WMOD00495

• Title: Night Vesus Day Cloud Seeding in Langmuir's Periodic Experiment
• Weather Modification - Doc Type: Report

Data published in the sources already referenced permitted comparisons
to be made among nighttime-seeded, daytime-seeded, and nonperiodic-
seeded 28-day cycles with respect to periodicity of precipitation
over regional subdivisions comprising a large part of the United
States and with respect to periodicity of temperature changes at the
70 kPa level over Omaha, Nebraska. The results of these comparisons
are shown in table 1 and figure 1. (A comparison of precipitation
amounts in South Carolina, Georgia, and Florida between nighttime-
seeded and daytime-seeded cycles 13 through 19 was also made from
the data of Brier's figure 4, but yielded only nonsignificant excess
during the nighttime-seeded cycles.)

The measure of periodicity in each instance is the coefficient of
correlation between the measured quantity for each day of the 28-day
cycle (precipitation or temperature) and the quantity for the same
day obtained from a 7-day sinusoidal wave fitted to the same data by
the method of least squares. The measure of amplitude is the amplitude
of the fitted wave. The measure of probability was obtained by
applying the Mann-Whitney rank test. This test has the advantages
of simplicity, freedom from assumptions about a distribution function,
and power appropriate to initial exploration of the data. Its
disadvantages are less power than more sophisticated tests and a
lack of accuracy for samples as small as the 10 cycles involved in
the nighttime-daytime comparison. The tabulated probabilities are,
therefore, to be considered only as suggestive, not definitive. A
(+) indicates that the measure for the first-mentioned set exceeded
that for the second-mentioned, and a (-) indicates the opposite.

The Mann-Whitney test for periodicity of temperature aloft at Omaha
involving nonseeded cycles was performed after first reconstructing
a rank list of nonseeded cycles from Brier's table 5 plotted as a
cumulative graph, and combining this reconstructed list with the
values for cycles 13 through 22 tabulated in Brier's table 6.

In his analyses, Langmuir distinguished between two nonperiodic-
seeded epochs. The first epoch embraced cycle 1 (which was to have
been the first seeded cycle but was called nonseeded because diffi-
culties with equipment and supplies prevented keeping the seeding
schedule) and cycles 23 through 29 immediately following termination
of the experiment. The second epoch embraced cycl es 30 through 3i',
and included two cycles, 31 and 33, when very high 7-day. periodicity
occurred, one of which was the object of special study by the Weather
Bureau (Hawkins, 1952; Brier, 1954). Even so, 7-day periodicity \'/as
weaker during the second epoch as a whole, with a mean correlation
coefficient of 0.14 compared with 0.28 during the first epoch.

2