1: \documentclass[12pt,preprint]{aastex}

2: 

3: %% My own definitions

4: \def\eg{\emph{e.g. }}

5: \def\et{\emph{et al.}}

6: \def\ie{\emph{i.e. }}

7: \def\cf{\emph{cf. }}

8: \def\apj{\emph{ApJ}}

9: \def\apjl{\emph{ApJ Letters}}

10: \def\apjs{\emph{ApJ Supp.}}

11: \def\aj{\emph{AJ}}

12: \def\aa{\emph{A \& A}}

13: \def\araa{\emph{ARA \& A}}

14: \def\pr{\emph{Phys. Rev.}}

15: \def\nat{\emph{Nature}}

16: \def\mnras{\emph{MNRAS}}

17: \def\x{$\times \, \,$}

18: \def\about{$\sim$}

19: 

20: \newcommand{\vdag}{(v)^\dagger}

21: \newcommand{\myemail}{carson@slac.stanford.edu}

22: 

23: \shorttitle{The Energy Spectrum of Mrk 421}

24: \shortauthors{Carson, \et}

25: 

26: \begin{document}

27: 

28: \title{The Energy Spectrum of the Blazar Markarian 421 \\

29: Above 130 GeV}

30: 

31: \author{J.E. Carson\altaffilmark{1,a}}

32: 

33: \author{J. Kildea\altaffilmark{2,b}}

34: 

35: \author{R. A. Ong\altaffilmark{1}}

36: 

37: \author{J. Ball\altaffilmark{1}}

38: 

39: \author{D. A. Bramel\altaffilmark{3,c}}

40: 

41: \author{C. E. Covault\altaffilmark{4}}

42: 

43: \author{D. Driscoll\altaffilmark{4}}

44: 

45: \author{P. Fortin\altaffilmark{5}}

46: 

47: \author{D. M. Gingrich\altaffilmark{6,7}} 

48: 

49: \author{D. S. Hanna\altaffilmark{2}} 

50: 

51: \author{T. Lindner\altaffilmark{2,d}}

52: 

53: \author{C. Mueller\altaffilmark{2}}

54: 

55: \author{A. Jarvis\altaffilmark{1}}

56: 

57: \author{R. Mukherjee\altaffilmark{5}}

58: 

59: \author{K. Ragan\altaffilmark{2}}

60: 

61: \author{R. A. Scalzo\altaffilmark{8,e}}

62: 

63: \author{D. A. Williams\altaffilmark{9}}

64: 

65: \author{J. Zweerink\altaffilmark{1}}

66: 

67: \altaffiltext{1}{Department of Physics and Astronomy, University of California, Los Angeles, Los Angeles, CA 90095}

68: \altaffiltext{a}{Current address: Stanford Linear Accelerator Center, MS 29, Menlo Park, CA 94025}

69: \altaffiltext{2}{Department of Physics, McGill University, Montreal, QC H3A 2T8, Canada}

70: \altaffiltext{b}{Current address: Fred Lawrence Whipple Observatory, Harvard-Smithsonian Center for Astrophysics, Amado, AZ 85645}

71: \altaffiltext{3}{Department of Physics, Columbia University, New York, NY 10027}

72: \altaffiltext{c}{Current address: Interactive Brokers, Greenwich, CT}

73: \altaffiltext{4}{Department of Physics, Case Western Reserve University, Cleveland, OH 44106}

74: \altaffiltext{5}{Department of Physics and Astronomy, Barnard College, Columbia University, New York, NY 10027}

75: \altaffiltext{6}{Department of Physics, University of Alberta, Edmonton, AB T6G 2G7 Canada}

76: \altaffiltext{7}{Also at TRIUMF, Vancouver, BC V6T 2A3 Canada}

77: \altaffiltext{d}{Current address: Department of Physics and Astronomy, University of British Columbia, Vancouver, BC V6T 1Z1, Canada}

78: \altaffiltext{8}{Department of Physics, University of Chicago, Chicago, IL 60637}

79: \altaffiltext{e}{Current address: Lawrence Berkeley National Laboratory, Berkeley, CA 94720}

80: \altaffiltext{9}{Santa Cruz Institute for Particle Physics, University of California, Santa Cruz, Santa Cruz, CA 95064} 

81: 

82: \begin{abstract}

83: 

