Potter T.D., Colman B.R. (co-chief editors). The handbook of weather, climate, and water: dynamics, climate physical meteorology, weather systems, and measurements

Подождите немного. Документ загружается.

the embryos. They range in size from 1 to 4 or 5 mm. The second stage involves the

growth of the embryo to hailstone size. This requires the presence of ample super-

cooled water at sufﬁciently cold temperatures to grow the hailstone. The most

efﬁcient growth occurs between the temperatures of 10 to 30



C, at elevations

of 5 to 8 or 9 km in the atmosphere. Unfortunately, the source regions of the embryos

are not well known and can come from several locations in a storm. This makes it

difﬁcult to modify the hailstone growth process. This two-stage scenario indicates

that the microphysical and dynamical processes in a storm must be matched in a

special way to produce hail.

2 MODIFICATION METHODS

Rain and Snow

Knowledge and understanding of the precipitation processes lead to ﬁnding ways of

modifying the processes. The two techniques used for rain and snow increases for

cold clouds are called microphysical (or static) and dynamic seeding methods. The

ﬁrst method rests on the assumption that clouds lack the proper number of rain or ice

embryos for the maximum precipitation to occur. Hence, great emphasis is placed on

increasing the precipitation efﬁciency of the cloud, making sure that as much as

possible of the cloud water is converted to precipitation. To affect the ice process,

artiﬁcial nuclei, such as silver iodide, can be added or powdered dry ice (as cold as

80 to 100



C) can be dropped through the cloud, forming ice crystals by both

homogeneous and heterogeneous nucleation. (Homogeneous freezing refers to the

change of phase of water from vapor to ice without the assistance of a nucleus.

Heterogeneous freezing occurs with the aid of a nucleus, which allows freezing at

warmer temperatures than in homogeneous freezing.) The most striking effect of this

microphysical seeding mode is the production of precipitation from marginal-type

clouds, those on the brink of producing precipitation. However, the most effective

precipitation increase may come from treating more vigorous but inefﬁciently rain-

ing or snowing clouds.

The concept of the second process, dynamic seeding, is that seeding a super-

cooled cloud with large enough quantities of artiﬁcial ice nuclei or a coolant such as

solid carbon dioxide (dry ice) will cause rapid glaciation of the cloud. The resultant

latent heat release from the freezing of supercooled drops (modiﬁed by the adjust-

ment of saturation with respect to ice instead of with respect to liquid) will then

increase the buoyancy of a cloud. The increased vigor of the cloud may result in a

taller cloud, stronger updraft, more vapor and water ﬂux, broader and longer lasting

rain area, stronger downdrafts, and greater interaction with neighboring clouds,

resulting in more cloud mergers and, hopefully, more rain. The reasoning sounds

convincing, but there are many junctures where the process may go astray, and much

research is needed to verify the hypothesis. Recent ﬁe ld studies of this seeding

process in convective cloud systems in Texas (Rosenfeld and Woodley, 1993)

436 WEATHER MODIFICATION

have shown nearly twice as many mergers in the seeded complexes compared with

the unseeded cloud complexes.

Theoretical results have indicated that dynamic seeding can also affect relatively

dry stratus-type clouds as well as wet cumul us clouds (Orville et al., 1984, 1987).

The action of the seeding is to cause embedded convection in the stratus clouds. One

of the challenges with both of these seeding methods is to identify those clouds and

cloud environments that will respond positively to the seeding.

The amount of seeding agent to use depends on the type of seeding agent, the

seeding method, and the vigor of the cloud. One gram of silver iodide can produce

about 10

nuclei when completely activated. The activation starts at 4



C and

becomes more efﬁcient as the atmosphere cools, about one order of magnitude

more activated nuclei for each 4



drop in temperature.

Normally the goal is to supply about one ice nucleus per liter of cloudy air to

affect the precipitation process through the microphysical seeding method. Up to

100 times this number may be needed for the dynamic seeding method to be

effective. The total amount of seeding material needed depends on the volume of

cloud to be seeded, which will depend on the strength of the updraft. In general, a

few tens of grams of silver iodide are needed for microphysical seeding and about

1 kg or more for dynamic seeding. Dry ice produces about 10

ice crystals per gram

of sublimed CO

in the supercooled portion of the cloud. Normally about 100 g of

dry ice are used per kilometer of aircraft ﬂight path for the microphysical seeding

method, with only about 10 to 20% of the dry ice being sublimed before falling

below the 0



C level in the cloud.

