Simultaneous multi-wavelength observations of the TeV Blazar Mrk 421 during February - March 2003: X-ray and NIR correlated variability
Alok C. Gupta, B. S. Acharya, Debanjan Bose, Varsha R. Chitnis, Jun-Hui Fan
aa r X i v : . [ a s t r o - ph ] J a n Chinese Journal of Astronomy and Astrophysics manuscript no.(L A TEX: ms.tex; printed on November 5, 2018; 0:30)
Simultaneous multi-wavelength observations of theTeV Blazar Mrk 421 during February − March2003: X-ray and NIR correlated variability
Alok C. Gupta ⋆ , B. S. Acharya , Debanjan Bose , Varsha R.Chitnis and Jun-Hui Fan Center for Astrophysics, Guangzhou University, Guangzhou 510006, China Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba,Mumbai - 400 005, India
Abstract
In the present paper, we have reported the result of si-multaneous multi-wavelength observations of the TeV blazar Mrk421 during February − March 2003. In this period, we have ob-served Mrk 421 using Pachmarhi Array of ˇCerenkov Telescopes(PACT) of Tata Institute of Fundamental Research at Pachmarhi,India. Other simultaneous data were taken from the published lit-erature and public data archives. We have analyzed the high qual-ity X-ray (2-20 keV) observations from the NASA Rossi X-RayTiming Explorer (RXTE). We have seen a possible correlated vari-ability between X-ray and J band (1.25 µ ) near infrared (NIR)wavelength. This is the first case of X-ray and NIR correlated vari-ability in Mrk 421 or any high energy peaked (HBL) blazar. Thecorrelated variability reported here is indicating a similar originfor NIR and X-ray emission. The emission is not affected muchby the environment of the surrounding medium around the centralengine of the Mrk 421. The observations are consistent with theshock-in-jet model for the emission of radiations. Key words: galaxies: active - galaxies: blazars: general - galaxies:blazars: individual: Mrk 421
A small subgroup of radio-loud active galactic nuclei (AGNs) show significantflux variability in the complete electromagnetic (EM) spectrum, variable polar-ization and their radiation at all wavelengths is predominantly non-thermal. Theyare known as blazars, which is a collective name of subclasses (BL Lac objects,optically violent variables OVVs, high polarization quasars HPQs and flat spec-trum radio quasars FSRQs) of radio-loud AGNs. On a unified model of radio-loud AGNs based on the angle between the line of sight and the emitted jetfrom the source, blazars jet make angle of < ◦ from the line of sight (Urry ⋆ E-mail: [email protected]
Gupta et al. & Padovani 1995). Since, blazars emit radiation in the complete EM spectrumwhich gives an excellent opportunity to study the spectral energy distribution(SEDs). It is found from observations that blazars SEDs have two peaks. Thefirst component peaks any where from IR to optical in so called red blazars orlow energy blazars (LBLs) or radio selected blazars (RBLs) and at UV/X-ray inso called blue blazars or high energy blazars (HBLs) or X-ray selected blazars(XBLs). Its origin is synchrotron emission from high energy electrons in the jet.The second component extends up to γ − rays, peaking at GeV energies in RBLsand at TeV in XBLs. The electromagnetic emission is dominated by synchrotroncomponent at low-energy and at high-energy by inverse Compton component(Coppi 1999, Sikora et al. 2001, Krawczynski 2004).Mrk 421 is the nearest detected TeV BL Lac object (redshift z = 0.031). Itwas first noted to be an object with blue excess which later turned out to be anelliptical galaxy with bright point like nucleus (Ulrich et al. 1975). Since theenergy of synchrotron peak of the source is higher than 0.1 keV, it is classifiedas a HBL. Mrk 421 was the first extragalactic object discovered at TeV ener-gies (Punch et al. 1992). This source was later confirmed by the high energygamma ray astronomy (HEGRA) group (Petry et al. 1996). It is also one of theTeV blazars detected by energetic gamma ray experiment telescope (EGRET)instrument in the 30 MeV - 30 GeV energy range by the Compton gamma rayobservatory (CGRO) (Thompson et al. 1995). This source has been detected bythe other detectors like, the imaging Compton telescope (COMPTEL) on boardCGRO at the 3.2 σ level in the 10-30 MeV energy range (Collmar et al. 1999)and the solar tower atmosphere ˇCerenkov effect experiment (STACEE) in the 140GeV energy band (Boone et al. 2002).Mrk 421 variability has been studied in all EM regimes in isolation. An ex-haustive compilation of radio data at 22 and 37 GHz, spanning for about 25 years,for several extragalactic sources including Mrk 421 were reported by (Tar ¨ a srantaet al. 2004, 2005). NIR data for three decades, for several blazars including Mrk421, were given by (Fan & Lin 1999). A much systematic and comprehensivestudy of this source was done by Gupta et al. (2004) in the same period of thecampaign for which the present paper is written. In the compiled optical datafor long term observations, variation of 4.6 mag was reported by Stein et al.