MULTI-WAVELENGTH OBSERVATIONS OF THE FLARING GAMMA-RAY BLAZAR 3C 66A IN 2008 OCTOBER

VERITAS is an array of four 12 m diameter imaging Cherenkov telescopes in southern Arizona, USA (Acciari et al. 2008b). 3C 66A was observed with VERITAS for 14 hr from 2007 September through 2008 January and for 46 hr between 2008 September and 2008 November. These observations (hereafter 2007 and 2008 data) add up to ∼32.8 hr of live time after data quality selection. The data were analyzed following the procedure described in Acciari et al. (2008b).

As reported in Acciari et al. (2009), the average spectrum measured by VERITAS is very soft, yielding a photon index Γ of 4.1 ±0.4_stat ± 0.6_sys when fitted to a power law dN/dE ∝ E^−Γ. The average integral flux above 200 GeV measured by VERITAS is (1.3 ± 0.1) × 10⁻¹¹ cm⁻² s⁻¹, which corresponds to 6% of the Crab Nebula's flux above this threshold. In addition, a strong flare with night-by-night VHE-flux variability was detected in 2008 October. For this analysis, the VERITAS spectrum is calculated for the short time interval 2008 October 8–10 (MJD 54747–54749; hereafter flare period), and for a longer period corresponding to the dark run¹²⁴ where most of the VHE emission from 3C 66A was detected (MJD 54734–54749). It should be noted that the flare and dark run intervals overlap and are therefore not independent. Table 1 lists the relevant information from each data set.

As shown in Figure 1, the flare and dark run spectra are very soft, yielding nearly identical photon indices of 4.1 ± 0.6_stat ± 0.6_sys, entirely consistent with that derived from the full 2007 and 2008 data set. The integral flux above 200 GeV for the flare period is (2.5 ± 0.4) × 10⁻¹¹ cm⁻² s⁻¹, while the average flux for the dark run period is (1.4 ± 0.2) × 10⁻¹¹ cm⁻² s⁻¹. The extragalactic background light (EBL) de-absorbed spectral points for the dark run calculated using the optical depth values of Franceschini et al. (2008) and assuming a nominal redshift of z = 0.444 are also shown in Figure 1. These points are well fitted by a power-law function with Γ = 1.9 ± 0.5.

Zoom In Zoom Out Reset image size

The LAT on board the Fermi Gamma-ray Space Telescope is a pair-conversion detector sensitive to gamma rays with energies between 20 MeV and several hundred GeV (Atwood et al. 2009). Since launch the instrument has operated almost exclusively in sky survey mode, covering the whole sky every 3 hr. The overall coverage of the sky is fairly uniform, with exposure variations of ⩽15% around the mean value. The LAT data are analyzed using ScienceTools v9r15p5 and instrument response functions P6V3 (available via the Fermi science support center¹²⁵). Only photons in the diffuse event class are selected for this analysis because of their reduced charged-particle background contamination and very good angular reconstruction. A zenith angle <105° cut in instrument coordinates is used to avoid gamma rays from the Earth limb. The diffuse emission from the Galaxy is modeled using a spatial model ($\tt gll\_iem\_v02.fit$) which was refined with Fermi-LAT data taken during the first year of operation. The extragalactic diffuse and residual instrumental backgrounds are modeled as an isotropic component and are included in the fit.¹²⁶ The data are analyzed with an unbinned maximum likelihood technique (Mattox et al. 1996) using the likelihood analysis software developed by the LAT team.

Although 3C 66A was detected by EGRET as source 3EG J0222+4253 (Hartman et al. 1999), detailed spatial and timing analyses by Kuiper et al. (2000) showed that this EGRET source actually consists of the superposition of 3C 66A and the nearby millisecond pulsar PSR J0218+4232 which is 096 distant from the blazar. This interpretation of the EGRET data is verified by Fermi-LAT, whose improved angular resolution permits the clear separation of the two sources as shown in Figure 2. Furthermore, the known pulsar period is detected with high confidence in the Fermi-LAT data (Abdo et al. 2009a). More importantly for this analysis, the clear separation between the pulsar and the blazar enables studies of each source independently in the maximum likelihood analysis, and thus permits an accurate determination of the spectrum and localization of each source, with negligible contamination.

