4. Event reconstruction and data selection
The core position of an EAS is reconstructed independently from the data of the scintillator matrix and from AIROBICC where the latter data allow to tag core positions beyond the HEGRA boundary. If the core position lies inside the area covered with detector elements the scintillator derived core coordinates have a resolution of for protons (iron) at energies above 300 TeV (a little more accurate compared to AIROBICC mainly due to the smaller grid distances of the scintillator huts). The direction of the primary particle is reconstructed nearly independently from the scintillator and AIROBICC arrival time data (where the scintillator derived core position is used here).
The particle density measured by the scintillator array is fitted by the NKG formula (Greisen 1956) with a Moliére radius of 106 m, yielding the shower size and an age value. is a factor 1.6 larger than (the "true" shower size at detector level, denoting the number of charged particles above a kinetic energy of 3 MeV) due to coverage of the detector huts with a lead layer and the fact that the NKG function does not correctly parametrise the electromagnetic part of hadronic showers.
The dependence of the Cherenkov photon density , as measured by AIROBICC, on the distance r from the shower axis can be well described by an exponential in the region 20 m r 100 m (Patterson & Hillas 1983):
The parameter slope (in units of [1/m]) is the most important one in our analysis methods. As an illustration Fig. 1 shows the lateral charged-particle and Cherenkov-light distributions for a single shower.
The amplitude calibration of the scintillator array is done for samples of 50000 events by comparing the ADC spectra of the individual huts - which display a single peak essentially corresponding to the energy deposited by minimum ionising electrons and muons - with the result of MC simulations for identical conditions. The absolute amount of the air Cherenkov light registered by AIROBICC was calibrated by comparing the energy inferred from the lateral Cherenkov light density in the spectral range from 300 nm to 500 nm registered at a shower core distance of 90 m (referred to as in the following) and the energy derived from and slope in the interval (refer to Sect. 5.1 below for energy reconstruction methods). The absolute Cherenkov-light calibration thus depends on the CR mass composition, because we do not apply a primary-mass independent energy reconstruction here. We used the low-energy composition at 100 TeV as specified by Wiebel-Sooth et al.(1998) (60 light elements, see below Sect. 5.2 for details) for this calibration. If a pure proton (iron) composition is assumed the energy reconstructed from Cherenkov light alone is shifted by 3 (13)% to higher (lower) energies.
To select well-reconstructed events the EAS core positions and directions as reconstructed with AIROBICC and the scintillator matrix are demanded to be consistent. Additional cuts ensure the quality of the directional as well as the fits to the lateral particle and Cherenkov light density distributions. Events with the true shower-core position within the HEGRA array boundaries for the detector components used in this analysis (distance to edge of array 10 m) and a zenith angle below 150 are used for the further analysis. The efficiency to select EAS events with true core positions in the regarded 160160 m2 area is about 98% for primary energies above 300 TeV (independent of the primary mass). The contamination of the sample with EAS, where the true shower cores lay beyond the HEGRA boundary but which were erroneously reconstructed to fulfil the cuts is less than 1% from our simulations.
Nights with perfect weather conditions are selected by data of the Carlsberg Meridian Cycle (B.Argyle, priv.comm.) and by comparing the Cherenkov light measurements with data from the scintillators for samples of 50000 events (accumulated in about one hour with the used setup and trigger conditions). The data set solely contains nights without any technical problems of the used detector stations. In total it comprises (dead-time corrected) an on-time of 208 h. This corresponds to about 150000 events after all cuts with an energy above about 100 TeV and a zenith angle below 150.
© European Southern Observatory (ESO) 2000
Online publication: July 7, 2000