(1976) and rapid variability of 1.4 mag in 2.5 hours was reported by Xie et al.(1988). There are several simultaneous X-ray and gamma-ray as well as multi-wavelength campaigns for the source (Makino et al. 1987, Macomb et al. 1995,Takahashi et al. 2000, Katarzynski et al. 2003, Blazejowski et al. 2005).In the present paper, we aimed to search for correlated multi-wavelength vari-ability in Mrk 421. This kind of study will be an important tool for understandingthe emission mechanism of blazars. This paper is structured as follows. In sec-tion 2, we present multi-wavelength observations and data reduction; section 3and 4 give the results and discussions of the present work. We have used Pachmarhi Array of ˇCerenkov Telescopes (PACT), for observationof Mrk 421 in TeV gamma rays. PACT is located in Central India (latitude 22 ◦ ′ N, longitude 78 ◦ ′ E, altitude 1075 m ). We use wavefront sampling tech-nique to detect TeV γ -rays from astronomical sources. There are 24 telescopesspread over an area of m × m . Each telescope has 7 para-axially mountedparabolic mirrors (f/d ∼
1) of diameter 0.9m with a PMT (EMI 9807B) mountedat the focus of each mirror. Each telescope is independently steerable, controlled ariability of Mrk 421 3
Fig. 1
Space-angle distribution of source and background events for atypical run.remotely and monitored throughout the observation (Gothe et al. 2000). The en-tire array is sub-divided into 4 sectors with 6 telescopes in each. Each sectorhas its own data acquisition system (DAQ) where data on real time, relative ar-rival time of PMT pulses (using TDCs) and photon density (using ADCs) arerecorded. A Master DAQ at the center of the array is also used for recordinginformations of an event relevant to entire array. The 7 PMT pulses of a tele-scope are also linearly added to form a telescope pulse for trigger generation.Data recording is initiated when a coincidence of 4 out of any 6 telescope pulsesgenerates an event trigger for a sector. Typical trigger rate was about 1-3 Hz persector. The details of this array are given in (Bhat et al. 2000, Majumdar et al.2003).Observations are carried out in ON-OFF mode on clear moonless nights.In year 2003 from 26th February to 5th March there was a world-wide multi-wavelength campaign in several wavebands for this source including PACT.During these nights 2 sectors out of 4 were aligned along the source directionand remaining 2 were looking at a background region simultaneously. The back-ground region is chosen to be a dark region with the same declination as that ofthe source but with different RA. Background region is chosen in such a way thatthere is substantial overlap of zenith angle range between the source and back-ground runs. The typical run span was about 1-3 hours. The sectors that look atsource and background were interchanged on a daily basis.A number of preliminary checks were carried out on the data before doing ac-tual analysis. It was found that data taken on 26th and 27th February, 3rd and 5thMarch were very bad therefore rejected. Observations taken on 28th February,1st, 2nd and 4th March are analyzed. The arrival direction of each shower isdetermined by reconstructing shower front using the relative arrival times ofˇCerenkov photons at various telescopes (or PMTs). ˇCerenkov photon front isthen fitted with a plane, normal to this plane gives the direction of the showeraxis. Then, for each shower or event, this space angle is estimated as an anglebetween the direction of shower axis and the source direction. Thus space angles
Gupta et al. are obtained for all the events for source as well as background runs. Space an-gle distributions of source runs are compared with respective background runsover the same zenith angle region. Due to some technical problem same night’ssource and background runs could not be compared for these runs. Each sourcerun is compared with previous nights or next nights background run so that thegeometry of the telescope setup is also same for Source and Background runs.For this comparison, the shapes of space angle distributions in 2.5 ◦ to 6.5 ◦ regionof source and background were normalized, as we do not expect any signal be-yond 2.5 ◦ (Majumdar et al. 2003). Normalization of distributions correspondingto the background with that of the source is necessary as these two data sets weretaken at different times. Differences between the number of source and normal-ized background events within 2.5 ◦ gives the estimate of γ -ray events. Figure 1shows the space angle distributions for a typical pair of source and backgroundruns. Details of analysis procedure are given in (Bose et al. 2005). During thecampaign nights no excess of events over the background is detected in any ofthose four nights, implying γ -ray flux is close to or below the sensitivity limit ofPACT.