Zoom In Zoom Out Reset image size

Figure 2 also shows the localization of the Fermi and VERITAS sources with respect to blazar 3C 66A and radio galaxy 3C 66B (see caption in Figure 2 for details). It is clear from the map that the Fermi-LAT and VERITAS localizations are consistent and that the gamma-ray emission is confidently associated with the blazar and not with the radio galaxy. Some small contribution in the Fermi-LAT data from radio galaxy 3C 66B as suggested by Aliu et al. (2009) and Tavecchio & Ghisellini (2009) cannot be excluded, given the large spillover of low-energy photons from 3C 66A at the location of 3C 66B. This is due to the long tails of the Fermi-LAT point-spread function at low energies as described in Atwood et al. (2009). Nevertheless, considering only photons with energy E>1 GeV, the upper limit (95% confidence level) for a source at the location of 3C 66B is 2.9 × 10⁻⁸ cm⁻² s⁻¹ for the dark run period (with a test statistic¹²⁷ (TS) = 1.3). For the 11 months of data corresponding to the first Fermi-LAT catalog (Abdo et al. 2010a), the upper limit is 4.9 × 10⁻⁹ cm⁻² s⁻¹ (TS = 5.8).

As in the analysis of the VERITAS observations, the Fermi-LAT spectrum is calculated for the flare and for the dark run periods. The Fermi flare period flux F(E>100 MeV) = (5.0 ± 1.4_stat ± 0.3_sys) × 10⁻⁷ cm⁻² s⁻¹ is consistent within errors with the dark run flux of (3.9 ± 0.5_stat ± 0.3_sys) × 10⁻⁷ cm⁻² s⁻¹. In both cases, the Fermi-LAT spectrum is quite hard and can be described by a power law with a photon index Γ of 1.8 ± 0.1_stat ± 0.1_sys and 1.9 ± 0.1_stat ± 0.1_sys in the flare period and dark run intervals, respectively. Both spectra are shown in the high-energy SED in Figure 1.

3C 66A was observed by the Chandra observatory on 2008 October 6 for a total of 37.6 ks with the Advanced CCD Imaging Spectrometer (ACIS), covering the energy band between 0.3 and 10 keV. The source was observed in the continuous clocking mode to avoid pile-up effects. Standard analysis tools (CIAO 4.1) and calibration files (CALDB v3.5.0) provided by the Chandra X-ray center¹²⁸ are used.

The time-averaged spectrum is obtained and re-binned to ensure that each spectral channel contains at least 25 background-subtracted counts. This condition allows the use of the χ² quality-of-fit estimator to find the best-fit model. XSPEC v12.4 (Arnaud 1996) is used for the spectral analysis and fitting procedure. Two spectral models have been used to fit the data: single power law and broken power law. Each model includes galactic H i column density (N$_{\rm H, \rm Gal}=8.99\times 10^{20}$ cm⁻²) according to Dickey & Lockman (1990), where the photoelectric absorption is set with the XSPEC model phabs.¹²⁹ An additional local H i column density was also tried but in both cases the spectra were consistent with pure galactic density. Consequently, the column density has been fixed to the galactic value in each model, and the results obtained are presented in Table 2. An F-test was performed to demonstrate that the spectral fit improves significantly when using the extra degrees of freedom of the broken power-law model. Table 2 also contains the results of the F-test.

Following the VERITAS detection of VHE emission from 3C 66A, Target of Opportunity (ToO) observations of 3C 66A with Swift were obtained for a total duration of ∼10 ks. The Swift satellite observatory comprises an UV–Optical telescope (UVOT), an X-ray telescope (XRT), and a Burst Alert Telescope (Gehrels et al. 2004). Data reduction and calibration of the XRT data are performed with HEASoft v6.5 standard tools. All XRT data presented here are taken in photon counting mode with negligible pile-up effects. The X-ray spectrum of each observation is fitted with an absorbed power law using a fixed Galactic column density from Dickey & Lockman (1990), which gives good χ² values for all observations. The measured photon spectral index ranges between 2.5 and 2.9 with a typical statistical uncertainty of 0.1.

UVOT obtained data through each of six color filters, V, B, and U together with filters defining three ultraviolet passbands UVW1, UVM2, and UVW2 with central wavelengths of 260 nm, 220 nm, and 193 nm, respectively. The data are calibrated using standard techniques (Poole et al. 2008) and corrected for Galactic extinction by interpolating the absorption values from Schlegel et al. (1998) (E_B−V = 0.083 mag) with the galactic spectral extinction model of Fitzpatrick (1999).