84: Markarian 421 (Mrk 421) was the first blazar detected at gamma-ray

85: energies above 300 GeV, and it remains one of only twelve TeV blazars

86: detected to date. TeV gamma-ray measurements of its flaring activity

87: and spectral variability have placed constraints on models of the

88: high-energy emission from blazars. However, observations between 50

89: and 300 GeV are rare, and the high-energy peak of the spectral energy

90: distribution (SED), predicted to be in this range, has never been

91: directly detected. We present a detection of Mrk 421 above 100 GeV as

92: made by the Solar Tower Atmospheric Cherenkov Effect Experiment

93: (STACEE) during a multiwavelength campaign in early 2004. STACEE is a

94: ground-based atmospheric Cherenkov telescope using the wavefront

95: sampling technique to detect gamma rays at lower energies than

96: achieved by most imaging Cherenkov telescopes. We also outline a

97: method for reconstructing gamma-ray energies using a solar heliostat

98: telescope. This technique was applied to the 2004 data, and we present

99: the differential energy spectrum of Mrk 421 above 130 GeV. Assuming a

100: differential photon flux $\frac{dN}{dE} \propto E^{-\alpha}$, we

101: measure a spectral index $\alpha = 2.1 \pm 0.2_{stat} \, {^{+0.2}

102: _{-0.1}}_{sys}$. Finally, we discuss the STACEE spectrum in the

103: context of the multiwavelength results from the same epoch.

104: 

105: \end{abstract}

106: 

107: \keywords{Galaxies: BL Lacertae Objects: Individual: Name: Markarian

108: 421, Galaxies: Active, Gamma Rays: Observations}

109: 

110: \section{\label{intro}Introduction}

111: 

112: The X-ray-bright, nearby (z = 0.03) blazar Markarian 421 was the first

113: extragalactic object detected at very high energies (VHE, $E > 100$

114: GeV) \citep{punch92} and remains one of a dozen VHE detections of

115: blazars to date. Since its discovery, it has produced several major

116: flares, providing many opportunities for multiwavelength campaigns

117: (\eg \citet{gaidos96,krennrich99,piron01,aharonian02}). During these

118: periods of days to weeks, it outshines the Crab Nebula by factors of

119: $2 - 10$.

120: 

121: The gamma-ray spectrum of Mrk 421 above 300 GeV has been measured

122: several times at different flux levels, and the VHE spectral index

123: $\alpha$ is known to vary significantly (differential flux

124: $\frac{dN}{dE} \propto E^{-\alpha}$). \citet{krennrich02} and

125: \citet{aharonian02} made the important observation that the spectral

126: index and the absolute flux level are correlated, suggesting that the

127: high-energy peak of the spectral energy distribution (SED) shifts to

128: higher energies during periods of heightened activity. Also, above a

129: few TeV, a curved deviation from a constant power law spectrum has

130: been detected for Mrk 421 \citep{krennrich01}.

131: 

132: The simultaneous radio to VHE SED of Mrk 421 has been successfully

133: described in terms of synchrotron self-Compton (SSC) models

134: \citep{maraschi99,fossati00,krawczynski01,blazejowski05}. In these

135: models, synchrotron photons from relativistic electrons in the inner

136: regions of the magnetized jet gain energy via inverse-Compton

137: scattering off the same electron population that radiated them. The

138: synchrotron spectrum peaks in the X-ray band, while the

139: inverse-Compton photons are radiated at GeV and TeV energies. Within

140: this framework, the X-ray and TeV data have constrained the magnetic

141: field $B$ and the size $R$ of the gamma-ray emitting region to within

142: factors of a few: $B \sim 0.1$ Gauss and $R \sim 10^{16}$ cm. The

143: measurements are also beginning to constrain the Doppler factor of the

144: bulk material in the jet, $\delta_{jet} \sim 10-50$, and the shape of

145: the electron energy distribution.

146: 

147: Although well-studied at other energies, the flux of Mrk 421 between

148: 100 and 300 GeV has only been measured three times

149: \citep{boone02,piron03,albert06}, and only two measurements have been

150: made of its spectrum around 100 GeV

151: \citep{piron03,albert06}. Measurements in this waveband are of

152: particular scientific interest, because most SSC models predict that

153: the high-energy SED will peak around 100 GeV. Testing this prediction

154: can place valuable constraints on the various electron emission

155: models, especially in describing the energy distribution of the

156: relativistic electrons.

157: 

158: We report the detection of Mrk 421 above 100 GeV using the Solar Tower

159: Atmospheric Cherenkov Effect Experiment (STACEE, \citet{gingrich05})

160: and present a measurement of the integral flux and differential energy

161: spectrum between 130 GeV and 2 TeV. Section \ref{sec: observations}

162: briefly describes the STACEE experiment itself and the Mrk 421 data

163: set. Section \ref{sec: analysis} presents the basic data analysis, and

164: Section \ref{sec: int results} presents the integral flux

165: results. Section \ref{sec: spec results} describes the spectral

166: analysis and results, and Section \ref{sec: discussion} discusses the

167: scientific implications of the results.