ˇCerenkov photon showers initiated by γ -rays and protons were simulated us-ing CORSIKA air shower simulation (Heck et al. 1998) code to estimate triggerrate, energy threshold, collection area etc for the PACT setup. For γ -rays incidentvertically the energy threshold, defined as the peak of differential rate curve, isestimated to be 750 GeV and the corresponding collection area is 1.58 × m .For Mkn421, which is at an angle of 20 ◦ w.r.t. zenith the energy threshold isestimated to be 1.2 TeV and the collection area as 1.8 × m . We have analyzed Mrk 421 data observed with RXTE during 26/2/2003 -6/3/2003. We have extracted archival data sets corresponding to this multi-wavelength campaign under the guest observing program 80172. RXTE has twotypes of detectors viz, Proportional Counter Array (PCA) and High-Energy X-ray Timing Explorer (HEXTE) on-board along with All Sky Monitor (ASM).The PCA consists of five identical xenon filled proportional counter units (PCUs)covering an energy range of 2-60 keV. During these observations only PCU 0 andPCU 2 were used. Since PCU 0 lost pressure in the top veto at the beginning ofEpoch 5, we have used only data from PCU 2. HEXTE consists of two clustersof phoswich scintillation detectors covering an energy range of 15-250 keV, butis less sensitive. We do not discuss HEXTE data here. ASM consists of threexenon filled position sensitive proportional counters with field of view of 6 × ariability of Mrk 421 5 Near Infrared data in J band used in the present paper is taken from (Gupta etal. 2004). They have done observations from 1.2 meter optical/NIR telescopeat Gurushikhar observatory, Mount Abu, India using NICMOS-3 detector andJ band filter. The detail about NIR observations and data reductions is given in(Gupta et al. 2004).The radio data is taken from the recent paper by (Tar ¨ a sranta et al. 2005). Theyobserved the source during the campaign by their 17.7 meter Mets ¨ a hovi radiotelescope at 22 and 37 GHz. The detail about radio data is given in (Tar ¨ a srantaet al. 1998). Fig. 2
Multi-wavelength data of Mrk 421 as a function of time for allbands from radio to gamma-rays observed during February 25 − March05, 2003. Vertical lines in panel (c) − (g) show simultaneous variabilityin NIR and X-ray bands. In general uncertainties are smaller than thesymbols, the error bars have been omitted. Figure 2 gives the radio to gamma-ray light curves for the multi-wavelength cam-paign during February 25 - March 05, 2003. The data plotted here for differentbands of the EM spectrum is daily average. Daily average of a specific date is
Gupta et al.
Fig. 3
NIR and X-ray bands data of Mrk 421 as a function of time ob-served during February 26 − March 05, 2003.reported at 00h 00m 00s UT. The radio flux seems to be in stable state, impliesvariability timescale may be longer than the duration of the campaign. On theother hand gamma-ray data is noisy. The figure shows highly correlated variabil-ity among the different energy bands of the PCA data.Figure 3 gives the NIR and X-ray light curves (with 5 minutes binning) forthe observations during February 26 - March 05, 2003. X-ray coverage was muchlonger than the NIR coverage therefore we have selected that portion of X-raydata which was approximately simultaneous with NIR. For this plot PCA data of2 energy bands, 2-4 keV and 4-9 keV are combined.
We computed ZDCF (Z-transformed discrete correlation function) (Alexander1997) from light curves in X-ray and NIR bands. The ZDCF is a method fordetermining the cross-correlation function (CCF) of light curves in different en-ergy bands which have non-evenly sampled data. The ZDCF makes use of theFisher’s z-transform of the correlation coefficient. Fisher’s z-transform of thelinear correlation coefficient, r, is used to estimate the confidence level of themeasured correlation. This method attempts to correct the biases that affect theoriginal DCF (discrete correlation function) by using equal population binning.The ZDCF involves the following three steps:(i) All possible pairs of observations, (a i , b j ), are sorted according to their time- ariability of Mrk 421 7 Fig. 4
NIR and X-ray correlations. Positive lags imply that the secondlight curve lags the first.lag t i − t j , and binned into equal population bins.(ii) Each bin is assigned its mean time-lag and the intervals above and below themean that contain 1 σ of the each point.(iii) The correlation coefficients of the bins are calculated and z-transformed.The error is calculated in z-space and transformed back to r-space.The time-lag corresponding to the ZDCF is assumed as the time delay be-tween both components. This function is much more efficient in detecting anycorrelation present also it takes care of the data gaps. The ZDCF seems to peakat a negative lag − ∼ τ for each combination of light curves are as follows:(i) For PCA (2 − A max(ZDCF) = 0.530 +0 . − . , τ = − −
20 keV) vs 12500 ˚ A max(ZDCF) = 0.460 +0 . − . , τ = − − −
20 keV)max(ZDCF) = 0.747 +0 . − . , τ = − Gupta et al.
