Statistical Investigation of Anomalies in the Winter Temperature Record of Boston, Massachusetts

Abstract

Normal maximum and minimum temperatures are computed for each date of the winter season from the 92-year Boston temperature record. The trend is removed from the normal and individual yearly values and standard deviations and errors are computed. Statistical significance levels are calculated and a count of significant values is made. The probability of obtaining these by chance is computed and found to be quite high. Average maximum and minimum values are computed for each date for each phase of the double sun-spot cycle in addition to a major-half phase and minor-half phase. A procedure similar to that mentioned earlier is followed and four phases out of 20 are found to have 1 per cent significance counts with a high degree of confidence. The standard deviations of both maximum and minimum temperatures for the eight phases exhibit cycles that correlate with the double sunspot cycle. The result of applying statistical tests indicates that the cycles are real. An apparent division of the winter season is examined in which singularities tend to occur in the early part of the winter during the minor-half phases and in the latter part of the winter during the major-half phases.