The R magnitude of the host galaxy of 3C 66A is ∼ 19 in the optical band (Wurtz et al. 1996). Its contribution is negligible compared to the typical AGN magnitude of R ≲ 15; therefore, host-galaxy correction is not necessary.

GASP-WEBT. 3C 66A is continuously monitored by telescopes affiliated to the GLAST-AGILE support program of the Whole Earth Blazar Telescope (GASP-WEBT; see Villata et al. 2008, 2009). These observations provide a long-term light curve of this object with complete sampling as shown in Figure 3. During the time interval in consideration (MJD 54700–54840), several observatories (Abastumani, Armenzano, Crimean, El Vendrell, L'Ampolla, Lulin, New Mexico Skies, Roque de los Muchachos (KVA), Rozhen, Sabadell, San Pedro Martir, St. Petersburg, Talmassons, Teide (BRT), Torino, Tuorla, and Valle d' Aosta) contributed photometric observations in the R band. Data in the J, H, and K bands were taken at the Campo Imperatore observatory. A list of the observatories and their locations is available in Table 3.

Zoom In Zoom Out Reset image size

MDM. Following the discovery of VHE emission, 3C 66A was observed with the 1.3 m telescope of the MDM Observatory during the nights of 2008 October 6–10. A total of 290 science frames in U, B, V, and R bands (58 each) were taken throughout the entire visibility period (approx. 4:30 – 10:00 UT) during each night. The light curves, which cover the time around the flare, are presented in Figure 4.

Zoom In Zoom Out Reset image size

ATOM. Optical observations for this campaign in the R band were also obtained with the 0.8 m optical telescope ATOM in Namibia, which monitors this source periodically. Twenty photometric observations are available starting on MJD 54740 and are shown in Figures 3 and 4.

PAIRITEL. Near-infrared observations in the J, H, and K_s were obtained following the VHE flare with the 1.3 m Peters Automated Infrared Imaging Telescope (PAIRITEL; see Bloom et al. 2006) located at the Fred Lawrence Whipple Observatory. The resulting light curves using differential photometry with four nearby calibration stars are shown in Figure 4.

Keck. The optical spectrum of 3C 66A was measured with the LRIS spectrometer (Oke et al. 1995) on the Keck I telescope on the night of 2009 September 17 under good conditions. The instrument configuration resulted in a full width half-maximum of ∼250 km s⁻¹ over the wavelength range 3200–5500 Å (blue side) and ∼200 km s⁻¹ over the range 6350–9000 Å (red side). A series of exposures totaling 110 s (blue) and 50 s (red) were obtained, yielding a signal-to-noise (S/N) per resolution element of ∼250 and 230 for the blue and red cameras, respectively. The data were reduced with the LowRedux¹³⁰ pipeline and calibrated using a spectrophotometric star observed on the same night. Inspection of the 3C 66A spectrum reveals no spectral features aside from those imposed by Earth's atmosphere and the Milky Way (Ca H+K). Therefore, these new data do not offer any insight on the redshift of 3C 66A and in particular are unable to confirm the previously reported value of z = 0.444 (Miller et al. 1978).

Radio observations are available thanks to the F-GAMMA (Fermi-Gamma-ray Space Telescope AGN Multi-frequency Monitoring Alliance) program, which is dedicated to monthly monitoring of selected Fermi-LAT blazars (Fuhrmann et al. 2007; Angelakis et al. 2008). Radio flux density measurements were conducted with the 100 m Effelsberg radio telescope at 4.85, 8.35, 10.45, and 14.60 GHz on 2008 October 16. These data are supplemented with an additional measurement at 86 GHz conducted with the IRAM 30 m telescope (Pico Veleta, Spain) on 2008 October 8. The data were reduced using standard procedures described in Fuhrmann et al. (2008). Additional radio observations taken between 2008 October 5 and 15 (contemporaneous to the flare period) are provided by the Medicina, Metsähovi, Noto, and UMRAO observatories, all of which are members of the GASP-WEBT consortium.

