Statistical Analyses of a Randomized Lightning Modification Experiment

Abstract

Analyses and interpretations of data from a randomized lightning modification experiment yield strong evidence that AgI seeding of mountain thunderstorms reduces the frequency of lightning and alters several lightning characteristics. Statistical procedures include likelihood ratio tests based on the log-normal distribution and two nonparametric tests (Wilcoxon and sum-of-squared ranks tests). Specifically, the estimated reduction in the long continuing current intervals in cloud-to-ground flashes is about one-fourth (two-sided p-values of both nonparametric tests are less than 0.04). Further, seeding appears to reduce the frequency of cloud-to-ground flashes and the duration of lightning activity by about one-half (two-sided p-values of both nonparametric tests are all less than 0.03). These results could have important applications in forest fire management programs.