There are other determinants for the growth of precipitation-sized particles that can,

in at least some cases, be artiﬁcially inﬂuenced. In the case of the collision–coales-

cence mechanism, opportunities sometimes exist for providing large hygroscopic

nuclei to promote initial droplet growth or much larger salt particles to form raindrop

embryos directly. The purpose of the hygroscopic seeding is thus to produce precipita-

tion particles either directly or by enhancing the collision–coalescence mechanism.

Two salt seeding methods are currently in use. One method applies hundreds of kilo-

grams of salt particles (dry sizes are 10 to 30 mm in diameter) near cloud base to

produce drizzle-size drops very soon after the salt particles enter the cloud. The second

method uses salt ﬂares to disperse 1 mm or smaller size particles into cloud updrafts, a

method currently receiving renewed interest in cloud seeding efforts (Mather et al.,

1996, 1997; Cooper et al., 1997; Bigg, 1997; Tzivion et al., 1994; Orville et al., 1998).

The salt material is released from kilogram-size ﬂares carried by aircraft; several ﬂares

are released per cloud cell. The salt particles change the size distribution of CCN in the

updraft, creating a more maritime-type cloud. Coalescence is enhanced; rain forms in

the seeded volume, eventually spreading throughout the cloud. This seeding thus

accelerates the warm rain process. In addition, if the updraft lifts the rain to high

enough altitudes, then the ice processes are also enhanced because of the larger

drops and droplets, making it more likely that freezing of some drops will occur.

Graupel and snow are more easily formed, increasing the precipitation efﬁciency of

the cell. These hygroscopic seeding methods are thought to work only on continental-

type clouds.

2 MODIFICATION METHODS 437

In the past, large water drops were added to clouds to accelerate the rain process,

but this is not an economically viable way to increase warm rain.

Hail

The suppression of hail from a hailstorm is a much more complex matter than the

increase of rain or snow from more benign cloud types . Two concepts appear to offer

the most hope for the suppressio n of hail. They are called beneﬁcial competition and

early rainout. To suppress hail according to the beneﬁcial competition theory, many

more hail embryos must be introduced into the cloud cell than would occur naturally.

(Cloud seeding with silver iodide is one way to provide the additional embryos.)

According to this hypothesis, the sharing of the available supercooled water among a

larger number of hailstones (i.e., their ‘‘competition’’ for the available water) reduces

their size. If enough embryos are present, it should be possible to reduce the local

supercooled liquid water content and hence the hailstone growth rates so that no

particle grows large enough to survive without melting during fallout to the ground.

Thus, less hail and more rain would be produced by the storm.

The consensus of many scientists is that this concept is most promising in storms

containing large supercooled drops, since these drops when frozen (caused by the

seeding) are large enough to act as efﬁcient collectors and to provide the necessary

competition. In storms containing only supercooled cloud droplets, this possibility is

absent, and there is greater difﬁculty in creating graupel embryos in the correct

place, time, and amount. Also, note that some storms may be naturally inefﬁcient

for the product ion of hail and the addition of potential hail embryos may actually

increase hail production.

Creating early rainout from the feeder cells of an incipient hailstorm also appears

to be an attractive method to suppress hail . Cloud seeding starts the precipitation

process earlier than it would naturally. The weak updrafts in the feeder cells cannot

support the rain particles and they fall out without participating in the hail formation

process. Thi s premature rainout removes liquid water and is accompanied by a

reduction of the updraft strength due to the downward force components caused

by both water loading in the lower par t of the cloud and n egative buoyancy caused

by cooling resulting from melting of ice particles and the evaporation of precipita-

tion beneath the cloud. These processes are thought to inhibit the hail generation

process. Some operational projects use this method with encoura ging results.

3 SOME SCIENTIFIC AND TECHNOLOGICAL ADVANCES IN

THE PAST 25 YEARS

Weather modiﬁcation as a scientiﬁc endeavor was last reviewed on a national basis in

the late 1970s. At that time the Weather Modiﬁcation Advisory Board (WMAB),

established by an act of Congress in 1976, produced a report and a two-volume

supporting document, detailing the status of weather modiﬁcation in both research

and operations (WMAB, 1978). Nearly $20 million per year was being spent on

438 WEATHER MODIFICATION

federal research in weather modiﬁcation at the time of the Congressional act. An

ambitious program was proposed to continue development of the technology, but

little action was taken. Weather modiﬁcation research funding dwindled to nearly

zero in the ensuing years. However, operational weather modiﬁcation continued and

in recent years has shown signs of invigoration. Much of the operational work is

being done in large areas of the midwestern and western United States (see Fig. 1).