The resulting multi-wavelength light curves from this campaign are shown in Figure 3 for those bands with long-term coverage and in Figure 4 for those observations that were obtained shortly before and after the gamma-ray flare. The VERITAS observations are combined to obtain nightly (E>200 GeV) flux values since no evidence for intra-night variability is observed. The highest flux occurred on MJD 54749 and significant variability is observed during the whole interval (χ² probability less than 10⁻⁴ for a fit of a constant flux).

The temporal dependence of the Fermi-LAT photon index and integral flux above 100 MeV and 1 GeV are shown with time bins with width of 3 days in Figure 3. For those time intervals with no significant detection, a 95% confidence flux upper limit is calculated. The flux and photon index from the Fermi-LAT first source catalog (Abdo et al. 2010a) are shown as horizontal lines for comparison. These values correspond to the average flux and photon index measured during the first 11 months of Fermi operations, and thus span the time interval considered in the figures. It is evident from the plot that the VHE flare detected by VERITAS starting on MJD 54740 is coincident with a period of high flux in the Fermi energy band. The photon index during this time interval is consistent within errors with the average photon index Γ = 1.95 ± 0.03 measured during the first six months of the Fermi mission (Abdo et al. 2010b).

Long-term and well-sampled light curves are available at optical and near-infrared wavelengths thanks to observations by GASP-WEBT, ATOM, MDM, and PAIRITEL. Unfortunately, radio observations were too limited to obtain a light curve and no statement about variability in this band can be made. The best sampling is available for the R band, for which variations with a factor of ≳2 are observed in the long-term light curve. Furthermore, variability on timescales of less than a day is observed, as indicated in Figure 4, and as previously reported by Böttcher et al. (2009) following the WEBT (Whole Earth Blazar Telescope) campaign on 3C 66A in 2007 and 2008.

The increase in gamma-ray flux observed in the Fermi band seems contemporaneous with a period of increased flux in the optical, and to test this hypothesis, the discrete correlation function (DCF) is used (Edelson & Krolik 1988). Figure 5 shows the DCF of the F(E > 1 GeV) gamma-ray band with respect to the R band with time-lag bins of 3, 5, and 7 days. The profile of the DCF is consistent for all time-lag bins, indicating that the result is independent of bin size. The DCF with time-lag bins of 3 days was fitted with a Gaussian function of the form $\rm DCF(\tau) = C_{\rm max}\times \exp {(\tau - \tau _{0})^{2}/\sigma ^{2}}$, where C_max is the peak value of the DCF, τ₀ is the delay timescale at which the DCF peaks, and σ parameterizes the Gaussian width of the DCF. The best-fit function is plotted in Figure 5 and the best-fit parameters are C_max = 1.1 ± 0.3, τ₀ = (0.7 ± 0.7) days and σ = (3.3 ± 0.7) days. An identical analysis was also performed between the F(E > 100 MeV) and the R optical band with consistent results. This indicates a clear correlation between the Fermi-LAT and optical energy bands with a time lag that is consistent with zero and not greater than ∼5 days. Despite the sparsity of the VERITAS light curve (due in part to the time periods when the source was not observable due to the full Moon), the DCF analysis was also performed to search for correlations with either the Fermi-LAT or optical data. Apart from the overall increase in flux, no significant correlations can be established. The onset of the E > 200 GeV flare seems delayed by about ∼5 days with respect to the optical–GeV flare but given the coverage gaps no firm conclusion can be drawn (e.g., the flare could have been already underway when the observations took place). No such lag is expected from the homogeneous model described in the next section but could arise in models with complex energy stratification and geometry in the emitting region.

Zoom In Zoom Out Reset image size

The broadband SED derived from these observations is presented in Figure 6 and modeled using the code of Böttcher & Chiang (2002). In this model, a power-law distribution of ultra-relativistic electrons and/or pairs with lower and upper energy cutoffs at γ_min and γ_max, respectively, and power-law index q is injected into a spherical region of comoving radius R_B. The injection rate is normalized to an injection luminosity L_e, which is a free input parameter of the model. The model assumes a temporary equilibrium between particle injection, radiative cooling due to synchrotron and Compton losses, and particle escape on a time t_esc ≡ η_esc R_B/c, where η_esc is a scale parameter in the range ∼250–500. Both the internal synchrotron photon field (SSC) and external photon sources (EC) are considered as targets for Compton scattering. The emission region is moving with a bulk Lorentz factor Γ along the jet. To reduce the number of free parameters, we assume that the jet is oriented with respect to the line of sight at the superluminal angle so that the Doppler factor is equal to D = (Γ [1 − β cos θ_obs])⁻¹ = Γ, where θ_obs is the angle of the jet with respect to the line of sight. Given the uncertainty in the redshift determination of 3C 66A, a range of plausible redshifts, namely z = 0.1, 0.2, 0.3, and the generally used catalog value z = 0.444, are considered for the modeling. All model fits include EBL absorption using the optical depth values from Franceschini et al. (2008).

