Characterization of Low-dt Non-poisson Noise in the Icecube Neutrino Detector
Open Access
- Author:
- Stanisha, Nick A
- Area of Honors:
- Physics
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Tyce De Young, Thesis Supervisor
Dr. Richard Wallace Robinett, Thesis Honors Advisor - Keywords:
- IceCube
Neutrino
Noise
PINGU
DeepCore - Abstract:
- The IceCube Neutrino Observatory is a 1-cubic-kilometer neutrino detector constructed within the glacial ice at the South Pole. Two sub-detectors, DeepCore and the proposed Precision IceCube Next Generation Upgrade (PINGU), have been designed to probe low energy events in IceCube. However, these low-energy events are dim compared to higher energy events and require more advanced noise simulation and removal methods. This thesis details the methods by which distinctly non-Poissonian noise hits were extracted from 117.38 seconds of IceCube data. These noise hits, identifiable in the IceCube data stream by a sharp increase in hit rate over a short period of time, are referred to as “correlated noise.” By extracting a sizeable correlated noise sample, a full noise profile was created which shows the timing distribution of all observed noise in IceCube. The time separations of noise hits in this profile extend from 10-9 seconds to 10-1 seconds. There are four main components which contribute to the full IceCube noise signal—thermal noise, afterpulses, correlated noise, and long-timescale correlated noise. Thermal noise and afterpulses have been characterized in previous studies—it is confirmed in this thesis that thermal noise is accurately modeled by a Poisson process with a rate of 220 Hz, and afterpulses are accurately modeled by a Gaussian distribution with a mean time separation of 6 µs and a standard deviation of 2 µs. It is shown here that the correlated noise component is approximately modeled by a log-normal distribution with a mean of -6 and a standard deviation of 0.848, in units of log10(δt/sec). Currently, there are no physical hypotheses explaining long-timescale correlated noise timing distribution, but its smoothness suggests that it might be well-approximated by a closed-form expression. Further studies are needed to determine the origin of long-timescale correlated noise and quantify DOM-to-DOM variations in the low-δt correlated noise region.