The visual inspection of the data in the figure 2 show, a strong correlation inNIR and X-ray bands at JD 2452697.5, 2452698.5, 2452701.5 and 2452702.5.In particular, the source has tendency to come to faint stage at JD 2452697.5and a flaring activity at JD 2452701.5. Thus the positive correlation is seenin both in flaring state and quiescent state. At JD 2452699.5, NIR has shownanti-correlation with PCA data (this anti-correlation is responsible for loweringthe correlation coefficients mentioned above). On March 01, 2003 observationscould not be taken in NIR J band due to bad weather condition. So, NIR data forJD 2452700.5 is not present in the panel (c) of the figure 2.
The SED of Mrk 421 is plotted in the figure 5 in the form log ν F ν vs. log ν . Allfrequencies used here are observed frequencies. The synchrotron component ofthe SED was fitted using NIR and X-ray data with a parabolic function y = Ax + Bx + C (1)the synchrotron peak frequency is determine by ν peak = − B/2A. The ν peak = Fig. 5
Spectral energy distribution of the Mrk 421. ariability of Mrk 421 9
We have found here the evidence of a possible correlation between X-ray andNIR wavelengths in Mrk 421 which is an HBL blazar.So far, correlated X-ray and NIR variation was observed in a LBL blazar 3C273, a flat spectrum radio quasar (FSRQ) (McHardy et al. 1999). For anotherLBL blazar AO 0235 + ∼ γ − ray flares in Mrk 421 in two different campaigns. In othercampaign of Mrk 421 in 1998 (Takahashi et al. 2000), complex variability, posi-tive and negative lags were found which authors report may not be real, If the lagfrom both signs are real, these imply that particle acceleration and X-ray coolingtimescales are similar. Katarzynski et al. (2003) show a well defined correlationbetween observed radio outburst in Mrk 421 with a corresponding X-ray outburstand a γ − ray flare in TeV range. In simultaneous TeV and optical observationsof blazar 1406 − ∼ γ − rayflaring events (Tosti et al. 1998 and references therein). Ghisellini & Maraschi(1996) have shown that the overall emission of the BL Lac Mrk 421 could beexplained using a homogeneous SSC model, in which the equilibrium particledistribution is found balancing continuum injection, cooling and particle escape.In this, single population of relativistic electrons emit synchrotron radiation upto the UV or X-ray band and soft photons upto the IR-optical bands undergoupscattering with the most energetic electrons to form TeV γ − rays. However,Blazejowski et al. (2005) have found TeV flares reached its peak days before X-ray flare during a giant flare or outburst in 2004. Spectral energy distribu-tion (SED) generated by Blazejowski et al. (2005) was not fitted with one-zonesynchrotron self-Compton (SSC) model but could be fitted well with additionalzones.Following Ghisellini & Maraschi (1996), Marscher (1996) and other authors,we note that the lags in the multi-frequency light curves of Mrk 421 (of the orderof hours between soft and medium X-ray photons, near simultaneous X-ray and γ − ray flares, IR leading X-rays by couple of days) require energy stratification inthe source. Similar delays were noticed in 3C 279 (IR leads X-ray by 0.75 ± −
304 (a day between EUV and X-rays, 2 days between UVand X-rays) and of the order of few hours between these (Edelson et al. 1995) aswell. Frequency stratification and different time scales for the duration of theseflares (shorter times for higher frequencies) are possible with the shock-in-jetmodel (Marscher 1996 and references therein).Within the context of the shock-in-jet model of AGNs, we attribute the NIRemission component to the internal shock driven into the jet by the variation ofthe central engine. The correlated X-ray and NIR variability with time lag of fewdays is a strong probe of the jet emission not affected much by the environmentof the surrounding medium. The anti-correlation seen in the NIR and soft X-raylight curve at JD 2452699.5 may be due to the reverse shock arising by the jet’scollision with the surrounding medium.
Acknowledgements
We thankfully acknowledge the referee for useful com-ments. The work of A. C. Gupta and J. H. Fan is supported by the NationalNatural Scientific Foundation of China (grant no. 10573005 and 10125313).We gratefully acknowledge the use of RXTE data from the public archive ofGSFC/NASA. Also, we are grateful to all the members of PACT collaborationfor their respective contributions.
References