168: 

169: 

170: \section{\label{sec: observations}STACEE Observations}

171: 

172: STACEE is a ground-based atmospheric Cherenkov telescope that utilizes

173: the wavefront sampling technique \citep{smith06} to measure gamma rays

174: above about 100 GeV. Gamma rays and cosmic rays striking the Earth's

175: atmosphere induce extensive air showers, and Cherenkov light is one

176: product of the showers. The Cherenkov light pool typically illuminates

177: an area approximately 300 m in diameter on the ground. In the

178: wavefront sampling technique as employed by STACEE, the Cherenkov

179: light is focused by heliostat mirrors -- large reflectors built

180: originally for solar energy research -- onto secondary mirrors and

181: then onto photomultiplier tubes (PMTs). The digitized PMT signals form

182: the raw data product. There is one PMT viewing each heliostat, so the

183: technique samples the Cherenkov wavefront at the location of each

184: heliostat. Figure \ref{fig: schematic} illustrates this technique. A

185: key feature of all atmospheric Cherenkov detectors is that the

186: effective collecting area for high-energy gamma rays is much larger

187: than the physical size of the mirrors used to collect the Cherenkov

188: light. Wavefront sampling detectors like STACEE, which use large

189: heliostats as opposed to focusing dishes, have larger mirror areas and

190: can therefore work at lower gamma-ray energies where Cherenkov photon

191: densities on the ground are lower.

192: 

193: STACEE uses 64 of the 220 heliostats at the National Solar Thermal

194: Test Facility in Albuquerque, New Mexico. These 64 reflectors are

195: distributed fairly evenly over an area of \about 150 m \x 240 m. Each

196: heliostat is composed of a 5 \x 5 array of square mirror facets, and

197: each facet has a collecting area of 1.5 m$^2$. The light from a

198: heliostat is reflected onto one of five secondary mirrors. Three large

199: secondaries (\about 2 m in diameter) focus light onto one of three

200: cameras holding 16 PMTs each, and two smaller secondaries (\about 1.1

201: m in diameter) focus light onto one of two smaller cameras holding

202: eight PMTs each. STACEE also operates a custom-made atmospheric

203: monitoring system during observations. This system includes a weather

204: station and a photometer for measuring atmospheric transmission.

205: 

206: The signal from each PMT is fanned out to a discriminator and to an

207: eight-bit flash analog-to-digital converter (FADC). The FADCs

208: continuously digitize the 64 PMT signals with a sampling rate of 1

209: sample/ns; a typical Cherenkov pulse is several nanoseconds wide, so

210: these digitizers record the shape of each pulse. The pulse shape

211: information from the FADCs allows precise pulse times and sizes to be

212: extracted. Relative pulse timing is used to reconstruct the direction

213: of the gamma ray, and pulse sizes are used to reconstruct gamma-ray

214: energy. A set of dynamic delays ensures that the 64 pulses remain

215: synchronized as the source is followed across the sky.

216: 

217: The Cherenkov light from air showers arrives on the ground as a

218: ``pancake'' a few nanoseconds thick, and over that brief interval it

219: is brighter than the night sky background. STACEE employs a two-level

220: trigger system and a tight timing coincidence requirement to pick out

221: Cherenkov-emitting air showers from the night sky glow. The trigger

222: system is designed as follows. First, the heliostats are grouped

223: geographically into eight clusters of eight heliostats each. The

224: cluster-level, or $L1$, trigger requires that the signals from five of

225: the eight heliostats within a cluster pass the discriminator

226: threshold, typically about four photoelectrons, within a timing window

227: of $8-24$ ns. Then, the array-level, or $L2$, trigger requires that

228: five of the eight clusters generate an $L1$ trigger within a 16-ns

229: window. When the $L2$ trigger condition is satisfied, the digitized

230: traces from all 64 PMTs are recorded.

231: 

232: The vast majority of the air showers detected by STACEE are initiated

233: by protons or light nuclei. In order to extract a gamma-ray signal

234: from data dominated by this cosmic-ray background, STACEE employs an

235: ``on-off'' observing strategy in which each 28-minute on-source

236: observation is paired with an equal-exposure off-source

237: observation. The off-source observation is at the same declination as

238: the on-source observation and is 30 minutes (7.5 degrees) away in

239: right ascension, so it covers the same trajectory in local sky

240: coordinates. The number of cosmic-ray events in the off-source

241: observation is subtracted from the number of on-source events to find

242: the statistical excess of gamma-ray events from the direction of the

243: source. Many on-off pairs of observations for a given source taken

244: over several weeks are used together to establish a gamma-ray signal

245: from that source. The observations of Mrk 421 presented here were

246: taken on 18 nights from January 29 to May 15, 2004. The total

247: on-source exposure time was 20.5 hours, with an equal amount of

248: off-source exposure time.

