Influence of natural rainfall variability on the evaluation of artificial precipitation enhancement

Evaluating cloud seeding effects is one of the most critical issues in artificial precipitation enhancement experiments. However, the evaluation is not straightforward because there is natural rainfall variability, which subjects the atmosphere to spatiotemporal instabilities. The aim of this study is to analyze natural rainfall variability using the modern statistical simulation method, “bootstrap”, to analyze its influence on the evaluation of seeding activities and to take proper measures to control the influence. The study is based on the 1997–2007 airborne seeding macro records and the daily precipitation data in Jilin Province. The influence of natural rainfall variability can be reduced through three approaches: the increase of the supposed “seeded” sample size N , the rejection of outliers, and the selection of similar control units. A larger N leads to smaller calculated precipitation variability and detectable lower limits of seeding effects. When N is near 470 and the seeding effect is between 20% and 30%, the confidence level reaches 90%. For a single seeding operation, the case deletion model that rejects strong influence points and selects similar control units is established to control the influence of natural precipitation variability, which obviously improves the evaluation of artificial precipitation enhancement. The results demonstrate that the relative seeding effect in Jilin Province is concentrated mainly in the range of 0 to 30%, with an average of 11.95%, and has no significant linear relationship with the actual precipitation amount. However, the fluctuation amplitude of the relative effect decreases as the precipitation amount rises.