WMOD00478

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

V2

-

Journal of the American Statistical Association, March 1979

B. Comparison of Average Precipitation on Seeded
and Not-Seeded Experimental Days In Concentric
. .
Rings
.3

!
;: .2
~
i:
U
III
f .I

>-
~
ct .1
...
z
III
~
IE
III
~
...

~
I .3
..
!

~ .2
:!
II:
~
It .1

"

>-
C(
o
.:. .I
...
Z
III
~
~
~
)(
III

It:
~ '
~ .3
!
~

~
;:: .2
j!
II:
U
III
2: .I

All days
--..__ .)ib SeMIlI .

. ------.----------L .--.-.----

-,.
- "'5eMId

A30 860 ClOD 120 E IlIO
MJUGE DISWCE FROM CENTER (.....1

.3

2

No' Seeded
~
----.....
-.....

---,

/

Seeded

Br days

.A 30

8 60

c

o 120

E

10

not-Br days

Not Seeded
._~

.-......~
....-.------.--- ---.---- ---.-------
-

~ -----
Seeded .

" 30

8 10

10

c

o 120

E

AVERAGE DISTANCE fROM CENTER (...Ie$1

.3

No' Seeded
'\.

.2

W days

A 30

C

8

D .120

E

60

90

Not Seeded

."\

",
.........-- ---
----- -----~- -----.-----

E days

Seeded
. \

A 30

860 C90 DI20E
AVERAGE DISTANCE FROM CENTER ......1

150

150

...
.-

150 F

150

",'
"

several sources, we located additional gages to increase
the total first to 174, and finally to 253, including all
reliable gages read in the morning around 8 A.M. Re-
peating the analysis, we found that the percentage de-
crease ascribable to seeding changed very little, first to. a
21 percent. decrease and then to a 20 percent d'ecrease,
with the two-tail significance probability (0.14) still not
significant. Although the rain gages were not placed at
random over the entire area, but rather were located for
other reasons before the experiment started and con-
tinued steadily throughout the experiment, the fact that
the .analysis is very insensitive to the increases in the
110 number of gages, even though their number was almost
doubled, suggests that the density of gages was sufficient
for the present purpose, at least, contrary to Braham's
assertion in this article.
If the apparent effect of seeding is real and the esti-
mated losses ascribable to seeding are roughly correct,
then the seeding in the Whitetop experiment caused the
region to lose about four million acre-feet of water per
season on the average (for each season except 1964, the
regional effect is negative; in 1961, 1962, and 1963, every
ring shows a negative effect). If the seeding had been
operll.tional rather than a50-percent seeded randomized
110 trial, the expected loss would have been twice as large-a
very large amount of water worth about $800 million
per s:eason at the rate of $100 per .acre-foot. Even if the
water rate were only $10, a very conservative figure, the
loss to the area in water alone would have been $80
million per season. It appears important to search for the
caUSHS and extent of the negative apparent effects of
seeding.
The experimental days were selected by Braham on the
basis of the wind direction at 8 A.M. at 4,000 feet level
above West Plains being between 1700 and 3400, plus
110 restriiciions on the amount of precipitable water at the
6 A.M. readings at the nearest U.S. Weather Bureau
radiosonde stations, Columbia, Missouri and Little Rock,
Arkansas. An immediate question is: Do the experi-
mental days correspond to the days hypothesized by
Braham (1966) as suitable for seeding on the basis of his
preliminary investigations-in particular, the west wind
aloft and the air-mass weather type? Or did the wind
shift during the seeding period?
For some time, the weather type was not available, nor
were the observations of wind measured by the pilot
balloon (pibal) at two-hour intervals throughout the day
180 at West Plains. Therefore, we used the twice-daily
radiosonde measurements of wind direction to designate
the experimental days which satisfied the wind criterion
at most levels throughout the seeding period. These were.
called Braham-days, or Braham radiosonde days, Br-
days for short. When we received the pibal measures for
West Plains, we repeated the stratification into Braham
pibal days and not-Braham pibal days, now labeled
Bp-wl.Ys and not-Bp days. The stratification "satisfied
at most levels" is a little subjective; we refined the'
definition of Braham days to be obje,..tive (Lovasich et al.
180 1971b) and labeled the resulting Cfl.tegories W-days and

F

F

F