Most of the following material comes from the BASC workshop summary

(BASC, 2000).

Hygroscopic Seeding of Convective Clouds

An exciting breakthrough in the development of rain augme ntation technology was

made within the past decade. Three randomized experiments in three different parts

of the world showed that hygroscopic seeding increased rainfall from convective

clouds. Statistically signiﬁcant increases in radar-estimated rainfall were achieved by

hygroscopic ﬂare seeding of cold convective clouds in South Africa, its replication

on cold convective clouds in Mexico, and by hygroscopic particle seeding of warm

convective clouds in Thailand. These efforts included physical studies as well as

statistical experiments and resulted in strong evidence indicating that the increases

in rainf all were due to seeded clouds lasting longer and producing rain over a

larger area.

The common elements of the three randomized experiments were: (1) seeding

with hygroscopic particles, (2) evaluation using a time-resolved estimate of storm

rainfall based on radar measurements in conjunction with an objective software

package for tracking individual storms (different software was used for each experi-

ment), (3) statistically signiﬁcant increases in radar-estimated rainfall, and (4) the

necessity to invoke the occurrence of seeding-induced dynamic effects to explain the

results.

The great signiﬁcance of the Mexican experiment was that it replicated the results

of the South African experiment in another area of the world. Several past programs

in the world failed when attempts were made to replicate previously successful

programs. Figure 2 displays a comparison of the quartile values of radar-derived

rain mass in seeded (solid lines) and nonseeded (dashed lines) cases for the three

different quartiles as a function of time after ‘‘decision time’’ (the time at which the

treatment decision was made) for the South African (dark lines) and Mexican (gray

lines) experiments.

The Mexican data were averaged over 5-min periods, while the South African

data were averaged over 10 min. To be consistent, the combined results are plotted in

Figure 2 as radar-derived ‘‘rain mass accumulated per minute.’’ The indicated differ-

ences in rain mass for both the South African and Mexican experiments were

statistically signiﬁcant after 20 to 30 min and remained signiﬁcant for the remainder

of the period. It is clear that the results from both experiments are in good agree-

ment. The main difference is that the storms in South Africa tended to last somewhat

longer than those in Mexico.

3 SOME SCIENTIFIC AND TECHNOLOGICAL ADVANCES IN THE PAST 25 YEARS 439

Figure 1 Active cloud seeding projects in the mid-western and western United States and Canada in the year 2000. (Figure supplied

by Bruce Boe, Weather Modiﬁcation, Inc.). See ftp site for color image.

440

Glaciogenic Seeding Effects in Convective Clouds

The seeding results on these clouds are more controversial and more difﬁcult to

document (Rangno and Hobbs, 1995, 1997; Silverman, 2001; but also note Nicholls,

2001). Strong glaciogenic seeding signatures, which appear 2 to 3 min after

initial seeding, have been documented in treated clouds by aircraft (Woodley and

Rosenfeld, 2000) and, more recently, from space. Glaciation times after seeding are

halved in both maritime and continental seeded clouds . These are consistent with the

conceptual model of dynamic seeding guidi ng the glaciogenic seeding experiments.

In addition, there is now limited ﬁeld evidence and some numerical modeling results

that indicate ice-phase seeding invigorates the internal cloud circulation concurrent

with the growth of graupel to precipitation size and the depletion of the cloud water.

These too are consistent with expectations from the conceptual model.

Radar estimation of the properties of treated and nontreated convective cells

shows indications that under certain conditions the seeded cells produce more rain-

fall as shown by the increasing maximum radar reﬂectivities, maximum areas,

maximum rain-volume rates, duration, clustering and merging of cells, and inferred

maximum rainfall rates. The indicated effects on rainfall quantities require further

Figure 2 Quartile values of radar-derived rain mass (precipitation ﬂux integrated over 1 min)

versus time after ‘‘decision time’’ for seeded (solid lines) and nonseeded (dashed lines) cases for

the South African (dark lines) and Coahuila (gray lines) experiments. First quartile is the value

of rain mass that is larger than the value for 25% of the storms; second quartile value exceeds the

value for 50% of the other storms; and third quartile value exceeds the value for 75% of the

storms. (Figure supplied by Roelof Bruintjes, National Center for Atmospheric Research).

