VERY-HIGH ENERGY OBSERVATIONS OF THE GALACTIC CENTER REGION BY VERITAS IN 2010–2012

The strong radio source Sgr A* located in the center of our galaxy is believed to coincide with a 4 × 10⁶ M_☉ black hole. While molecular clouds and dust hide the view toward the Galactic center at optical wavelengths, transient X-ray events with a 2–10 keV energy output up to 10³⁵ erg s⁻¹ are observed from Sgr A* on a regular basis (Degenaar et al. 2013; Neilsen et al. 2013; Barriere et al. 2014), as well as transient events at MeV/GeV energies (Vasileiou et al. 2011). Flares from X-ray binaries located in the Galactic center region can reach luminosities up to 10³⁷ erg s⁻¹ (Muno et al. 2005; Porquet et al. 2005; Sakano et al. 2005; Wijnands et al. 2006; Degenaar et al. 2012). Various astrophysical sources located close to the Galactic center are potentially capable of accelerating particles to multi-TeV energies, such as the supernova remnant Sgr A East or the pulsar wind nebula (PWN) G 359.95–0.04 (Wang et al. 2006).

A recently discovered gaseous object, G 2, heading toward the immediate vicinity of the Galactic center (Gillessen, et al. 2012) is expected to start merging into the black hole accretion stream some time in 2013–2014. This potential merger is a once-in-a-lifetime event that will allow observers to test magneto-hydrodynamical accretion models and their potential link to emission at the highest energies. Simulations show that the expected change in accretion and emission strongly depend on the origin/properties of the object (Abarca et al. 2014; Ballone et al. 2014; Saitoh et al. 2014) which have not yet been constrained well enough for accurate predictions. The merging process can potentially last for several decades and represents strong motivation for establishing a baseline for the γ-ray emission (as presented here) as well as for ongoing long-term monitoring of this region.

Observations of the Galactic center region also provide an avenue for dark-matter detection (Abramowski et al. 2011). Cold dark matter is widely viewed to be an essential component of the universe in our current standard cosmological model. A 100 GeV to TeV scale thermal relic with weak-scale interactions (or weakly interacting massive particle, WIMP) could provide the cold dark matter required to explain the observed structure in the universe as well as the matter density derived from cosmic microwave background measurements. However, to effectively search for a dark matter annihilation signal in the GeV/TeV regime, it is necessary to first understand the distribution, angular extent, and energy spectrum of the astrophysical sources near the Galactic center.

Several astrophysical sources located in the vicinity of the Galactic center can potentially emit γ-rays at MeV/GeV/TeV energies. Definite associations, on the other hand, are hampered by the limited angular resolution of instruments in these wave bands, ranging from ${\simeq }0.1 \, \deg$ at TeV energies to several degrees in the MeV regime. The EGRET MeV/GeV γ-ray source 3EG J1746-2851 is spatially coincident with the Galactic center (Hartman et al. 1999). More than one MeV/GeV source were resolved by the Fermi/LAT instrument (20 MeV < E ≲ 100 GeV) in the inner ${\sim }3 \deg$ region around the Galactic center (Abdo et al. 2010; Nolan et al. 2012), with the strongest source being spatially coincident with the Galactic center (see sky map in Section 3). Uncertainties in the diffuse Galactic background models and the limited angular resolution of the Fermi/LAT limits the ability to study the morphologies of these MeV/GeV sources in great detail.

At GeV/TeV energies a detection of a source coincident with the position of the Galactic center was first reported by the CANGAROO II collaboration which operated a ground-based γ-ray telescope in the southern hemisphere. The CANGAROO collaboration reported a steep energy spectrum dN/dE∝E^−4.6 (250 GeV < E ≲ 2.5 TeV) with an integral flux above 250 GeV at the level of 10% of the Crab Nebula flux (Tsuchiya et al. 2004). Shortly after, evidence for emission above 2.8 TeV from the Galactic center at the level of 3.7 standard deviations (s. d.) was reported from 1995 to 2003 large zenith angle (LZA) observations ($2.8 \, \rm {TeV} < {E} \lesssim 10 \, \rm {TeV}$) with the Whipple 10 m γ-ray telescope (Kosack et al. 2004). (See Section 2 for an explanation of LZA observations.)

