Time Series Analysis of Low Level Gas Counting Data

Abstract

We demonstrate the feasibility of pulse time of arrival information for early detection of periodic events in low level counting. Time of arrival data allows us to apply time series analysis and serial correlation tests which, in graphic form, give the user an illustrative view of the parameters affecting the validity of counting statistics. The decision to discontinue counting can already be made on the basis of less than 100 counts from the time information alone if more than 10 of these are non-Poisson periodic counts. The analyses also serve as a means of quality control for low level counting, being directly based upon the interval distribution of the Poisson process.