3 SOME SCIENTIFIC AND TECHNOLOGICAL ADVANCES IN THE PAST 25 YEARS 441

validation by direct surface measurements. Earlier experimentation had also indi-

cated that increases in the top height of the treated clouds accompanied the apparent

rainfall increases. In recent years, however, this apparent height signal has been

much weaker, due probably to a change in the method of estimating cloud top. In

the early years, cloud tops were measured directly with a jet aircraft, while radar has

been used in later years to estimate cloud top. Because radar may underestimate the

tops of seeded clouds relative to nonseeded clouds as a consequence of the seeding-

induced glaciation (glaciated clouds are less radar reﬂective than unglaciated clouds

in the absence of hail), satellite measurements of cloud tops must be used to recheck

this link in the conceptual chain.

Snowpack Augmentation

A signiﬁcant accomplishment in recent snowpack augmentation research is the

establishment of the direct link between the seeding activity and the water reaching

the ground in the form of snow. The incre ases in precipitation rate caused by silver

iodide seeding have been documented several times in the reviewed scientiﬁc litera-

ture (Reynolds, 1988; Super and Holroyd, 1997). The link has been established by

physical and chemical techniques. The snow falling at particular targeted sites is

connected directly to the seeding material (see Fig. 3).

Figure 3 Observed concentrations of precipitating ice crystals, ice nuclei, and precipitation

rate during one hour of AgI seeding between 0945 and 1045, December 15, 1994 in Utah

(Super and Holroyd, 1997).

442

WEATHER MODIFICATION

The methodologies used to establish this direct chemical linkage have been

described by Warburton et al. (1985, 1994, 1995a, b), Super and Heimbach

(1992), Chai et al. (1993), Stone and Huggins (1996), Super and Holroyd (1997),

and McGurty (1999).

One big advantage of snowpack work is that the scientists are dealing with solid-

state preci pitation that can be sampled in ﬁxed and targeted areas both during and

after storm events and stored in the frozen state until analyzed.

Hail Suppression

In recent years, crop hail damage in the United States has typically been around $2.3

billion annually (Changnon, 1998). Susceptibility to damage depends upon the

crop type, the stage of development, the size of the hail, and also the magnitude

of any wind accompanying the hailfall.

Property damage from hail in recent years has been on the same order as crop hail

damage, usually topping the $2 billion mark, sometimes more. A recent report by

the Institute for Business and Home Safety (IBHS, 1998) indicated that losses

from wind storms involving hail, from June 1, 1994 through June 30, 1997, totaled

$13.2 billion. While some of this damage resulted from wind, hail certainly

accounted for a signiﬁcant fraction of the total damage.

Some of the recent high-dollar hailstorms include: Denver, 1990, $300 million;

Calgary, 1991, $350 million; Dallas–Fort Worth, 1992, $ 750 million; Bismarck, ND,

1993, $40 million; Dallas–Fort Worth, 1995, $1 billion; and Calgary, 1996,

$170 million. Wichita (Kansas) , Orlando (Florida), and northern (Arlington) Virgi-

nia are just a few of the other U.S. locales that have recently been hard hit by

hailstorms.

Results from North American hail suppression programs vary. The North Dakota

Cloud Modi ﬁcation Project (NDCMP) reports reduct ions in crop hail damage on the

order of 45% (Smith et al., 1997), while the Western Kansas Weather Modiﬁcation

Program (WKWMP) reports reduced crop hail losses of 27% (Eklund et al., 1999).

Neither program reports any statistically signiﬁcant changes in rainfall.

Both of these projects are operational, nonrandomized programs, and the evalua-

tions are based upon analyses of crop hail insurance data. Projects elsewhere in the

world (e.g., Dessens, 1986) have generally reported similar reductions in damage.

Considerable success has been achieved using numerical cloud models to simu-

late hailstorms and hail development (Farley et al., 1996; Orville, 1996). This has

contributed signiﬁcantly to the development of the contemporary conceptual models

for hail suppression. Contemporary numerical models contain microphysical compo-

nents, as well as cloud dynamics.

If a cloud model, after programming with actual atmospheric conditions, can

successfully reproduce a cloud or storm like that actually observed in those same

atmospheric conditions, the model has reinforced the physical concepts employed

therein. Such cloud models can be used to test concepts for hail suppression. If the

model employing a certain concept gets it right, this strengthens the conﬁdence in

the concept.