Observations by H.E.S.S. in 2004–2006 confirmed the Galactic center as a GeV/TeV γ-ray source in the energy range of 100 GeV to several tens of TeV (Aharonian et al. 2004). The measured energy spectrum is described by a power law dN/dE∝E^−2.1 with a cutoff at ∼15 TeV (Aharonian et al. 2009). No evidence for variability was found in the H.E.S.S. or Whipple data over a time span of more than ten years. The difference between the energy spectrum measured by CANGAROO compared to the spectra measured by the other ground-based GeV/TeV instruments could only be explained if the different instruments observed different astrophysical sources in different states of activity or if the CANGAROO results were affected by a measurement error (see, e.g., Yoshikoshi et al. 2009).

Using the high-precision pointing system of the H.E.S.S. telescopes (reducing the pointing uncertainty to 6'' per axis), the position of the supernova remnant Sgr A East could be excluded as the source of the TeV γ-ray emission (Acero et al. 2010). A diffuse GeV/TeV γ-ray emission was identified after subtracting the point source located at the position of the Galactic center (Aharonian et al. 2006a). Its intensity profile is aligned along the Galactic plane (see sky map in Section 3) and follows the structure of molecular clouds. The energy spectrum of the diffuse emission is described by a power law dN/dE∝E^−2.3. It can be explained by an interaction of local cosmic rays (CRs) with the matter in the molecular clouds—indicating a harder spectrum and a higher flux of CRs in this inner region of the Galaxy as compared to the local CR spectrum (${d}N/ {d}E \propto E^{-2.7}$) observed at Earth. Recently, an additional, unresolved diffuse component of γ-ray emission was identified by the H.E.S.S. collaboration along the extended Galactic plane (Egberts et al. 2013). The MAGIC collaboration detected the Galactic center in 2004/2005 observations performed at large zenith angles at the level of 7 s. d. (0.5–10 TeV) (Albert et al. 2006), confirming the energy spectrum measured by H.E.S.S.

VERITAS first reported a >10 s. d. detection of the Galactic center in 2010 LZA observations, covering an energy range of 2.5 TeV to several tens of TeV (Beilicke et al. 2011). In this paper, we report on the results of three years (2010–2012) of VERITAS observations of the Galactic center region at large zenith angles at energies above ∼2.5 TeV. This paper focuses on the central TeV γ-ray source coincident with the Galactic center. The data were analyzed with the displacement method which substantially improves the angular resolution and sensitivity for data taken at large zenith angles (see Section 2). The VERITAS observations and results are discussed in Section 3. A discussion, comparison to models, and prospects of future GeV/TeV γ-ray observations of the Galactic center region are presented in Section 4. A study of the surrounding regions and the dark matter upper limit will be discussed in a second publication.

The stereoscopic method of shower reconstruction in ground-based atmospheric Cherenkov telescopes (such as VERITAS) is based on the intersection of the major axes of the parameterized Cherenkov images (Hillas) recorded in individual telescopes (Hofmann et al. 1999). In general, this method is very powerful since it makes use of the full capabilities of the stereoscopic recording of air showers. In the following this method is referred to as the geometrical method.

An alternative technique has been developed for data taken with single-telescopes (e.g., Whipple 10 m), using an estimate of the displacement parameter which is measured along the major axis of the image between the center of gravity of the Hillas ellipse and the shower position in the camera system (Buckley et al. 1998; Kosack et al. 2004; Domingo-Santamaria et al. 2005). For γ-ray showers the displacement parameter has a certain characteristic expectation value (derived from Monte Carlo simulations). Its value can be parameterized as a function of the Hillas parameters (Hillas 1985) of the corresponding image: the length l, the width w, and the amplitude s. Throughout this paper this method is referred to as the displacement method.

