In the first part of this work it was investigated whether it is possible to facilitate a direction reconstruction for cascade-like events. This would increase the impor-tance of the cascade detection channel with the possibility to perform point-source analyses.
In the course of this investigation the simulation of electromagnetic and hadro-nic cascades was improved. Instead of a point-like light emitter cascades are now treated more realistically by taking into account the longitudinal development.
This is a major improvement for electromagnetic cascades with energies above 10 PeV as the LPM effect increases the shower length significantly. At these en-ergies the energy loss profile cannot be parameterized due to strong fluctuations, hence a one-dimensional simulation of the shower development has been developed.
It incorporates parameterizations of bremsstrahlung and pair production cross sec-tions including the LPM suppression effect. The resulting electromagnetic shower profiles can have lengths of more than 100 m.
Based on the modified light yield of elongated cascades (see Figure 4.14) a direction reconstruction for cascade-like events was developed, which incorporates the elongation of cascades. For the vertex and energy reconstruction the perfor-mance of the algorithm is as good as existing algorithms. The aimed reconstruction of the cascade direction with better than 10◦ was not achieved within the avail-able time. However, positive indications were seen and the methods introduced will be incorporated in the IceRec software framework [110]. In particular the
“Deep-Core” extension of IceCube could improve the direction reconstruction due to denser module spacing. The decision to build this extension came too late to be included in the studies of this work, however, a dedicated analysis should be performed in the near future.
Consequently, without a direction reconstruction available, a diffuse flux analy-sis was developed to study the sensitivity of the completeIceCubedetector in the νe detection channel. The sensitivity was assessed introducing new cut variables that exploit the time structure of photons arriving at individual optical modules.
The resulting sensitivity reached within one year of data taking for an E−2 flux
φsens= 1.5−33% ·10 E GeV s sr cm .
Sensitivity is defined as the average upper limit that could be set in absence of a signal at 90 % confidence level. Without the use of sophisticated likelihood recon-structions the achieved sensitivity is already approximately one order of magni-tude better than a more evolved cascade analysis using five years of Amanda-II data [13].
At higher energies O(100 PeV) the sensitivity of theIceCube detector in the νe detection channel is limited by low expected fluxes of ultra-high-energetic neu-trinos and the finite detector volume. The quasi-differential limit derived in this work shows this performance decrease. The cascade channel is mostly restricted to the instrumented volume and only events within a maximum distance of∼300 m outside of the geometrical volume of the detector can contribute to the signal.
The effective volume reaches ∼0.9 km3. Existing Amanda-II analyses reach al-most half this effective volume, since events outside of the detector contribute.
Therefore, the improvements in the high energy region are not as good as in the 10 TeV to 10 PeV energy range.
The comparison with results from other experiments like Auger and Anita-lite is more difficult. Of course experiments with larger detection volumes are superior at the highest energies. However, for comparisons in the same energy re-gion, the assessment is subtle due to different detection channels and background contributions. With respect to many models, IceCube would cover lower ener-gies, whereas Auger and Anita-litewould cover the higher energy part of the spectrum. They are therefore complementing each other.
For the detection of GZK neutrinos most likely much larger detectors are re-quired. Current development efforts address the possibility of radio and acoustic detection of high energy neutrino events [41]. A simulation of a detector instru-menting an area of 10×10 km2 with radio and acoustic sensors, presented in [36], yields∼20 GZK neutrino events per year assuming the flux predicted by [52]. Fol-lowing the analysis in this workIceCube would detect at most one event within ten years of the same flux.
Improvements of the Analysis
Despite the improvements already achieved in this work, there are several things that can be optimized before the analysis is applied to data. The limited back-ground statistics have to be increased and compared to a fraction of data taken with the detector in order to justify the filtering and classification based on simulated events. It is probably not necessary to use a dedicated high-energy background sample since the sensitive energy range is within the range where the standard Poly-Gonato background simulation provided by the collaboration is valid, i.e.
below ∼100 PeV. However, a small high-energy sample should be used to study
background contributions from cosmic rays with energies above ∼100 PeV. The signal simulation should be extended to include theνµandντ interactions as these can also be classified as cascade-like due to the hadronic cascade in the final state of the neutrino-nucleon interaction and the decay signature of the tau, as well as high energy losses along the muon track.
The analysis does not include any likelihood reconstructions which provide parameters to discriminate cascade-like from track-like events [98]. In particular the reconstructed energy can be used to filter efficiently signal and background events and should also be included in a search for cascade-like events. Finally, the filtering could be optimized by introducing more filter levels.
In conclusion, the analysis presented in this work introduces new methods to improve existing cascade searches by exploiting the capabilities of the waveform readout system. A robust classification scheme has been utilized which yields promising results. This work suggests to focus on energies in the range from 10 TeV to 10 PeV in searches for cascade-like events. With the inclusion of all neutrino-flavors and energy reconstructions, an improvement of a least one order of magnitude is expected, reaching the sensitivity of diffuse muon analysis.