Monitoring the data quality of the HiSPARC detector array

Lam, Ching Bon (2013) Monitoring the data quality of the HiSPARC detector array.

Abstract:HiSPARC is an experiment with an detector array of almost 100 stations in the Netherlands for researching cosmic rays. A sound analysis depends on good data. Therefore the detectors in the array have to be in good health at any time. A monitoring and notification system has been implemented by extending the current public website (Publicdb) and monitoring software (Nagios). The most probable value (MPV) of the pulseheight distribution is chosen to be the health metric, but other variables can be added relatively easy to the system. The MPV is extracted from data nightly by means of a fit. The data is considered to be good if the fitted MPV is within 4 standard deviations of the mean of the MPV over time. A systematic approach is taken to determine the MPV by fitting the pulseheight distribution with a Gauss function in the range where a maximum-minimum pair is found in the derivative. Fits are rejected when either the reduced �2 of the fit is smaller than 0.1 or when the fit range is lesser than 90 ADC. Toy experiments are conducted to measure properties of the fit method and selection. Data is used from station 505 on 20 January 2010 containing 39674 events between 50 and 1550 ADC. The pull distribution of the MPV for 20000 events per toy is a standard Gaussian. When there are 5000 events per toy, both a bias to the left (μ = −0.2215) and a tail on the left side are introduced. This might be due to the method of determining the fit range and requires further investigation. The selection efficiency drops from > 99% to 88% when the number of events of a fit drops from 20000 to 5000 events. Similarly, the systematic error increases from 1.2%/ADC to 3.4%/ADC. The contamination, however, stays below 1% for both cases. When the data is bad the rejection efficiency drops from > 99% to 91% for 20000 and 5000 events respectively. The contamination increases from < 1% to 9%. The standard deviation of the MPV over time is determined for 31 plates for data taken in 2010, 2011 and 2012. Data is fitted per day. Fits are selected when �2/N > 0.1 and fit range > 90ADC. Configurations are selected when the absolute MPV drift rate is less than 2 ADC per month. The weighted average of the standard deviation of the MPV over time is 1.54 ± 0.30%/ADC. iii
Item Type:Essay (Bachelor)
Faculty:TNW: Science and Technology
Subject:33 physics
Programme:Applied Physics BSc (56962)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page