Tests of the dynamic seeding concept for increasing cloud size through massive seeding with silver iodide generated by pyrotechnics were conducted in Pennsylvania, Arizona, in the Caribbean, and in southern Florida during the late 1960s. The tests confirmed the increased growth of the clouds, which was sometimes dramatic, as a result of massive seeding--the increase in cloud size in accord with predictions made by the numerical cloud model. Subsequently, the effect of dynamic seeding on rainfall in Florida cumulus was investigated in two projects in 1968 and 1970. A comparison of radar-measured precipitation from individual clouds showed an increase by a factor of 2-3 that was significant at the 0.5% level. The area-wide precipitation increase was almost surely not so large.

Conceivably, as a result of increased growth, compensating downdrafts could suppress the growth of other cumuli in the vicinity. Thus, even though precipitation increases from individual clouds might be proven, the rainfall over a larger area could be unchanged or even decreased. A Florida Area Cumulus Experiment (FACE) was organized in 1970 to look into this possibility, and the project is continuing. A high degree of selectivity, based on numerical cloud models, was used in the treatment process. Days with widespread precipitation were not seeded.

The Statistical Task Force reviewed the materials from the exploratory phase of Project FACE (1970-76). They concluded: "Results suggest the possibility of a seeding effect that, under some conditions, increases rainfall. At this point, because

exploratory phase), and the possibility of subjective influences of seeding knowledge, conclusions remain uncertain. The confirmatory phase began in the summer of 1978. Warm Season Cumulus clouds

In an area-type, warm-season cumulus cloud experiment in South Dakota, precipitation increases were reported for days characterized by cumulus showers. On days with large thunderstorms, there was essentially no seeding effect. The investigators ascribed the results to dynamic seeding although the seeding rates used were much smaller than those of either the FACE or Caribbean experiments. This points up the fact that there is no hard and fast line between static and dynamic seeding. All ice formation releases heat of fusion. Perhaps in some very marginal situations, even a small amount of additional glaciation could promote substantial additional cloud growth. This is an area where additional research is sorely needed.

A randomized study of seeding warm-season cumulus clouds for rain enhancement and hail suppression was carried out in North Dakota during summers of 1969 through 1972. The project was randomized by day with silver iodide released into cloud updrafts at cloud bases. Two different seeding rates were employed. If the project meteorologist judged a storm to be a potential hail storm on a day selected for seeding, the seeding rate was increased to a higher rate thought necessary for hail suppression.

The Statistical Task Force reports for this project, "Overall differences in rain for seeded and unseeded days are negligible and nonsignificant. When seeded days are separated into days judged suitable and unsuitable

for dynamic seeding, the mean difference in average rainfall between (suitable) seeded days and unseeded days approaches significance." They also found "When days are separated by 500-mb (millibar) temperature, results show significantly less rainfall under seeding on colder days [temperatures between -15°C (+5°F) and -20°C (-4°F)]. . .

At

་་

The economic significance of making adjustments in precipitation during the cropgrowing season in the agricultural heartland of the United States makes it very important to resolve the uncertainties about the effects of seeding warm-season cumulus clouds. present, the strongest scientific basis for believing that rain enhancement from such clouds is possible is the demonstrations, at high statistical significance, that dynamic seeding resulted in substantial growth of cumulus clouds in Florida and in the Caribbean. The conditions under which such growth occurs is understood and predictable. The FACE data suggest an associated increase in rainfall-a result that is not really unexpected considering the larger clouds. However, for a number of reasons discussed in detail in Volume II, the FACE results are not regarded as conclusive, and confirmation is sought in the renewed project begun in 1978.

There are several additional reasons for recommending an expansion of research into the seeding of warm-season cumulus clouds. Rapid advancements are currently being made in numerical modeling of convective cloud precipitation, from both natural and seeded clouds. One very important recent study shows a strong interrelationship between the warm cloud coalescence processes and the cold cloud ice-phase processes. This new insight into the inner workings of cumulus clouds helps explain the disparate results

often obtained in seeding seemingly similar clouds in different parts of the country. The improved numerical models will help improve the sensitivity of future seeding experiments. It is also important to note that many scientists believe that the rainfall decreases found in Project Whitetop are

now understood and that we should be able to avoid such results in future experiments.

