Estimation of daily area-average rainfall in central Florida using arithmetic averaging and kriging

Results are compared of using arithmetic averaging and kriging to estimate daily area-average rainfall during the summer 1991 Convection and Precipita tion/Electrification Experiment (CaPE) in central Florida, where convective rainstorms often produce rain-fields of high spatial variability. Four kri ging methods were applied: ordinary kriging using exponential and spherical semivariograms and universal kriging using linear and quadratic trends. Th e following results were obtained. First, the differences between daily are a-average rainfall estimated by arithmetic averaging and kriging are statis tically insignificant when the arithmetic mean is less than 12 mm. Second, for days with daily arithmetic means greater than 12 mm, kriging estimates are systematically less than the arithmetic means. Third, kriging estimates for daily area-average rainfall are insensitive to spatial resolution amon g grid spacing varying from 500 m to 20 km. Fourth, universal kriging with quadratic trends provides the best estimate of daily rainfall at a given lo cation among the four kriging methods.