3 SOME SCIENTIFIC AND TECHNOLOGICAL ADVANCES IN THE PAST 25 YEARS 443

It is impossible to ﬁnd two identical clouds, seed one, and leave the other

untreated as a control, since no two clouds are exactly the same. However, the effects

of seeding can be examined by modeling a cloud beginning with natural conditions

and simulating seeding in a second run. Any differences in behavior can then be

attributed to the seeding. In fact, this option is very attractive, as the timing (relative

to the life cycle of the subject cloud), locations, and amounts of seeding can be

varied in a succession of model runs to better understand the importance and effects

of targeting.

Figure 4 gives an example of numerical model output, showing a decrease in hail

in the large sizes and an increase in graupel size particles due to silver iodide cloud

seeding. Total hailfall was decreased by 44%, hail impact energy by 58% in the

seeded case as compared with the nonseeded case.

Advances in Technology

Instrumentation Unusual opportunitie s for determining, by direct measurement,

the physical processes and water budgets of cloud systems and their changes due to

purposeful and inadvertent cloud modiﬁcation reside with new technologies,

particularly remote-sensing technologies. Many other instruments and techniques

have been developed that are applicable, including satellite systems and in situ

systems including aircraft platforms. However, the main point to be noted is that

none of these technologies were available during the period of about 1965–1985

when signiﬁcant funding was directed toward weather modiﬁcation research. There

has been a paradigm shift in observational capabilities, and cloud modiﬁcation can

now be revisited with the new and emerging tools.

Figure 4 Size distributions of graupel and hail particles striking the ground (at the region of

maximum hailfall from computer-simulated hailstorm). Categories 11, 14, 17, 20 represent 2-,

6-, 16-, 42-mm hailstones, respectively (Farley et al., 1996).

444

WEATHER MODIFICATION

An abbreviated list of signiﬁcant tools that have advanced our ability to observe

cloud systems is summarized in Table 1. This table indicates that the microphysical

and dynamic history of seeded and nonseeded clouds can now be docum ented and

results compared between the two types of clouds.

Cloud Seeding Agents Practical capabilities have been established over the

past 50 years, and especially since about 1980, for generating highly effective ice

nucleating aerosols with well-characterized behaviors for both modeling and

detecting their atmospheric fate after release for cloud modiﬁcation. In the same

period, our understanding of natural ice nucleation and our prospects for further

elucidating ice formation processes have improved. Improvements in understanding

how hygroscopic aerosols interact with clouds and how to use them to increa se

precipitation have also occurred in recent years.

By 1980, a clear need was recognized to account for the complex mechanisms of

ice formation by speciﬁc ice nuclei used in weather modiﬁcation ﬁeld programs. In

particular, the chemical and physical properties of aerosols were established to be

very important in determining ice formation rates as well as efﬁciency. This recogni-

tion led to the development of new, highly efﬁcient silver chloro-iodide ice nuclei

(DeMott et al., 1983). These same nuclei can be generated with a soluble component

to enhance the action of a fast condensation-freezing ice nucleation mecha nism

(Feng and Finnegan, 1989). The development of similar, fast-acting and highly

efﬁcient ice nuclei from pyrotechnic generation methods followed by the early

1990s (Fig. 5). These new nucleating agents represent substantial improvements

TABLE 1 Some Remote Sensors for Hydrometeorology*

Winds, air motions in clouds, clear air Doppler radar, wind proﬁling radar,

Doppler lidar

Temperature proﬁles Radio acoustic sounding system (RASS)

Water vapor, cloud liquid water Microwave radiometer

Cloud boundaries Cloud (mm-wave) radar

Water vapor and liquid ﬂuxes and proﬁles (Combinations of the above)

Thermodynamic and microphysical

cloud structures

‘‘Retrievals’’ from Doppler radar data

Cloud phase, ice hydrometeor type

and evolution

Dual-polarization cloud and precipitation

radars, microwave radiometer,

polarization microwave radiometer,

polarization multi-ﬁeld-of-view lidar

Transport and dispersion, mixing

in clouds

Chaff wind tracking dual-circular-polarization

radar, gaseous tracers

Precipitation trajectories Chaff wind tracking and Doppler radar

Snowﬂake size, snowfall areal rate Dual-wavelength radar

Rainfall rate, hail differentiation Dual-polarization radar

*Table provided by Roger Reinking, National Oceanic and Atmospheric Administration.

3 SOME SCIENTIFIC AND TECHNOLOGICAL ADVANCES IN THE PAST 25 YEARS 445