WMOD00550

• Title: Scientific Weather Modification Experimentation in the United States
• Weather Modification - Doc Type: Report

3. Preci~itation development pro-
ceeded as hypothesized in those clollds
with sustained updrafts such that the
main precipitation growth occurred at
temperatures colder than -10 oC (above
~he seeding level). These results are
supported by calculations (Cooper and
Lawson, 1984; Cooper, 1984) w~ich show
that accretiona1 growth is more rapid
and more efficient at about -12 and
-20 oC. In this respect, the seedin9
hypothesis which emphasized the warm
temperature region of the cloud was in
error and the choice of seeding level
was, perhaps, too low since it failed
to take advantage of the region of
rapid development of graupel from ice
crystals in most cases. The conclu-
sion that seeding would be more effec-
tive at a temperature level of -12 oC
and colder supports the notion that
there is a warm-side cloud top tem-
perature limit for seedabi1ity of con-
vective clouds (Grant and Elliott,
1974; Gagin and Neumann, 1981).

3. FACE'-2

i
I
. {
I
I
I
I
\
I
;
I
!
!

The Florida Area Cumulus Experiment (FACE)
was a long-term program to determine the
potential of dynamic seeding for increasing
convective rainfall over a fixed target area.
The physical concept underlaying FACE was
that massive seeding would increase areal .
rainfall by promoting growth, mesoscale orga-
nization and merger of cumulus in south
Florida through dynamic invigoration of cloud
towers with adjacent cloud systems. Both an
exploratory experime~t (FACE-I) and a confir-
matory experiment (FACE-2) were conducted to
test whether seeding in this manner increased
rainfall on the ground. Although a concep-
tual model and chain of physical events
following seeding were specified (Woodley et
al., 1982b), no attempt was made to verify it
through direct observation and measurement of
the intermediate processes. Both FACE-1 and
FACE-2 were primarily "black-box" experi-
ments, with the evaluation of the results
based on the statistical differences in rain-
fall characteristics of the seed and non-seed
populations.

Candidate experimental days were selected by
evaluating a daily suitability index that was
based on maximum model-predicted seedability
and morning shower activity as observed by
radar. Individual cloud suitability was
based on visual appearance and aircraft
observations of cloud top height, updraft
v~locity and liquid water. When the suitabi-
,'Iity criteria were met, pyrotechnic flares
that produced either 70 g of AgI or sand were
ejected from an aircraft in the top of the
clouds on randomly selected seed and non-seed
days, respectively. Typically 5 to 10 flares
per seeding pass were released in the acti-
vely growing portion of the treated clouds.
Ground-level rainfall was estimated using S-
band radar observations after adjustment by
rain gages.

The results of FACE-l (WOOdley et a1., 1982b)
provided evidence of precipitation increases
due to seeding. Analyses without predictors
suggested apparent increases in both location
(mean and median) and the dispersion
_._---~~---

~
:::;:.:;:: ';':, GO

(standard"deviation and interquartile range)
ch~racteristics of rainfall due to seeding, in

the total target (H) and floating target ,"
(FT), the most intensely treated portion of
th~ target. Approximately 50% and 25%
increases ;n the means for the FT and TT
variables, respectively, were found, with
substantial statistical support for the FT
results and lesser statistical support for
the TT results (i.e., P values of 0.01 to
0.09, respectively). Time profiles of FT and
TT rainfall composited with respect to the
time of the initial treatment pass also pro-
vided substantial evidence for a treatment
effect in FACE-1, with rainfall peaking
higher and later in the seeded cases.
Analyses of covariance using meteorologically
based covariates suggested an even larger.
effect of seeding than the overall results
with stronger statistical support. The
FACE-2 experiment was conducted in an effort
to confirm the indications of positive
seeding effects on rainfall from FACE-I.

FACE-2 was conducted during the summers of
1978, 1979, and 1980. There were 61 days of
experimentation of which 10 were A days (days
on which less than 60 AgI flares were used)
and 51 were B days (days on which 60 or more
AgI flares were used). Of the 10 A days, 3
were seed days and 7 were non-seed days. Of
the 51 B days, 25 were seed days and 26 were
non-seed days.

The confirmatory specifications (Woodley et
al., 1982a) and final data sets (Barnston et
al., 1982) for FACE-2 were identified prior
to any disclosure of the treatment decisions.
The confirmatory specifications were:

1. The first and weakest level of
confirmation will have been attained
if the seed FT and TT rainfall time
profiles are greater in magnitude and
peak later than the corresponding
control rainfall profiles with
substantial probability support in the
adjusted (for multiplicity) P values.
Failure to achieve this first level of
confirmation precludes moving to the
second level.

2. The second and somewhat stronger
level of confirmation will have been
achieved if the adjusted P values also
indicate positive treatment effects in
FT and TT rainfall in tne six hours
after initial seeding for the B days.
Again, failure to satisfy this level
of confirmation precludes moving on to,
the third and final level.

3. The third and highest level of
confirmation will have been achieved
if the adjusted P values also indicate
positive treatment effects in FT and
TT rainfall in the six hours after
initial ,seeding for t~e A & B days.

According to these specifications, FACE-2
failed to confirm the results of FACE-l
(WOOdley et al., 1983). The first and
weakest level of confirmation was not
achieved. Contrary to the confirmatory spe-
cifications and the hypothesized conceptual
mo~e1 of seeding effects, the mean rainfall
time profiles pe~.~~..::ar1 ie~_~:.? days

~,Jd!