Large zenith angle observations are observations where the Cherenkov telescopes are pointed to low elevation angles, increasing the average distance to the detected showers. This results in a larger footprint for the Cherenkov lightpool (increasing the effective area), but a decrease of the Cherenkov light intensity also results in an increase of the energy threshold. The larger distance to the shower also decreases the parallactic displacement between images in the different telescopes. Moreover, given the large inclination angle, the angular separation of the telescopes projected into the shower plane are foreshortened in one dimension. The net effect is a strong reduction in the average stereo angle between the major axes of pairs of images, causing a large uncertainty in the determination of the geometrical intersection point. This, in turn, leads to a considerable reduction of the angular resolution in the reconstruction of the shower direction and impact parameter. The displacement method, on the other hand, does not rely on the intersection of axes, making it independent of the stereo angle between images. Therefore, no substantial drop in performance is expected with increasing zenith angle.

The displacement parameter as implemented in the VERITAS analysis (Aliu et al. 2012) was parameterized as a function of l, w, s, the zenith angle z, the azimuth angle Az, as well as the pedestal variance (a measure for readout noise fluctuations) of the image. In contrast to earlier realizations of the method, the parameterization is done in an orthogonal six-dimensional parameter space stored in the form of a lookup table that was trained with an extensive set of Monte-Carlo simulations of γ-ray showers. For each image, the displacement parameter is read from the lookup table and results in two most likely points of the shower direction with respect to the image center of gravity (CoG, in camera coordinates): $\rm {CoG} \pm \rm {displacement}$ along the major axis of the parameterized image. The combination of the points of all images involved in the event resolves the two-fold ambiguity.³⁰ The reconstruction of the shower impact parameter proceeds in a similar way, again making use of a multi-dimensional lookup table.

Figure 1 shows the angular resolution of both methods (geometrical and displacement) as a function of the cosine of the zenith angle z, derived from Monte Carlo simulations. While the angular resolution of the displacement method remains almost independent of cos (z), the angular resolution of the standard geometrical method becomes increasingly worse at large zenith angles. A further improvement is achieved if both methods are combined: d = d_geo · (1 − w') + d_disp · w', with the weight being calculated as w' = exp (− 12.5 · (cos (z) − 0.4)²) and w' = 1 for cos (z) < 0.4, respectively. Both methods benefit in similar ways from an additional requirement of N ⩾ 3 images in the event reconstruction. The method was applied to 6.5 hr of LZA Crab Nebula data (Figure 2). The data are in good agreement with the simulations and illustrate the clear improvement the displacement method provides in the case of LZA observations. The spectrum reconstructed from the Crab Nebula observations is shown in Figure 5 (in Section 3) and is found to be in reasonable agreement with the H.E.S.S. measurements obtained from lower zenith angles. The LZA Crab data set indicates an improvement in sensitivity of 30%–40 % when using the combined displacement/geometrical method compared to the geometrical method alone.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

VERITAS is an array of four 12 m diameter imaging atmospheric Cherenkov telescopes and is located at the base camp of the Fred Lawrence Whipple Observatory in southern Arizona at an altitude of 1280 m (Holder et al. 2008). VERITAS is sensitive to γ-rays in the energy range of 100 GeV to several tens of TeV. For observations close to zenith, sources of 10% (1%) of the strength and spectrum of the Crab Nebula are detected at the level of 5 s. d. in 0.5 hr (26 hr), respectively.

The Galactic center was observed by VERITAS in 2010–2012 for 46 hr (good quality data, dead-time corrected). Given the declination of the Galactic center, the observations were performed at large zenith angles in the range of z = 60.2°–66.4°, resulting in an average energy threshold (energy corresponding to the peak detection rate for a Crab-like spectrum) of E_thr ≃ 2.5 TeV. The shower direction and impact parameter were reconstructed with the geometrical/displacement method as described in Section 2. Other than using the displacement method, the standard analysis procedure was applied with event selection cuts a priori optimized for weak, hard-spectrum sources: angular separation between source position and reconstructed shower direction of $\Delta \theta ^{2} \le 0.012 \deg ^{2}$, mean scaled width/length ⩽1.04/1.25, and N ⩾ 3 images per event.

