, 125, 984. doi:10.1086/345888Nesci R., Massaro E., Maesano M., Montagni F., Sclavi S., Venturi T., Dal-lacasa D., et al., 2002, AJ, 124, 53. doi:10.1086/341038Nesci R., Maselli A., Montagni F., Sclavi S., 2012, AcPol, 52, 39Padovani P.,Richards J. L., Max-Moerbeck W., Pavlidou V., et al., 2011, ApJS, 194, 29.Giommi P., 1995, ApJ, 444, 567. doi:10.1086/175631Paiano S., Falomo R., Treves A., Scarpa R., 2020, MNRAS, 497, 94.doi:10.1093/mnras/staa1840Rector T. A., Stocke J. T., 2001, AJ, 122, 565. doi:10.1086/321179Röken C., Schuppan F., Proksch K., Schöneberg S., 2018, A&A, 616, A172.doi:10.1051/0004-6361/201730622Roland J., Britzen S., Kudryavtseva N. A., Witzel A., Karouzos M., 2008,A&A, 483, 125. doi:10.1051/0004-6361:20078521Li W., Filippenko A. V., Chornock R., Jha S., 2003, PASP, 115, 844.doi:10.1086/376432This paper has been typeset from a TEX/L A TEX file prepared by the author.MNRAS000