Astron. Astrophys. 349, 11-28 (1999)

2. The HEGRA system of imaging atmospheric Cherenkov telescopes

2.1. The HEGRA Cherenkov telescope system

The VHE -ray observatory of the HEGRA collaboration consists of six imaging atmospheric Cherenkov telescopes (IACTs) located on the Roque de los Muchachos on the Canary island of La Palma, at 2200 m above sea level. A prototype telescope (CT1) started operation in 1992 and has undergone significant hardware upgrades since then. This telescope continues to operate as an independent instrument. The stereoscopic system of Cherenkov telescopes consists of five telescopes (CT2 - CT6), and has been taking data since 1996, initially with three and four telescopes, and since 1998 as a complete five-telescope system. Four of the telescopes (CT2, CT4, CT5, CT6) are arranged in the corners of a square with roughly 100 m side length, and one telescope (CT3) is located in the center of the square. During 1997, when the data discussed in this paper were taken, CT2 was still used as stand alone detector.

The telescopes have an 8.5 m² tessellated reflector, focusing the Cherenkov light onto a camera with 271 photomultipliers (PMTs), covering a field of view of in diameter. A telescope is triggered when the signal in at least two adjacent PMTs exceeds an amplitude of 10 (before June 1997) or 8 (after June 1997) photoelectrons; in order to trigger the CT system and to initiate the readout of data, at least two telescopes have to trigger simultaneously. Typical trigger rates are in the 10-16 Hz range. The PMT signals are digitized and recorded by 120 MHz Flash-ADCs. The telescope hardware is described by Hermann (1995); the trigger system and its performance are reviewed by Bulian et al. (1998).

2.2. Reconstruction of air showers with the HEGRA IACT system

The routine data analysis (see Paper 1 for details) includes a screening of data to exclude data sets taken at poor weather conditions or with hardware problems. In particular, the mean system trigger rate proved to be a sensitive diagnostic tool. Reconstruction of data involves the deconvolution of Flash-ADC data (Heß et al. 1998), the calibration and flat-fielding of the cameras, the determination of Hillas image parameters, and the reconstruction of geometrical shower parameters based on the stereoscopic views of the air shower obtained with the different telescopes (Daum et al. 1997, Aharonian et al. 1997a). The characteristic angular resolution for individual -rays is ; by sophisticated procedures -ray sources can be located with sub-arcminute precision (Pühlhofer et al. 1997).

The separation of hadronic and electromagnetic showers is based on the shape of the Cherenkov images, in particular using the width parameter. The width of each image is normalized to the average width of a -ray image for a given impact parameter of the shower relative to the telescope, and a given image intensity. Here, impact distances are obtained from the stereoscopic reconstruction of the shower geometry. Cuts are then applied to the mean scaled width obtained by averaging the scaled width values over telescopes. For a point source both the pointing information and the image shape information are used. Each of them provides a cosmic-ray background rejection of up to 100.

The reconstruction of the energy of air showers is based on the relation between the shower energy and the image intensity (size ) at a given distance from the shower axis (Aharonian et al. 1997c). This relation is tabulated based on Monte Carlo simulations, with the zenith angle of the shower as an additional parameter. The distance between a given telescope and the shower core is known from the stereoscopic reconstruction of the shower, with a typical precision of 10 m or less, for not too distant showers. The energy estimates from the different telescopes are then averaged, taking into account the slightly different sensitivities of the telescopes. These sensitivities are calibrated to 1% by comparing the light yield in two telescopes for events with cores halfway between the two telescopes (Hofmann 1997). According to Monte Carlo simulations, this energy reconstruction provides an energy resolution of 15% to 20%, depending on the selection of the event sample.

2.3. Monte Carlo simulations of air showers and of the telescope response

Any quantitative analysis of IACT data has to rely on detailed Monte Carlo simulations to evaluate the detection characteristics of the instrument.

The simulation of air showers and of Cherenkov light emission (Konopelko et al. 1999) includes all relevant elementary processes. On their trajectory to the detector, photons may be lost by ozone absorption, Mie scattering, and Rayleigh scattering. Atmospheric density profiles, ozone profiles and aerosol densities have been checked against local experimental data where available (e.g. Hemberger 1998).

On the detector side, the simulations include the wavelength dependence of the mirror reflectivity, of the light collection system, and of the PMT quantum efficiency. The point spread function of the mirror system is modeled after measurements of images of bright stars. The readout electronics is simulated in significant detail. PMT output waveforms are modeled by superimposing the response to single photoelectrons, with their relative timing and amplitude smearing. These signals are then sampled, quantitized, and fed into the same analysis path as regular Flash-ADC data. The simulation includes the measured saturation effects both in the PMT/preamplifier and in the Flash-ADC. Details concerning the Monte Carlo simulation used here, the performance of the system, and the comparison with experimental data are described by Konopelko et al. (1999).

© European Southern Observatory (ESO) 1999

Online publication: August 25, 1999
helpdesk.link@springer.de