In the case of ground-based Cherenkov astronomy, the atmosphere acts as a calorimeter and its influence on the transmission of Cherenkov light is the single largest contributor to the systematic uncertainty in the estimate of the reconstructed TeV γ-ray energy E. The overall uncertainty for close-to-zenith observations is estimated to be on the order of ΔE/E ≃ 0.2, with an atmospheric contribution of about 0.15. For a spectrum dN/dE∝E^−2.5, this translates into an error in flux of ΔΦ/Φ ≃ 0.2. The column density of the atmosphere changes with 1/cos (z) and, conservatively, the systematic error in the energy/flux reconstruction can be expected to scale accordingly. For the Galactic center observations the contribution of the systematic effect induced by the atmosphere therefore roughly doubles as compared to low-zenith angle observations and we estimate the systematic error on the LZA flux normalization to be ΔΦ/Φ ≃ 0.4.

This section describes the results of the TeV γ-ray source coincident with the Galactic center. In the three-year data set, an excess on the order of 18 s. d. is found at the position of Sgr A* (see Figure 3). The background for the excess study (as well as for the energy spectrum, see below) was estimated from seven regions placed at the same radial camera distance as the source region (reflected background model, see Berge et al. 2007). The tail in the angular excess distribution can likely be explained by a contribution from the surrounding diffuse emission (Aharonian et al. 2006a) which becomes increasingly important at higher energies (Viana & Moulin 2013). More detailed studies on the diffuse emission morphology will be presented in a second paper.

Zoom In Zoom Out Reset image size

Figure 4 shows the VERITAS sky map of the Galactic center region. The background in this figure was estimated using a ring-like region (045 ⩽ r ⩽ 07) surrounding each test position (ring background model, see Berge et al. 2007), with a correction term taking into account the camera acceptance. Both background models exclude known sources from the background estimate (HESS J1741-302, HESS J1745-303, G 0.9+0.1, and the Galactic center itself). A fit of the point spread function to the uncorrelated excess sky map results in a centroid position of the excess in Galactic coordinates of $\rm {long} = (-0.077 \pm 0.006_{\rm {stat}} \,{\pm}\, 0.013_{\rm {sys}}) \deg$ and $\rm {lat} = (-0.049 \pm 0.003_{\rm {stat}} \pm 0.013_{\rm {sys}}) \deg$ with a fit quality of χ²/d.o.f. = 77.1/61. We name the VERITAS source VER J1745-290. This position is compatible with the Galactic center position (long = −0056 and lat = −0046) and the position measured by H.E.S.S. Both positions are indicated in Figure 4 which also shows the contour lines of the diffuse emission measured by H.E.S.S. (Aharonian et al. 2006a). Furthermore, Figure 4 shows the positions of the MeV/GeV sources taken from the 2FGL Fermi catalog (Nolan et al. 2012), as well as the contour lines of the 1–100 GeV diffuse emission after subtraction of point sources, extragalactic, and Galactic backgrounds (Yusef-Zadeh et al. 2013).

Zoom In Zoom Out Reset image size

The VERITAS energy spectrum obtained from the three-year data set is shown in Figure 5 and is found to be compatible with the spectra measured by Whipple (Kosack et al. 2005), H.E.S.S., and MAGIC. It can be described (χ²/dof = 2.1/4) by a power law dN/dE = I₀(E/5 TeV)^−Γ with a normalization at the decorrelation energy of 5 TeV of $I_{0} = (6.89 \pm 0.64_{\rm {stat}} \pm 2.75_{\rm {syst}}) \, 10^{-14} \, \rm {{\rm photons } } \, \rm {cm}^{-2} \, \rm {s}^{-1} \, \rm {TeV}^{-1}$ and a photon index of Γ = 2.57 ± 0.14_stat ± 0.2_syst. Since the LZA effective area of the VERITAS observations compensates for the shorter exposure (46 hr) as compared to the low-zenith H.E.S.S. observations (93 hr), the statistical errors of the VERITAS E > 2.5 TeV data points are comparable to those of the H.E.S.S. measurements. Recently, the H.E.S.S. collaboration reported an updated energy spectrum that was corrected for the energy-dependent contribution from the surrounding diffuse emission, leading to a lower cutoff energy around 10 TeV (Viana & Moulin 2013).