Downwind Effects

People often ask what effect cloud seeding might have on precipitation downwind of the target area. Post-hoc analyses of several summer cumulus seeding projects have given some information, but the overall picture is far from clear. A positive precipitation effect was found up to several hundred kilometers downwind from a generator site in a Swiss hail-suppression project. Later, negative precipitation effects were found downwind and to either side of the Project Whitetop plume. Downwind precipitation effects of a negative character were claimed more than 100 kilometers downwind of the Arizona summer cumulus target area. In the Necaxa Basin project in Mexico, there appeared to be a positive precipitation effect downwind and a negative effect upwind.

target, separated from the first by a buffer zone, was to the south in Israel. Seeding of either the north or south target was randomized by day. Thus one target became the control area whenever the other was seeded. This is known as the "cross-over" seeding design. Most attention has been given to the northern target experiment.

Phase I indicated an increase in precipitation of about 15% (with statistical significance) when cloud tops were between about -10°C (+14°F) and -25°C (-13°F). Phase II, considered to be a confirmatory experiment testing the conclusions of Phase I, indicated a 13%-15% increase in rainfall at significance levels of 4% and 9% (in two different significance tests). Analysis suggested that the increases came from increased duration of rain periods rather than from increased rain intensity. When the results of Phase II became known, the Israeli Government decided to stop experimenting and launched a program of operational seeding at every cloud opportunity. A post-hoc analysis of effects downwind of the target indicated a positive response to seeding as far as 160 kilometers (about 100 miles) downwind.

The Statistical Task Force reviewed the only report thus far published about Phase II. They concluded that "If closer critical analysis of the study and its data fails to detect flaws, the results of this study would be judged to constitute confirmatory evidence that rainfall amounts have been increased by cloud seeding."

As far as we know, there are no exact analogs of the Israel-type cloud systems in this country. The closest analogs may be the winter cyclonic storms along the California coast, and in the U.S. Pacific Northwest,

which are known to contain numerous convective rain bands within which the highest intensity precipitation is located. There is evidence that convective bands also appear elsewhere, as in New England. They are typically about 40 kilometers (about 25 miles) wide, over 100 kilometers (about 62 miles) long, and may be spaced about 100 kilometers apart. They are normally found up to several hundred kilometers ahead of cold fronts or surface

occluded fronts. They also occur in the cold

Their

air behind the surface cold front. orientation tends to be parallel to the nearest front. They are easily observed by weather radars and can be traced through a mountainous region by means of rainfall analysis of a rain-gage network. The convective bands contain high concentrations of supercooled liquid water and are therefore prime candidates for artificial nucleation. A randomized seeding project involving winter clouds was carried out near Santa Barbara, California, during the period between 1967 and 1974. During the first 4 years (Phase 1 of Santa Barbara II), seeding was done by means of a high-output pyrotechnic device at a single site on a 1000-meter (about 3,280-foot) ridge line. The emphasis was

on individual convection bands embedded within winter cyclonic storms. Seeding was randomized band by band. Evaluation of Phase 1 indicated precipitation increases of 50% or greater from seeded bands. Seeding seemed to be most effective when the 500-millibar temperatures (presumed to be related to cloud-top temperatures) were between -12.8°C to -17.5°C (+9°F-0°F), tapering off tc negative results beyond -22.5°C (-9°F). There was also indication of precipitation increases up to 50% extending out to 160 kilometers (about 100 miles) downwind of the generator

site.

Phase 2 shifted the scene of action 100 kilometers (about 62 miles) to the west and further explored the downwind effects. Seeding was carried out from aircraft flying within convective bands upwind of the target area. Seeding was randomized in 48-hour periods, but the test was abandoned before an adequate sample was collected. Rainfall in the target and in a downwind region out to about 250 kilometers (about 155 miles, and 45° to the right of the 700-millibar windflow) was reported to have been increased 50% or more by seeding. There was good evidence that the duration of precipitation at any location within seeded bands was enhanced by an expansion of the width of the band. analysis of the pressure patterns based on standard National Weather Service microbarographs in the downwind area indicated that surface pressure beneath seeded bands was as much as 1.5 millibars less than in the nonseeded bands. These results suggest that the dynamic effect was realized in the seeded precipitation bands. Further confirmation of this observation could be sought in randomized tests using high-powered weather radar and airborne cloud physics observations to focus on the specific effects of seeding on band structure.

