Development of reconstruction algorithms for large volume neutrino telescopes and their application to the KM3NeT detector

Abstract

Neutrino-astronomy is one of the most interesting frontiers of the astro-particle physics. Neutrino telescopes detect neutrinos indirectly through charged leptons produced in weak charged current interaction. In transparent media, tracks of relativistic particles can be reconstructed detecting light produced via Cherenkov effect, with 3D arrays of optical sensors. At present the most sensitive neutrino telescope in the world is the IceCube detector at the South Pole, that recently has reported evidence of a flux of cosmic high energy neutrinos exceeding the expected flux of atmospheric neutrinos by a statistically significant factor, de facto opening a new era of neutrino astronomy. KM3NeT will be the next-generation cubic kilometers scale neutrino telescope to be installed in the depths of the Mediterranean Sea. With several cubic kilometers of sea water instrumented with thousand of optical sensors, KM3NeT will be the most sensitive high energy neutrino telescope, with a sensitivity exceeding that of IceCube by a substantial factor. The construction of KM3NeT is based on a novel type of optical sensor, the Digital Optical Module, realised by arranging many small photomultipliers inside a pressure resistant glass sphere. This concept allows better performances with respect to the standard single large-area photomultiplier used in Antares and IceCube, since photon counting and directionality are possible due to photocatode segmentation. The aim of this work was to develop a new muon track reconstruction procedure appropriate for such a detector and evaluate the performances of the telescope through a complete Monte-Carlo simulation. Using this algorithm, the detector sensitivity (flux that can be excluded at 90% CL) and discovery potential (flux that can be detected at 5 sigma or 3 sigma above the background noise) for two galactic sources that appear to be the best candidates neutrino sources, the Supernova Remnants RXJ1713.7-3946 and Pulsar Wind Nebula Vela X, have been evaluated. The observation time required for the discovery of these sources at 5 sigma is about 5 year (2 years at 3 sigma) for the RXJ1713.7-3946 and about 3 years (1 year at 3sigma) for the Vela X. This analysis leads to the conclusion that at least the more intense galactic sources are at reach for KM3NeT. Recently it has been also proposed to exploit underwater Cherenkov neutrino telescopes to investigate the neutrino mass hierarchy by studying atmospheric neutrino oscillations at low energies (about 10 GeV). To perform such studies an as much as possible accurate determination of the neutrino energy and of the zenith angle are crucial. This requires a much denser array of photosensors with adequate containment conditions. First simulations show that with an effective volume of 3 Mton and an observation time of 5 years a significance of 3 sigma (5 sigma) can be achieved if the energy resolution is 25% (10%), assuming only contained events and a perfect knowledge of the muon zenith angle. To approach these assumptions an ad hoc reconstruction algorithm has been developed and is presented in this thesis. This algorithm allows to identify with an error of a few meters the interaction vertex that is used for the containment conditions. The error on the reconstructed zenith angle is less than 1 degree above about 8 GeV. The muon energy is reconstructed through the muon track length allowing to achieve a neutrino energy resolution of about 35% at 10 GeV. Further improvements are expected by including also the reconstruction of the hadronic shower, which will require a dedicated algorithm that has not yet been developed. The contamination due to electron and tau neutrino still needs to be considered.