Zoom In Zoom Out Reset image size

Most VHE blazars known to date are high synchrotron peaked (HSP) blazars, whose SEDs can often be fitted satisfactorily with pure SSC models. Since the transition from HSP to ISP is continuous, a pure SSC model was fitted first to the radio through VHE gamma-ray SED. Independently of the model under consideration, the low-frequency part of the SED (<10²⁰ Hz) is well fitted with a synchrotron component, as shown in Figure 6. For clarity, only the high-frequency range is shown in Figures 7 and 8, where the different models are compared. As can be seen from the figures, a reasonable agreement with the overall SED can be achieved for any redshift in the range explored. The weighted sum of squared residuals has been calculated for the Fermi-LAT and VERITAS flare data (8 data points in total) in order to quantify the scatter of the points with respect to the model and is shown in Table 4. The best agreement is achieved when the source is located at z ∼ 0.2–0.3. For lower redshifts, the model spectrum is systematically too hard, while at z = 0.444 the model spectrum is invariably too soft as a result of EBL absorption. It should be noted that the EBL model of Franceschini et al. (2008) predicts some of the lowest optical depth values in comparison to other models (Finke et al. 2010; Gilmore et al. 2009; Stecker et al. 2006). Thus, a model spectrum with redshift of 0.3 or above would be even harder to reconcile with the observations when using other EBL models.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 4. Parameters Used for the SSC Models Displayed in Figure 7

Model Parameter	z= 0.1	z= 0.2	z= 0.3	z= 0.444
Low-energy cutoff, γmin	1.8 × 104	2.0 × 104	2.2 × 104	2.5 × 104
High-energy cutoff, γmax	3.0 × 105	4.0 × 105	4.0 × 105	5.0 × 105
Injection index, q	2.9	2.9	3.0	3.0
Injection luminosity, Le (1045 erg s−1)	1.3	3.3	5.7	12.8
Comoving magnetic field, B (G)	0.03	0.02	0.02	0.01
Poynting flux, LB (1042 erg s−1)	1.1	4.9	8.5	13.7
B ≡ LB/Le	0.9 × 10−3	1.5 × 10−3	1.5 × 10−3	1.1 × 10−3
Doppler factor (D)	30	30	40	50
Plasmoid radius, RB (1016 cm)	2.2	6.0	7.0	11
Variability timescale, δtminvar (hr)	7.4	22.1	21.1	29.4
Weighted sum of squared residuals to VERITAS flare data	7.1	0.9	0.7	6.2
Weighted sum of squared residuals to Fermi-LAT flare data	1.6	1.6	1.3	1.4
Total weighted sum of squared residuals	8.7	2.5	1.9	7.6

Notes. All SSC models require very low magnetic fields, far below the value expected from equipartition (i.e., _B ≪ 1). The weighted sum of squared residuals to the VERITAS and Fermi-LAT data and the total value for the combined data set are included at the bottom of the table. The best agreement between the model and the data is obtained when the source is at redshift z = 0.2–0.3.

Download table as:  ASCIITypeset image

A major problem of the SSC models with z ≳ 0.1 is that R_B is of the order of ≳5 × 10¹⁶ cm. This does not allow for variability timescales shorter than ≲1 day, which seems to be in contrast with the optical variability observed on shorter timescales. A smaller R_B would require an increase in the electron energy density (with no change in the magnetic field in order to preserve the flux level of the synchrotron peak) and would lead to internal gamma–gamma absorption. This problem could be mitigated by choosing extremely high Doppler factors, D ≳ 100. However, these are significantly larger than the values inferred from VLBI observations of Fermi-LAT blazars (Savolainen et al. 2010).¹³¹ Moreover, all SSC models require very low magnetic fields, far below the value expected from equipartition (_B = L_B/L_e ∼ 10⁻³ ≪ 1), where L_B is the Poynting flux derived from the magnetic energy density and L_e is the energy flux of the electrons propagating along the jet. Table 4 lists the parameters used for the SSC models displayed in Figure 7.