249: 

250: \section{\label{sec: analysis}Analysis}

251: 

252: A set of criteria was applied to the data set to select and remove all

253: periods of time with unreliable data. First, data taken during a

254: hardware malfunction were removed. These malfunctions included

255: mispointings of one or more heliostats, trips of the high-voltage

256: power to the PMTs, or failures of the data acquisition system. Second,

257: the weather station data were checked for evidence of frost, which can

258: accumulate on the heliostats and seriously degrade their

259: reflectivities. A total of 5.7 hours of data was removed due to

260: hardware malfunctions and frost.

261: 

262: Data were also removed if marred by changes in background light

263: levels, \eg from clouds drifting through the field of view or from

264: lights from cars or airplanes passing near the experiment. The

265: observable quantities used to identify these time periods were the

266: $L1$ trigger rates. For every observing run, the $L1$ rates were

267: examined by eye and were required to vary smoothly during a run (\ie

268: no sudden spikes or drops). The stability of the $L1$ and $L2$ rates

269: were verified in the post-cut data. This step eliminated an additional

270: 5.7 hours of on-source data. After all data quality cuts, the final

271: data set comprised 9.1 hours of on-source data taken on 16 nights.

272: 

273: Before the on- and off-source observations could be compared to

274: determine the statistical excess of gamma-ray events, an important

275: correction was made to account for the effects of the different

276: background light levels between the on- and off-source sky

277: fields. These differences are largely due to the particular stars in

278: those fields; a single bright star in one of the fields can radically

279: alter the light level. Low-energy cosmic-ray showers that are just

280: below the threshold required to initiate an event trigger can be

281: promoted above threshold by large positive fluctuations of the night

282: sky background. Likewise, showers that are just above threshold can be

283: demoted by large negative background fluctuations. If there were an

284: equal number of showers at all energies, then there would be the same

285: number of showers promoted above threshold as demoted below

286: threshold. However, because there are more low-energy showers than

287: high-energy ones (the cosmic-ray spectrum decreases with increasing

288: energy), the net effect of the night sky background fluctuations is to

289: promote subthreshold showers. This effect leads to a higher cosmic-ray

290: rate in observations with brighter sky fields. If the area of sky

291: covered by the on-source observation is brighter than that of the

292: off-source observation, as is the case for Mrk 421, then this effect

293: can imitate a true gamma-ray signal. 

294: 

295: A technique called ``padding'' \citep{scalzo04} has been developed to

296: equalize the backgroud light levels in a two-step process. The

297: technique relies on a pre-compiled library of FADC waveforms

298: corresponding to various low light levels measured under controlled

299: conditions. In the first step, background noise from this library is

300: added to ``pad'' the FADC waveforms of the observation with the dimmer

301: background. This step equalizes the RMS fluctuations, measured from

302: the 400 ns preceeding the pulse itself, between the on- and off-source

303: observations. In the second step, the trigger condition is reimposed

304: at a higher discriminator threshold. This is necessary because the

305: showers in the dimmer-background observation that \emph{would} have

306: triggered the detector if the light levels had been equal are absent

307: from the data stream. Imposing a software trigger identical to the

308: hardware trigger cannot regain these lost showers. Therefore, the

309: threshold is raised in software to the level where the

310: dimmer-background observations are not missing promoted showers

311: relative to the brighter-background observations. The technique is

312: fully described in \citet{scalzo04}, \citet{scalzophd04}, and

313: \citet{carson05}.

314: 

315: \section{\label{sec: int results}Integral Flux Results}

316: 

317: After applying the quality cuts and field brightness correction, there

318: are 2897 excess events in 511.3 minutes, for a photon count rate of

319: $5.7 \pm 0.9$ photons $\textrm{min}^{-1}$. The detection significance

320: is 5.9 standard deviations above the background. The mean significance

321: per observed pair is 1.4 $\sigma$ in 28 minutes of on-source exposure

322: time.

323: 

324: The detector response is characterized by the effective area

325: $A_{eff}$, defined as the overall gamma-ray collecting area of the

326: experiment, as opposed to the physical area of the heliostat

327: mirrors. It is determined from Monte Carlo simulations of the air

328: shower physics and the detector performance, as described in

329: \citet{scalzo04} and \citet{bramel05}. The effective area for the Mrk

330: 421 data set is shown in Figure \ref{fig: trig rate}. The data points

331: reflect simulations of observations at five different zenith angles,

332: exposure-weighted to reflect the composition of the real data set. The

333: solid line is a fit to the data assuming $A_{eff} = A_0 \left(1 -

334: e^{\frac{-(E-E_{min})}{E_r}}\right)$, where $A_0$, $E_{min}$, and

335: $E_r$ are constants.

