The IceCube observatory, at the South Pole, includes an array of optical sensors deployed between 1450 and 2450 m below the surface and a surface array of cosmic ray detectors. The deep array, IceCube-InIce, is the world’s largest neutrino detector and consists of over 5000 optical sensors which detect the optical Cherenkov light emitted by the charged particles produced when neutrinos interact in the ice. Neutrinos are unique astronomical messenger particles as they can travel across the Universe without interference or being absorbed by matter. In contrast to charged particles, whose trajectories are affected by magnetic fields, neutrinos’ trajectories can be used to point back to their sources to aid the identification of particle acceleration sites.

Although the main aim of IceCube-InIce is to search for astrophysical neutrinos, the majority of events recorded are background events originating from the interaction of cosmic rays in the Earth’s atmosphere. In this thesis, the use of the IceTop surface array as a veto to identify background events is investigated. The focus of this thesis is the study of the implementation of an IceTop veto within an IceCube-InIce cascade analysis. A cascade analysis focuses on selecting IceCube-InIce events where the neutrino interaction has resulted in a shower or cascade of secondary particles rather than a long-range particle. The resulting pattern of light detected across the optical sensors appears globular rather than elongated and track-like. There are two variations of cascade analyses; contained and uncontained analyses. Contained cascade analyses concentrate on events where the central point of the light pattern is well-contained within the IceCube-InIce array while the uncontained analyses focus on the remaining events.

In this thesis it is shown that there is potential for an IceTop veto to be used in a way that allows other filter criteria to be relaxed in an uncontained cascade analysis in order to aid signal retention and for the veto to be used to remove additional high-energy background events. In order to link hits in the IceTop array with events in IceCube-InIce it is necessary to determine a time window within which hits will IceTop hits will be associated with IceCube- InIce event. The various factors which affect the choice of the time window length were investigated and it was found that the most limiting factor is the poor direction reconstruction of cascade events. A time window of 2000 ns was chosen and the performance of the IceTop veto with this time window was studied for simulated background events. In addition, the IceTop veto was applied to a sample of data events.