Long-term variability of monthly total precipitation

A methodology was developed to quantify long-term temporal variability of monthly total precipitation. Ninety-three years of data from four weather stations on the Florida peninsula were used. This peninsula, covered by a single 8 degrees x 10 degrees grid of the NASA/GISS atmospheric GCM, is a transitional climatological region, influenced on a seasonal basis by mid-latitude continental fronts, convective thunderstorms and tropical disturbances. Traditional statistical descriptors suggested that the distribution of monthly total precipitation was highly non-Gaussian. A gamma probability density function was successfully fit to the historical data and used to investigate the possible presence of long-term trends. These findings suggested that the distribution of monthly total precipitation rather than the mean and standard deviation, should be used in weather and climate model development and validation. This methodology also may be applicable to sub-grid scale spatial variability of monthly precipitation.