Zoom In Zoom Out Reset image size

The night-by-night integral fluxes above 2.5 TeV are calculated by folding a fixed spectral slope (derived from the energy spectrum for the full data set: dN/dE∝E^−2.6) with the effective area for the zenith angle of the corresponding night and comparing it with the excess counts above the threshold found in the data. The fluxes were binned according to observation periods of ∼3 week duration (as summarized in Table 1) and are shown in Figure 6 together with the yearly integral fluxes obtained from integrating the reconstructed energy spectra obtained for the individual years. No evidence for flux variability was found in the three-year data (the fit of a constant function to the run-by-run light curve (20 min segments) yields a fit quality of χ²/dof = 117/150). The H.E.S.S. collaboration reported a fit quality of χ²/dof = 233/216 as a result of a comparable study based on their 2004–2006 data set divided into 28 min segments (Aharonian et al. 2009).

Zoom In Zoom Out Reset image size

4.1. VERITAS Results in the Context of Multi-wavelength Data

The following potential counterparts are located³¹ within the 1' position uncertainty (statistical plus systematic) of the VERITAS excess: (1) the Galactic center Sgr A*, (2) the supernova remnant Sgr A East, (3) the PWN G 359.95–0.04, (4) the low mass X-ray Binary AX J1745.6−2901, (5) nine maser objects, and (6) about 150 X-ray sources. Here, (1)–(3) have the highest potential for TeV γ-ray emission. Sgr A East, however, was excluded as TeV counterpart by H.E.S.S. (Acero et al. 2010). PWN G 359.95−0.04 is discussed as potential counterpart in Section 4.2. In general, a contribution to the measured TeV γ-ray flux from more than one object cannot be excluded.

Sgr A* is known to exhibit 2–10 keV X-ray flares above the quiescent level on a regular basis, as for example observed in the 2006–2011 Swift monitoring campaign (Degenaar et al. 2013). A bright X-ray flare (flare #6, MJD 55359) was detected by Swift during the 2010 VERITAS monitoring but no TeV data were taken on that particular night. The time of the X-ray flare is indicated in Figure 6 and its spectrum is shown in the spectral energy distribution (SED) in Figure 7 together with a baseline measurement of the continuum emission from the extended region surrounding the Galactic center (including the contribution of Sgr A*). High spatial resolution X-ray observations were conducted in 2012 by Chandra (Neilsen et al. 2013) for a total of 3 Ms leading to the detection of 39 X-ray flares from the Galactic center with durations ranging from O(100 s) to O(8 ks). The observed flare luminosities in the 2–10 keV band reached 10³⁴ to $2 \times 10^{35} \, \rm {erg}\,\rm {s^{-1}}$. The time span of the Chandra campaign is indicated in Figure 6, where four nights had quasi-simultaneous coverage (same night) by VERITAS. Medium-intensity X-ray flares were detected in two out of those four nights at MJD 56047.7 and 56066.4. However, no significant flux changes were observed in the TeV band (see Section 4.4 for an estimate of the VERITAS sensitivity to detect variability). In an earlier campaign an X-ray flare was observed during joint H.E.S.S./Chandra observations in 2005; but no increase in TeV γ-ray flux was measured (Hinton et al. 2008).

