Statistical analysis of freezing temperatures in Central and Southern Florida

The probability distributions of the number of freezing periods per year, their timing and severity are examined at 25 stations in southern and central Florida. The proposed distributions, derived from stochastic crossing theory, provide adequate fits with the exception of three variables at different stations. The estimated parameter values reflect the changing geography of the freeze hazard. The freeze variable that exhibits the greatest variability over the area appears to be the number of periods, while the average severity of freezes and their timings are fairly homogeneous. The theoretical basis of the method permits estimation of the distribution of the dates of first and last freezes and the distribution of the coldest temperature likely to be encountered during any period of interest. The risks of periods of extremely cold temperatures, rarely observed in the historic records, may also be obtained from those at the more commonly encountered level. The technique provides a thorough yet easily interpreted summary of cold‐spell risks at any location.