WMOD00478

• Title: Field Experimentation in Weather Modification
• Date: 3/1/1979
• Weather Modification - Doc Type: Report

~'~~ "~l-\'~ A"

Neyman: Field Experimentation in Weather Modification

affirmative case, (iii) to guess and to verify the governing
chance mechanism.
Our several publications on the Whitetop experiment
represent the confirmatory phase of research, intended
to verify the generality of a certain phenomenon we
noticed when studying the Swiss experiment Gross-
versuch III. Regretfully, instead of confirming or denying
the generality of this phenomenon (far-away effects of
local cloud seeding), our studies of Whitetop brought
evidence that there must have been some difficulty with
randomization, and that, therefore, "any conclusions
about the effectiveness of seeding, one way or the other,
that are based on the Whitetop experiment must be made
with extreme caution" (Lovasich et 801. 1971b).
With reference to the clo.ud seeding literature, and this
includes Professor Braham's present article, I have regrets
about the incompleteness of reports on the experiments,
about occasional discrepancies with facts, and about in-
adequacy of statistical evaluations. Following are two
examples: I

(1) In Section 3, when describing the Whitetop experi-
ment, Braham asserts: "This experiment was designed to
test the hypotheses that ... seeding of summer cumulus
clouds. . . would (a) alter the likelihood of precipitation,
or (b) alter the amount of rain reaching the ground."
This description, in terms of testing the two hypotheses,
is not quite consistent with Professor Braham's testimony
before the U.S. Senate Committee on Commerce on
March 8, 1966, as follows: "We felt that our major
problem, scientific problem, was to determine the condi-
tions under which the modest increases in precipitation
might occur ..." (emphasis added). This is what in-
terested Braham in 1966. The awareness of hypotheses
(a) and (b) came later. This awareness is relevant to the
question of adequacy of statistical methodology used by
Braham. Some details are discussed below.

(2) Braham's description of Whitetop gives the im-
pression of smoothness in the implementation of a care-
fully designed plan. However, he warns the reader that
his report is brief and refers to his own detailed "Final
Report" (Braham 1966). Being interested in the details,
we examined this report, and what we found determined
the direction of our several published studies. The follow-
ing numbered passages are taken from Braham's "Final
Report," where they are somewhat dispersed. For this
reason, each quoted passage is followed by the page
number.

(i) Poem oj efforts: "Project Whitetop efforts were con-
centrated upon afternoon convective clouds, sometimes
called air mass convective clouds" (p. 25).
When I read this, I did not know the meaning of the
term "air mass." Consultations with Dr. Morris
Neiburger, Professor of Meteorology at UCLA, revealed
that meteorologists distinguish two categories of summer
convective clouds, the "air mass" and the "frontal"
categories. This dictated that our studies of Whitetop

c;, J~.:,..._-;;_! ..._

.,,~~-"

91

be stratified accordingly (Lovasich et 801. 197180). The
actual stratification was performed by Dr. H.C. Chin,
working under the supervision of Professor Neiburger, to
whom I am very grateful. It was found that the Whitetop
seeding on the "air mass" days was marked by un-
beHevably widee.pread (up to the distance of ISO miles),
large deficiencies of precipitation on days with seeding
compared to controls. On the other hand, some marks of
"modest increases" of rainfall ascribable to seeding were
found for the "frontal days."
Curiously, the distinction between "air mass" and
"frontal" summer convective clouds does not appear to
be mentioned in Braham's article now discussed. There-
fore, on a recent. occasion, I took the liberty of checking
the opinion of Professor Neiburger. I asked two meteorol-
ogi.sts the question, is there just one type of summer con-
vective clouds or are there more? The meteorologists'
res.dy replies were identical. They mentioned the "air
ms.ss" and the "frontal" categories.
While ignoring the two categories, one of which was to
be the central effort of the Whitetop project, Braham
discusses a number of "post-factum partitions based upon
meteorological factors and on distance downwind from
the seeding line." His comments vary considerably: (a)
"Statistical support for decreases in rain was very weak
. ..." (b) "On days when ... the target-control differ-
en<:es indicate positive treatment effects of ... +68 to
+llOO in rain, with strong support." (c) "Since days with
these very tall echoes dominate the rainfall pattern in
so\llthern Missouri in summer, they over-balanced the
poElitive effects indicated for days with medium-height
echoes."
The definitive tone of Braham's assertions' on the
strength of statistical support attracts my attention:

One aspect of Whitetop which has received considerable atten-
tion is the fact that both target and control areas received sub-
stantially less rain on days we seeded than on days we did not
seed. .. The 'tatiltimlauppurt for tAil being due to anythifl{/ but
chance iI very tDtak. (Emphasis added.)

(ii) Plan oj Whitetop and its impl.ementation: The plan
of the experiment involved the following definition of
opE:rational days: .

"The defining ...riteria for operational days" were (a) certain
requirements about "precipitable water," and (b) "the wind
direction at 4,000 ft. msl over West Plains, Missouri, mUlt be
between 1700 and 3400 inclusive on the morning of interest."
(p. 12, emphasis added.)

However, the actual practice involved deviations:

(a) "There wenl days on which the criteria said 'go' but on which
the seeder airplanes were unable to leave the ground..."
etc. (p. 14). Also, (b) "Several days were included in the
experiment even though the criteria were just below mini-
mum standards because other available information indi-
dicated that the day would be suitable" ,(po 130). Finally,
(c) "in all of the situations we fell back upon the best judg-
ment of the meteorologist in charge. .." (p. 15).

The combination of the outspoken "must" referring to
the 4,000 ft. level wind directions and of deviations from
this "must" made it imperative for us to investigate the
relevance of the directions of winds aloft. Here, however,