ar X iv : a st ro - p h / 99 10 44 4 v 1 2 5 O ct 1 99 9 Data analysis techniques for stereo IACT systems

Based on data and Monte-Carlo simulations of the HEGRA IACT system , improved analysis techniques were developed for the determination of the shower geometry and shower energy from the multiple Cherenkov images. These techniques allow, e.g., to select subsamples of events with better than 3' angular resolution, which are used to limit the rms radius of the VHE emission region of the Crab Nebula to less than 1.5'. For gamma-rays of the Mrk 501 data sample, the energy can be determined to typically 10% and the core location to 2-3 m. Systems of imaging atmospheric Cherenkov telescopes (IACTs) for TeV gamma-ray astronomy allow the stereoscopic reconstruction of air showers, and provide improved angular resolution, energy resolution, and rejection of backgrounds such as showers induced by cosmic rays, local muons, or random triggers caused by night-sky background light. For systems with more than two telescopes, the shower parameters are overdetermined, allowing important cross-checks of the performance of the telescope system and of the reconstruction algorithms. In particular, the event-by-event determination of the position of the core permits to directly measure effective detection areas, and to estimate the systematic errors in the flux measurement [1,2]. In this talk, I will cover recent developments concerning improved algorithms to reconstruct shower direction and shower energy, and their tests using data from the HEGRA IACT system [3,2]. Detailed information as well as a more complete list of references can be found in [4–6]. Reconstruction of the shower geometry [4]. The traditional reconstruction algorithm used in HEGRA determines the shower direction by intersecting the axes of all Cherenkov images, regardless of the quality of individual images (Fig. 1(a)). In particular in events combining some bright images with dim images, the latter, with their poorly determined image parameters, can spoil the angular resolution. The angular resolution can be improved by estimating, for each image, the errors on the image parameters and by properly propagating these errors (Fig. 1(b)). In addition, one can use the shape of the image, in particular the width/length ratio, to estimate the distance d between the image centroid and the source, and