WMOD00473

• Title: Evaluation of Bridger Range Winter Cloud Seeding Experiment Using Control Gages
• Date: 12/12/1983
• Weather Modification - Doc Type: Report

DECEMBER 1983

ARLIN B. SUPER AND JAMES A. HEIMBACH, JR.

1991

,
'i.

Once an apparent type I error was detected by ex-
amining Wilcoxon test statistics at gages both in and
outside the expected target area, there was no known
objective scheme to overcome it. Since a major portion
of the BRE was apparently subjected to type I errors,
there seemed to be limited additional information to
be gleaned. Moreover, due to cutbacks in the funding
agency's budget, resources to pursue further analysis
were curtailed.
The problem of type I errors is common in ran-
domized seeding experiments; the Whitetop project
may have suffered from it (Decker et al., 1971), as well
as the Grand River experiment in South Dakota (Gel-
haus et al., 1'974). Neiburger and Chin (1969) discussed
the problem in relation to a Swiss hail suppression
project. Mielke (1979) reported that the Climax I and
II experiments were influenced by type I errors.
Recently, objective procedures have been developed
to overcome the type I error using upwind control gage
precipitation measurements (Mielke et al., 1981a,
1982). The analysis reported herein was largely inspired
by the availability of these procedures, which have
been applied to the BRE. Use of control gages greatly
increases the effectiveness of statistical designs over the
single-area approach as shown by Neumann and
Shimbursky (1972).
Because the BRE was clearly an exploratory exper-
iment according to the nomenclature of Gabriel (1981)
,and its analysis is of post hoc nature, the reader is
cautioned that problems of multiplicity of analyses
exist. This should be borne in mind in interpreting
probability values to be presented. The statistical results
cannot be considered scientifically conclusive without
a confirmatory experiment which has not been con-
ducted. Thus, they should be considered no more than.
suggestive.
The remainder of this paper is structured in the
following manner. Section 2 discusses the general
physical hypothesis envisioned at the time of the BRE
and'more recent findings which suggest that the hy-
pothesis requires revision. The equipment and mea-
surement systems employed in the BRE are considered
in Section 3, followed by a discussion of the evidence
for successful targeting of the seeding material in Sec-
tion 4. Section 5 summarizes the experimental design;
statistical analysis procedures are the topic of Section
6. Results of data partitions are shown and discussed
in Section 7. Section 8 considers supporting evidence
from snow courses while Section 9 gives a general
summary and recommendations.

2. Physical hypothesis

The BRE experimental design was strongly influ-
enced by the Colorado State University (CSU) Climax
I statistical results and CSU's physical studies and
modeling efforts associated with the then ongoing Cli- .
max II experiment. It was envisioned that the BRE

would test the concepts emerging from the CSU work
in. a higher latitude and over simpler topography. It
was assumed that the abruptly rising Bridger Range
(see Fig. 1) would provide strong uplift to moist east-
ward-flowing air, thereby producing liquid water con-
densate. It was also assumed that the concentration. of
natural ice nuclei and resulting ice particles was some-
times less than optimum for conversion of the con-
densate to snowfall. The latter assumption was based
on ice nuclei measurements made with both a Schaefer-
type mixing cold chamber and a NCAR ice nucleus
counter, on the Bangtail Ridge prior to the BRE. Both
instruments generally indicated less than one ice nu-
cleus per liter at -20DC (Super et al., 1969).
These ideas follow the classic paper by Ludlam
(1955). Chappell (1967) expanded on Ludlam's con-
cepts for the specific Climax situation and later mod-
eled the cold orographic cloud (Chappell, 1970). A
proposed physical explanation of the Climax statistical
results by Grant et al. (1969) followed similar reason-
ing. It was believed that cloud top temperature was
very. important because ground-based measurements
of natural ice nucleus concentrations were observed
to be highly temperature dependent. Cold cloud tops
were, therefore, expected to result in high concentra-
tions of ice nuclei and resulting crystals, which would
settle through the cloud effectively converting the
available condensate to snowfall. Conversely, as sug-
gested by the Climax I results, cloud tops wanner than
approximately - 20DC were expected to produce lim-
ited na.tural ice nucleus concentrations, in which case
AgI seeding might increase snowfall.
This physical concept has been questioned by Hobbs
and Rangno (1979). They point out that "direct air-
borne measurements in winter clouds over the Rockies
have since shown that the ice particle concentrations
bear little or no relation to the measured ice nucleus
concentrations, and that the ice particle concentrations
often exceed Grant et al. (1969) optimum concentra-
tions even at quite high temperatures". Cooper and
Saunders (1980) also note that ice nucleus measure-
ments were much lower than ice particle concentrations
measured in storms over the San Juan Mountains of
southern Colorado. However, in a companion paper,
Cooper and Marwitz (1980) suggest that seeding po-
tential may well exist in some storm stages where ice
particle concentrations were generally ~1O L-1 and 1
m S-1 updrafts were common.
Other recent observations also give cause to question
the importance of cloud top temperature. Rauber and
Grant (1981 a) present results of aircraft measurements
from two stably stratified storms over the Park Range
in northern Colorado. The liquid water was confined
to a narrow region over the windward slope and barrier
crest. Further, single ice crystals were horizontally
stratified, with their habit appropriate to their envi-
ronmental temperature and their limited fall velocities
apparently in approximate balance with the general