An

The Statistical Task Force stated that "the analysis of Phase 1 has been attempted in various ways Significance hovers. near the conventional 5% point." And "Phase 2 was terminated before it was reasonable to anticipate statistical significance."

Hail Suppression

Hailstones grow through the impact and freezing (riming) of supercooled cloud drops upon precipitation particles such as snow crystals and frozen raindrops. Very large

and destructive hail is formed within very deep and vigorous cumulus clouds in hot weather. The same types of clouds produce lightning and thunder, hence the association of hail with thunderstorms.

It has become clear that there are at least two different types of microphysical processes involved in the formation of hailstone embryos. One involves graupel, usually conical, growing by riming into a quasispherical hail form. The other involves freezing of raindrops and their further growth by riming. In either case, growth into large hailstones requires suspension of the hailstones for some period within a strong updraft.

The two types of embryos can be distinguished by examining the internal structure of hailstones. It appears that the freezing raindrop case occurs primarily in very moist air masses where low cloud bases are characteristic, whereas the graupel case occurs in relatively dry air masses with high-based clouds. Numerical models have succeeded in

making this distinction. They suggest that seeding with a large dosage of artificial nuclei may successfully reduce hail from the cases initiated on frozen drops but not in the graupel-initiated cases. However, there have been no adequate field tests of these distinctions.

There are two main hypotheses used in attempts to suppress hail by cloud seeding. One is to glaciate the upper parts of clouds to reduce the amount of supercooled water available for hail growth. The other is to add more hail embryos to distribute the supercooled water among a larger number of stones and thus reduce their individual sizes. Most experiments have been based

upon the latter (the competition hypothesis) since it requires less seeding material.

Efforts to suppress hail began in the United States in the 1950s. Privately supported projects over small areas first took place in high crop-hail loss areas in Nebraska and West Virginia well before experimentation had established a scientific approach to hail suppression. These efforts were conceptually based largely on the competition hypothesis. In those early years, there was no way to directly inject the materials for modification into the hailstone formation zones at high levels in storms, so the seeding materials were released either from the ground or from airplanes circling below the storm clouds.

Hail research efforts appeared in Canada in the 1950s, culminating in a major hailsuppression cloud seeding in the 1970s. However, pressure from research and operational groups led to a program in the Province of Alberta involving a mixture of operations and experimentation using cloud-base and cloud-top seeding techniques. Then, in 1976, farmer pressure for a fully operational project ended the randomized experiment after only 2 years. Although the Alberta Project was too short to provide firm scientific conclusions about seeding for hail suppression, it contributed. a great deal to the development of seeding systems and to the identification of hailproducing clouds in Alberta.

The first major U.S. experiment with hail suppression occurred in northeastern Colorado in 1959, but results were inconclusive. Operational programs of hail suppression, without any proof of great success or any foundation of sound scientific experimentation, continued into the 1960s and 1970s in Colorado, Kansas,

The Soviet Union has developed an ambitious hail-suppression effort and claims a 60%-80% effectiveness in reducing hail. Similar programs have since been fostered in several eastern European countries, and Soviet hailseeding equipment and supplies are marketed for sale around the world. The seeding technologies offered by the United States and the U.S.S.R. differ, the American approach using aircraft for seeding below or above storm clouds, sometimes combined with surface seeding devices, and the Soviet approach using ground-based rockets (or artillery) to shoot the material inside the storms.

An early experiment in Switzerland called Grossversuch III involved ground-based seeding with AgI to produce competing embryos. This 7-year experiment suggested a 66% increase in hail days on days when seeding was performed. Another experiment using ground-based AgI seeding was conducted in Argentina during a 3-year