336: 

337: The differential trigger rate $\frac{dR}{dE}$ (photons s$^{-1}$

338: GeV$^{-1}$) is $A_{eff}$ times the differential photon flux

339: $\frac{dN}{dE}$, expressed as a power law:

340: 

341: \begin{displaymath}

342: \frac{dN}{dE} = N_0 \,\, \left(\frac{E}{E_0}\right)^{-\alpha} \quad .

343: \end{displaymath}

344: 

345: \noindent Assuming a spectral index $\alpha = 2.0$, the curve of

346: $A_{eff} E^{-\alpha}$ for the Mrk 421 data set, which is proportional

347: to the differential trigger rate, is shown in the inset to Figure

348: \ref{fig: trig rate}.

349: 

350: Following the usual convention in VHE astronomy, we define the energy

351: threshold as the energy of the peak of the differential rate

352: curve. From Figure \ref{fig: trig rate}, the peak is at $E_{th} = 180$

353: GeV. It is clear from the figure that, like all VHE instruments,

354: STACEE has significant sensitivity to energies below $E_{th}$, as

355: defined in this way. The systematic error on the STACEE energy scale

356: is estimated to be \about 20\% \citep{scalzophd04}. Assuming $\alpha =

357: 2.0$, the integral flux of VHE gamma rays from Mrk 421 measured by

358: STACEE is $\Phi (E > 180 ~ \textrm{GeV} = (4.7 \pm 0.8_{stat}) \times

359: 10^{-10} \,\, \textrm{photons} \,\, \textrm{cm}^{-2} \,\,

360: \textrm{s}^{-1}$.

361: 

362: \section{\label{sec: spec results}Spectral Analysis and Results}

363: 

364: We have developed a technique for reconstructing gamma-ray energies

365: from the charges recorded on each of the 64 PMTs that constitute the

366: STACEE camera. The method relies on two physical properties of

367: gamma-ray showers: 1) the intensity of the Cherenkov light on the

368: ground is directly proportional to the energy of the initiating gamma

369: ray and 2) the light distribution on the ground is approximately

370: uniform out to the edges of the shower. These properties, along with

371: an independent estimate of the shower core location

372: \citep{scalzophd04}, allow us to convert a PMT charge to a number of

373: photons arriving at the corresponding heliostat. This conversion

374: relies on simulations of the STACEE electronics and optics that have

375: been thoroughly checked against calibration data \citep{scalzophd04,

376: carson05}. The reconstructed lateral photon distribution is then

377: converted to a gamma-ray energy using standard simulations of air

378: showers initiated by gamma rays. The energy reconstruction method

379: successfully reconstructs the energies of simulated gamma rays between

380: 130 GeV and \about 2 TeV with an energy resolution of $20-30\%$;

381: systematic errors in the energy estimate are less than $10\%$. The

382: details of the technique are described in \citet{carson05}. It was

383: applied to STACEE data taken on the Crab Nebula, the standard

384: calibration source for VHE astronomy, and the results are in good

385: agreement with previous measurements

386: \citep{hillas98,aharonian04,wagner05}.

387: 

388: The energy reconstruction method was also applied to the 2004 Mrk 421

389: data. The reconstructed energies were sorted into six energy bins:

390: $130-150$ GeV, $150-200$ GeV, $200-300$ GeV, $300-500$ GeV, $500-1000$

391: GeV, and $>1$ TeV. The bin choices were motivated by two

392: considerations: 1) ensuring that the effective area did not change

393: substantially over the bin, and 2) having a similar number of entries

394: in each bin. For each energy bin, the number of deadtime-corrected

395: entries in the off-source observations was subtracted from the number

396: of deadtime-corrected entries in the on-source observations, and the

397: gamma-ray excess in each energy bin was divided by the livetime for

398: the observations and the time-averaged effective area at that

399: energy. The resulting differential spectrum $\frac{dN}{dE}$ (photons

400: cm$^{-2}$ s$^{-1}$ GeV$^{-1}$) is shown in Figure \ref{fig: spectrum}.

