WMOD00278

• Title: Addendum to Comments on "Weather Modification in Arizona, 1971" by Herbert B. Osborn
• Prepared By: Wallace E. Howell
• Date: 1/1/1971
• State: AZ
• Weather Modification - Doc Type: Report

.....~. II "'i..

liOn our part, we believe that the conclusion reached by
Dr. Decker is partly due to the complexity of his tables and
partly to the significance he attaches to area-to-area
comparisons. These we think are irrelevant. From the point
of view of effectiveness of cloud seeding, the only relevant
comparisons are between the seeded and not seeded precipita-
tion amounts in properly defined areas. Our own preference
would be a fixed target, perhaps with the entire circular
experimental area."

(The experimental area referred to is the 60-mile circle centered
on the radar and was divided in the original experimental plan into
in-plume areas influenced by seeding and out-plume areas not so
influenced. )

No physical hypothesis was then or yet has been put forward to
explain cloud seeding effects outside the Whitetop in-plume
so Neyman and Scott appear to have committed the well-known fallacy
of reasoning from statistical inference to physical causality without
independent physical verification. A simple example of this fallacy
is to reason that the blooming of the daffodils causes the oak trees
to leaf.

Later, Neyman, et aI, (1969) after expanding their study to a circle
180 miles in diameter centered on the Whitetop radar site restated
their conviction that lithe evidence in support of a causal rela-
tionship between seeding and loss of rain appears quite strong"
but worried that liThe situation is aggravated by the absence of
an intelligible theory, even a hypothesis, on the mechanism through
which cloud seeding conducted as a 'local measure' could have large
effects some 150 miles away." After they had extended the study in
time as well as space and discovered that the randomly selected
seeded days were characterized by warmer, drier, and less cloudy
weather not only during and immediately after the hours of seeding
but also during the 10 or 11 hours immediately before seeding and
not only downwind but as much as 180 miles upwind (Lovasich, Neyman,
et aI, 1971a and 1971b), they concluded that "This difference could
not be due to seeding." They added, "In consequence, 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."
The same conclusion is repeated by Neyman, et aI, (1972): "Until
a satisfactory explanation is found, this circumstance discourages
us from quoting the Whitetop results in support of any hypothesis
regarding the effects of seeding." This is a decisive abandonment
of the earlier position that rainfall deficits outside the seeded
plume should be attributed causally to the seeding, which means
abandonment of the sole grounds from which statistical significance
for rainfall decrease was inferred. Decker, et aI, (1971) meanwhile,
after subjecting in-plume and out-plume areas to a conventional
covariance regression analysis, reaffirmed the conclusion:

2

/
/