Zoom In Zoom Out Reset image size

One possible origin of the observed X-ray flares is a change/disruption in accretion rate. In models where the TeV emission comes from particle acceleration near the black hole, one might expect some connection between the variation in the X-ray emission of the accretion and the TeV γ-ray production (see next section). The frequency of X-ray flares exceeding the quiescent state by a factor of 10 is estimated to be roughly one flare per day (Neilsen et al. 2013). The frequency of bright ($L_{\rm {X}} > 10^{35} \, \rm {erg}\,\rm {s^{-1}}$) X-ray flares is estimated to be 0.1–0.2 per day (Degenaar et al. 2013). Given (1) the sensitivity of the VERITAS LZA Galactic center observations, (2) the X-ray flare intensity and (3) flare frequency, it is challenging to correlate (with either direct or delayed response functions) the two wave bands on the basis of individual flares unless much stronger X-ray flares are observed (≫10 times the X-ray quiescent level). Most TeV emission models (see next section) predict a "smoothing out" of the accretion/flare activity in the TeV response, if there is any relationship at all.

Four more medium-intensity X-ray flares with durations of less than one hour and energies reaching up to 79 keV were detected in a NuSTAR campaign conducted in summer/fall 2012 (Barriere et al. 2014). Although the VERITAS 2012 observations had already ended by that time, the measured spectra of the two strongest flares J21₂ (07/2012) and O17 (10/2012) are shown in Figure 7 for reference.

At current times the emission level from the Galactic center is roughly 10 orders of magnitude below its Eddington luminosity (Terrier et al. 2010; Sunyaev et al. 1993). Spatial and temporal variations in the X-ray flux measured from molecular clouds surrounding the Galactic center have been found (Terrier et al. 2010; Ponti et al. 2010; Sunyaev et al. 1993). The results are interpreted as a bright $10^{39} \, \rm {erg}\,\rm {s^{-1}}$ outburst of a source coincident with Sgr A* that happened O(100) yr ago. These findings indicate long-term variations in accretion/brightness of the central source (Ponti et al. 2010). Recently, the Fermi/LAT instrument discovered two large γ-ray bubbles extending below and above the Galactic center (Su et al. 2010). Although the origin of the bubbles remains unclear so far, a significant increase in energy injection from the Galactic center on timescales of Myr is discussed as one of the possibilities (Su et al. 2010); for example, in the form of a plasma jet originating from the (previously brighter) active galactic nucleus in our galaxy (Yang et al. 2013; Guo et al. 2012).

4.2. Comparison to Models

4.3. Prospects for Dark Matter Limits

4.4. Prospects of Future VERITAS Observations

The implementation of the displacement method in the VERITAS data analysis chain has substantially improved the shower reconstruction and sensitivity for data taken at large zenith angles. It allows detection of the Galactic center at the level of 5 s. d. in roughly 3 hr with z > 60° observations. The measured energy spectrum is found to be in agreement with earlier measurements by H.E.S.S., MAGIC, and Whipple. At energies above 2.5 TeV the VERITAS measurements are competitive with H.E.S.S. Further constraints on emission models can be placed by future observations to measure the cutoff energy in the spectrum and to determine limits on the flux variability at the highest energies. The recently discovered gaseous object G 2 heading toward the immediate vicinity of the Galactic center black hole (Gillessen, et al. 2012) represents further motivation for future TeV γ-ray monitoring of this region. In addition to the potential for discoveries, the observations will establish a base line TeV γ-ray flux and spectrum that can be used to study possible changes caused by the merging process that can potentially last for several decades (Abarca et al. 2014; Ballone et al. 2014; Saitoh et al. 2014). An upper limit on diffuse γ-ray emission surrounding the Galactic center region and, in consequence, a limit on the photon flux initiated by the annihilation of dark matter will be presented in a separate publication.

This research is supported by grants from the U.S. Department of Energy Office of Science, the U.S. National Science Foundation, and the Smithsonian Institution, by NSERC in Canada, by Science Foundation Ireland (SFI 10/RFP/AST2748) and by STFC in the UK. We acknowledge the excellent work of the technical support staff at the Fred Lawrence Whipple Observatory and at the collaborating institutions in the construction and operation of the instrument.