401: Fitting the spectrum with the power law from Section \ref{sec: int

402: results} with $E_0 = 100$ GeV, we find

403: 

404: \begin{displaymath}

405: N_0 = (9.6 \pm 3.5_{stat} \, {^{+3.1} _{-2.0}}_{sys}) \times 10^{-12}

406: \quad \textrm{photons} \, \, \textrm{cm}^{-2} \, \, \textrm{s}^{-1} \,\, 

407: \textrm{GeV}^{-1}

408: \end{displaymath}

409: 

410: \noindent and

411: 

412: \begin{displaymath}

413: \alpha = 2.1 \pm 0.2_{stat} \, {^{+0.2} _{-0.1}}_{sys}  \quad .

414: \end{displaymath}

415: 

416: \noindent In the inset to Figure \ref{fig: spectrum}, the Mrk 421 SED

417: is plotted as $\nu F_{\nu} = E^2 \frac{dN}{dE}$ (ergs cm$^{-2}$

418: s$^{-1}$).

419: 

420: \section{\label{sec: discussion}Discussion}

421: 

422: As shown in the inset to Figure \ref{fig: spectrum}, the Mrk 421 SED

423: is flat in the STACEE band. The STACEE measurement suggests that the

424: high-energy peak of the SED of Mrk 421 was at or above 130 GeV (log

425: $\nu = 25.5$) in early 2004. This is a slightly higher energy for the

426: peak than predicted by most SSC models, such as the time-dependent SSC

427: model of \citet{krawczynski01}, which predicts a high-energy peak of

428: between 10 and 100 GeV. A shift of the high-energy peak above 100 GeV

429: is sometimes invoked in modeling of bright flares, however; for

430: instance, \citet{maraschi99} suggested that the peak shifted to

431: several hundred GeV during a flare in April 1998.

432: 

433: The STACEE observations of Mrk 421 were part of an extensive

434: multiwavelength campaign; during the first five months of 2004, the

435: source was monitored by radio, optical, X-ray (\emph{RXTE}), and TeV

436: (Whipple) telescopes simultaneously. These multiwavelength

437: measurements were presented in Figure 12 of \citet{blazejowski05} and

438: are summarized here. The data were divided into ``medium'' and

439: ``high'' states of activity based on their X-ray fluxes. It turned out

440: that this division separated the data into two time periods; the

441: medium state covers the period from January 27 to March 26 2004, and

442: the high state covers April $16-20$ 2004. The medium- and high-state

443: data were fit to one- and two-zone SSC models, respectively, where a

444: zone is a spherical emitting region characterized by a single magnetic

445: field and electron population. These models \citep{krawczynski04} are

446: snapshots of the SSC radiation emitted by a steady-state electron

447: population, and do not evolve the electron spectrum

448: self-consistently. Power law fits were performed on the Whipple data

449: assuming a high-energy cutoff of 4.96 TeV; these fits returned

450: spectral indices of $2.40 \pm 0.18$ and $2.11 \pm 0.14$ for the

451: medium- and high-state data, respectively. The STACEE data set covers

452: about 40\% of the nights represented in the multiwavelength SED

453: (Figure 12 of \citet{blazejowski05}), and \about 90\% of the STACEE

454: data set was taken during these nights. Approximately 25\% of the

455: STACEE data were from the TeV flaring state in April.

456: 

457: The STACEE spectral index is consistent with the high-state spectral

458: index determined by Whipple and marginally consistent with the

459: medium-state spectral index. However, it is naive to expect the

460: spectrum to follow a simple power law over such a large range in

461: energy, and the most useful comparison of the two data sets is

462: restricted to the energy range where they overlap, as shown in Figure

463: \ref{fig: tevcomp}. The open symbols are the Whipple data from the

464: high (diamonds) and medium (squares) states. The STACEE spectrum is

465: shown as asterisks. We note that the first STACEE energy bin is at 140

466: GeV, 120 GeV lower than the first Whipple energy bin. Figure \ref{fig:

467: tevcomp} shows that the STACEE and Whipple measurements are in good

468: agreement, especially considering that no cross-calibration was

469: performed on the data from the two instruments and that the errors

470: shown are statistical only. We expect the STACEE flux levels to

471: coincide more closely with the Whipple measurements in the medium

472: state, as seen, since the majority of the STACEE data were taken

473: before the April flare. The STACEE data show that the spectrum remains

474: approximately flat down to an energy of 130 GeV.

