An island method of image cleaning for near threshold events from atmospheric Cerenkov telescopes

This paper describes a new strategy for cleaning of IACT images based on assessing the significance of islands of signal in the image. Simulations show that this method removes contamination from night sky background more effectively than the standard method and selects a greater number of genuine signal pixels leading to a more accurate parameterisation for low energy events. The pointing axis, ALPHA, is better determined for low energy events. For many events of 100-200 GeV, normal cleaning does not retain sufficient pixels for parameterisation to be performed. Island cleaning increases the number of parameterised events in this region by a factor of 3. Most of the improvement is observed at energies close to the current hardware trigger threshold where interference from the local ration population is very high. Consequently, only small improvements in real data are observed. The method is applicable to all IACT images and will be most beneficial for the new generation of telescopes which do not have the muon contamination. (C) 2003 Elsevier B.V. All rights reserved.