Low-Energy Neutrino Astrophysics with Super-Kamiokande

Core-Collapse Supernovae (CCSNe) are among the most powerful events in the cosmos where neutrinos play key role carrying away about 99% of the energy available. Despite the progress in the numerical studies the full details of the explosion mechanism are still unknown and only the novel observation of the neutrino burst from the next nearby CCSN will enable us to shed light on this phenomenon. The Super-Kamiokande(SK) experiment is an underground water Cherenkov neutrino detector consisting in a very large volume cylinder filled with ultrapure water (50 kton) readout by about 11,000 photomultipliers and being located under 1,000m overburden of rock (2,700 m.w.e.). SK is part of the Supernova Early Warning System (SNEWS), network which is coordinating most of the large-scale neutrino telescopes around the globe for the detection of such bursts, with a short latency. The current system at SK is designed to send normal warning if more than 25 events are detected within 20 seconds, which allows the efficient detection of the CCSNe up to the Small Magellanic Cloud. Offline analyses are also performed where only 7-8 events are required, allowing the extension of the search horizon up to few hundreds of kpc. The main aim of the Thesis consists in exploring the differences between the time profiles of the SN neutrino signal and the standard poissonian background trend, in order to lower the current multiplicity thresholds of the real-time monitor, providing fast alerts within fixed false positive rate (1 false alarm per century). To this purpose, new original statistical methods have been developed introducing additional cuts based not only on the absolute number of events but also on the characteristic time scale of the candidate SN clusters. The performances with different clustering algorithms have been tested as well.