475: 

476: The most consistent interpretation of the combined STACEE and Whipple

477: data is that the peak of the SED is around $130-500$ GeV (log $\nu

478: \sim 25.5 - 26.1$). This is a higher peak energy than expected from

479: past modeling for a medium flux state, including the SSC model shown

480: in \citet{blazejowski05}. We note, however, that the

481: \citet{blazejowski05} medium-state model overpredicts the Whipple data

482: at the lowest energies, and a peak in the range indicated above is a

483: better reflection of the data than the one-zone model peak. A

484: detection of the peak indicates that we are measuring most of the

485: energy released in inverse-Compton cooling, and constrains the

486: inverse-Compton energy much more than a measurement of the falling

487: edge of the SED alone. Going a step further and combining the VHE data

488: with the X-ray data near the synchrotron peak then measures the total

489: energy budget of Markarian 421 and the relative contributions of

490: synchrotron and inverse-Compton cooling. During the period of these

491: observations, the combined STACEE and X-ray data suggest that the

492: synchrotron peak was a factor of 2-3 higher than the inverse-Compton

493: peak, suggesting that synchrotron cooling dominates. Because these

494: data represent a time average of the jet radiation, we suggest further

495: modeling that evolves the electron population as it cools via

496: synchrotron radiation.

497: 

498: During the preparation of this manuscript we learned of a measurement

499: of the Mrk 421 energy spectrum by the MAGIC collaboration

500: \citep{albert06}. The energy range covered by MAGIC includes much of

501: that explored by STACEE and they also report a flattening of the

502: energy spectrum at the low end. Their data are from a different epoch

503: (November 2004 to April 2005), so a direct comparison of data sets

504: from this highly episodic source is not easily made. However, it is

505: clear that many such observations will help with the detailed

506: understanding of Mrk421.

507: 

508: In summary, we have measured a spectrum of Markarian 421 above 130 GeV

509: in six energy bins. The STACEE data suggest that we are detecting the

510: high-energy peak of the SED for the first time above 100 GeV and that

511: the peak is around $130-500$ GeV. The direct detection of the

512: high-energy peak can strongly constrain future modeling.

513: 

514: \acknowledgments

515: 

516: We are grateful to the staff at the National Solar Thermal Test

517: Facility, who continue to support our science with enthusiasm and

518: professionalism. This work was supported in part by the National

519: Science Foundation, the Natural Sciences and Engineering Research

520: Council of Canada (NSERC), Fonds

521: Qu$\acute{\textrm{e}}$b$\acute{\textrm{e}}$cois de la Recherche sur la

522: Nature et les Technologies (FQRNT), the Research Corporation, and the

523: University of California at Los Angeles.

524: 

525: \begin{thebibliography}{}

526: 

527: \bibitem[Aharonian \et (2004)]{aharonian04} Aharonian, F.~ \et~ 2004,

528: \apj, 614, 897

529: 

530: \bibitem[Aharonian \et (2002)]{aharonian02} Aharonian, F.~ \et~ 2002,

531: \aa, 393, 89

532: 

533: \bibitem[Aharonian \et (2000)]{aharonian00} Aharonian, F.~ \et~ 2000,

534: \emph{New Astronomy}, 5, 377 

535: 

536: \bibitem[Albert \et (2006)]{albert06} Albert, J.~ \et~ 2006,

537: astro-ph/0603478, Submitted to \apj

538: 

539: \bibitem[Blazejowski \et (2005)]{blazejowski05} Blazejowski, M.~ \et~

540: 2005, \apj, 630, 130

541: 

542: \bibitem[Boone \et (2002)]{boone02} Boone, L.~M.~ \et~ 2002, \apjl, 

543: 579, L5

544: 

545: \bibitem[Bramel \et (2005)]{bramel05} Bramel, D.~A.~ \et~ 2005, \apj,

546: 629, 108

547: 

548: \bibitem[Carson (2005)]{carson05} Carson, J.~E.~ 2005, Ph.D. thesis, 

549: University of California, Los Angeles (unpublished)

550: 

551: \bibitem[Fossati \et (2000)]{fossati00} Fossati, G.~ \et~ 2000, \apj,

552: 541, 166

553: 

554: \bibitem[Gaidos \et (1996)]{gaidos96} Gaidos, J.~A.~ 1996, \nat,

555: 383, 319

556: 

557: \bibitem[Gingrich \et (2005)]{gingrich05} Gingrich, D.~M.~ \et~ 2005,

558: \emph{IEEE Trans. Nucl. Sci.}, NS-52(6), 2977-2985

559: %%astro-ph/0506613

560: 

561: \bibitem[Hillas \et (1998)]{hillas98} Hillas, A.~M.~ \et~ 1998, \apj,

562: 503, 744

563: 

564: \bibitem[Krawczynski \et (2004)]{krawczynski04} Krawczynski, H.~ \et~ 2004, 

565: \apj, 601, 151

566: 

567: \bibitem[Krawczynski \et (2001)]{krawczynski01} Krawczynski, H.~ \et~ 2001, 

568: \apj, 559, 187

569: 

