Abstract
Current data analysis methods leading to a climatological probability forecast of the first fall and last spring occurrences of preselected low temperatures do not work well in Florida, the Gulf Coast States, and the Southeastern Atlantic States where these seasons are not well defined. A proposal is made to recognize only a “cold season” beginning July 1 and ending June 30 and to construct data series of the first and last cold season events of low temperatures of interest to assign climatological probabilities to the dates of their occurrence. This proposal eliminates the analytic difficulties of present methods and extracts information from the data series more efficiently. A fit to the normal distribution is verified, and a computational example is given.