Subsequently, an external infrared radiation field with ad hoc properties was included as a source of photons to be Compton scattered. For all SSC+EC models shown in Figure 8, the peak frequency of the external radiation field is set to ν_ext = 1.4 × 10¹⁴ Hz, corresponding to near-IR. This adopted value is high enough to produce E ≳ 100 GeV photons from IC scattering off the synchrotron electrons and at the same time is below the energy regime in which Klein–Nishina effects take place. Although the weighted sums of squared residuals for EC+SSC models are generally worse than for pure SSC models, reasonable agreement with the overall SED can still be achieved for redshifts z ≲ 0.3. Furthermore, all SSC+EC models are consistent with a variability timescale of Δt_var ∼ 4 hr. This is in better agreement with the observed variability at optical wavelengths than the pure SSC interpretation. Also, while the SSC+EC interpretation still requires sub-equipartition magnetic fields, the magnetic fields are significantly closer to equipartition than in the pure SSC case, with L_B/L_e ∼ 0.1. The parameters of the SSC+EC models are listed in Table 5.

Table 5. Parameters Used for the EC+SSC Model Fits Displayed in Figure 8

Model Parameter	z= 0.1	z= 0.2	z= 0.3	z= 0.444
Low-energy cutoff, γmin	5.5 × 103	7.0 × 103	6.5 × 103	6.0 × 103
High-energy cutoff, γmax	1.2 × 105	1.51.2 × 105	1.51.2 × 105	1.51.2 × 105
Injection index, q	3.0	3.0	3.0	3.0
Injection luminosity, Le (1044 erg s−1)	1.1	4.2	6.0	10.4
Comoving magnetic field, B (G)	0.35	0.22	0.21	0.23
Poynting flux, LB (1043 erg s−1)	1.0	2.4	6.0	11.2
B ≡ LB/Le	0.10	0.06	0.10	0.11
Doppler factor, D	30	30	40	50
Plasmoid radius, RB (1016 cm)	0.5	1.2	1.5	1.5
Variability timescale, δtminvar (hr)	1.7	4.4	4.5	4.0
Ext. radiation energy density (10−6 erg cm−3)	5.4	2.4	1.2	1.3
Weighted sum of squared residuals to VERITAS flare data	4.8	3.6	7.9	15.7
Weighted sum of squared residuals to Fermi-LAT flare data	1.0	1.2	0.8	1.5
Total weighted sum of squared residuals	5.8	4.8	8.7	17.2

Notes. These model fits require magnetic fields closer to equipartition and allow for the intra-night variability observed in the optical data. The weighted sum of squared residuals to the VERITAS and Fermi-LAT data and the total value for the combined data set are included at the bottom of the table.

Download table as:  ASCIITypeset image

Models with and without EC component yield the best agreement with the SED if the source is located at a redshift z ∼ 0.2–0.3. Of course, this depends on the EBL model used in the analysis. An EBL model that predicts higher attenuation than Franceschini et al. (2008) would lead to a lower redshift range and make it even more difficult to have agreement between the SED models and the data when the source is located at redshifts z ≳ 0.4. Finally, it is worth mentioning that the redshift range z ∼ 0.2–0.3 is in agreement with previous estimates by Finke et al. (2008), who estimate the redshift of 3C 66A to be z = 0.321 based on the magnitude of the host galaxy, and by Prandini et al. (2010) who use an empirical relation between the previously reported Fermi-LAT and IACTs spectral slopes of blazars and their redshifts to estimate the redshift of 3C 66A to be below z = 0.34 ± 0.05.

A detailed study of hadronic versus leptonic modeling of the 2008 October data will be published elsewhere, but it is worth mentioning that the synchrotron proton blazar (SPB) model has been used to adequately reproduce the quasi-simultaneous SED observed during the 2003–2004 multi-wavelength campaign (Reimer et al. 2008). On that occasion rapid intra-day variations down to a 2 hr timescale were observed, while during the 2008 campaign presented here these variations seem less rapid. Qualitatively, the longer timescale variations may be due to a lower Doppler beaming at the same time that a strongly reprocessed proton synchrotron component dominates the high energy output of this source.