570: \bibitem[Krennrich \et (1999)]{krennrich99} Krennrich, F.~ \et~ 1999,

571: \apj, 511, 149

572: 

573: \bibitem[Krennrich \et (2001)]{krennrich01} Krennrich, F.~ \et~ 2001, 

574: \apjl, 560, L45

575: 

576: \bibitem[Krennrich \et (2002)]{krennrich02} Krennrich, F.~ \et~ 2002, 

577: \apjl, 575, L9

578: 

579: \bibitem[Maraschi \et (1999)]{maraschi99} Maraschi, L.~ \et~ 1999, 

580: \apjl, 526, L81

581: 

582: \bibitem[Piron \et (2001)]{piron01} Piron, F.~ \et~ 2001, \aa,

583: 374, 895

584: 

585: \bibitem[Piron \et (2003)]{piron03} Piron, F.~ \et~ 2003, 

586: \emph{Proceedings of the 28th International Cosmic Ray Conference}

587: 

588: \bibitem[Punch \et (1992)]{punch92} Punch, M.~ \et~ 1992, \nat,

589: 358, 477

590: 

591: \bibitem[Scalzo (2004)]{scalzophd04} Scalzo, R.~A.~ 2004, Ph.D. thesis, 

592: University of Chicago (unpublished)

593: 

594: \bibitem[Scalzo \et (2004)]{scalzo04} Scalzo, R.~A.~ \et~ 2004, \apj,

595: 607, 778

596: 

597: \bibitem[Smith (2006)]{smith06} Smith, D.~A.~ 2006, astro-ph/0608251

598: 

599: \bibitem[Wagner \et (2005)]{wagner05} Wagner, R.~M.~ \et~ 2005,

600: \emph{Proceedings of the 29th International Cosmic Ray Conference},

601: Mumbai: Tata Institute of Fundamental Research, 4, 163

602: 

603: \end{thebibliography}

604: 

605: \clearpage

606: 

607: \begin{figure}

608: \epsscale{1.0}

609: \plotone{f1.eps}

610: \caption{A schematic representation of the wavefront sampling

611: technique. A gamma ray incident on the atmosphere induces an air

612: shower that illuminates an area approximately 300 m in diameter on the

613: ground with Cherenkov light. The light is reflected by heliostat

614: mirrors onto secondary mirrors and then focused onto PMTs mounted on

615: cameras under the secondaries. \label{fig: schematic}}

616: \end{figure}

617: 

618: \clearpage

619: 

620: \begin{figure}

621: \epsscale{1.0}

622: \plotone{f2.eps}

623: \caption{\emph{main figure:} The effective area, the overall

624: collecting area of the experiment, for the Mrk 421 data set. The data

625: points represent an exposure-weighted average over simulated

626: observations at five different zenith angles, and the solid line is a

627: fit to the form $A_0 \left(1-e^{\frac{-(E-E_{min})}{E_r}}\right)$ where

628: $A_0$, $E_{min}$, and $E_r$ are constants. \emph{inset:} The

629: (unnormalized) differential trigger rate as a function of energy for a

630: spectral index $\alpha = 2.0$. The energy threshold is defined as the

631: peak of this curve (180 GeV). \label{fig: trig rate}}

632: \end{figure}

633: 

634: \clearpage

635: 

636: \begin{figure}

637: \epsscale{1.0} \plotone{f3.eps}

638: \caption{\emph{main figure:} The Mrk 421 spectrum between 130 GeV and

639: 2 TeV, as measured by STACEE in early 2004. The spectral index is

640: $\alpha = 2.1 \pm {0.2_{stat}} \, {^{+0.2} _{-0.1}}_{sys}$. The

641: vertical error bars represent the statistical errors in both the

642: signal and the effective area. The horizontal bars represent the

643: widths of the energy bins. \emph{inset:} The same spectrum plotted

644: using the standard convention, as $\nu F_{\nu} = E^2 \frac{dN}{dE}$

645: (ergs cm$^{-2}$ s$^{-1}$) vs. frequency $\nu$. \label{fig: spectrum}}

646: \end{figure}

647: 

648: \clearpage

649: 

650: \begin{figure}

651: \epsscale{1.0}

652: \plotone{f4.eps}

653: \caption{The spectral energy distributions of Mrk 421 from Whipple

654: (open symbols) and STACEE (stars). The former are divided into medium

655: (squares) and high (diamonds) states of activity based on their X-ray

656: flux, as in \citet{blazejowski05}. A typical Whipple error bar is

657: shown in the upper left corner. \label{fig: tevcomp}}

658: \end{figure}

659: 

660: \clearpage

661: 

662: \end{document}

663: 

664: