A NEW LOOK AT THE ISRAELI CLOUD SEEDING EXPERIMENTS

Two statistical experiments, carried out in Israel, appeared for a tim e to have provided a unique demonstration of the ability of cloud seed ing to increase rainfall. In this paper the authors examine the possib ility that both experiments were compromised by type I statistical err ors(i.e., ''lucky draws'' or false positives). It is concluded that in the first Israeli experiment a type I statistical error produced the appearance of statistically significant effects of artificial seeding on rainfall 1) in the buffer zone and the center target area, 2) in th e coastal region of Israel, a few kilometers downwind of the seeding, and 3) in portions of Lebanon, Syria, and Jordan. Analysis of the seco nd Israeli experiment using the original crossover design produced a n ull result. However, when the two target areas were evaluated separate ly, naturally heavier rainfall over a wide region on days when the nor th target area was seeded produced the appearance of increases in rain fall due to seeding in the north target area, and when the south targe t area was seeded, the appearance of decreases in rainfall due to seed ing was produced. Target-control (as contrasted with crossover) evalua tions of the second Israeli experiment for the north target area alone foundered when control stations were selected from a relatively small region of anomalously low seed/no-seed ratios that was situated withi n a much larger region of high seed/no-seed ratios, which included Leb anon, Jordan, and most of Israel. Thus, the north target area seed/no- seed ratios are not an isolated, seeding-induced anomaly. On the contr ary, it is the low Seed/no-seed ratios of the northern coastal control stations, selected after the experiment began, that are anomalous in a regional context and are virtually the only stations that yield an a pparently statistically significant effect due to seeding in the north target area. It is concluded that neither of the Israeli experiments demonstrated statistically significant effects on rainfall due to seed ing. Considerations of the rainfall climatology of Israel, recent repo rts concerning the microstructure of clouds in Israel, and the relativ ely small amount of seeding carried out in the first Israeli experimen t support the view that seeding was unlikely to have had significant e ffects on rainfall. Contrary to previous reports, clouds in Israel con tain large cloud droplets, precipitation-sized drops, and considerable concentrations of natural ice particles at quite high temperatures, a ll of which should obviate attempts to increase rainfall by artificial seeding in wintertime air masses.