WMOD00444

• Title: Evaluation of Idaho's 1992-1993 Winter Cloud Seeding Program
• Date: 10/1/1993
• Weather Modification - Doc Type: Report

Each of the eight target snowpack sampling sites was individually related to the average of
the six control sites of table 1, excluding sites 3 and 8. This arrangement was made to test
whether some portion of the target area might appear more impacted by seeding than the
overall target area. A particular combination of topography, AgI generator siting, etc. might
favor some portion of the overall target.

The R-values between the control sites and individual target sites ranged between 0.923 and
0.958, all relatively high values. Departures of the 1993 winter from the regression lines
ranged between 8 and 26 percent, the highest value corresponding with a R-value of 0.933.
The 1993 data points fell well below 2 standard errors of estimate in each case, and below
1 standard error of estimate in half the cases. Therefore, no statistically significant evidence
was provided that seeding increased the 1993 snowpack beyond natural variability at any
individual sampling site.

Because seeding did not start until mid-December 1992 in the Boise River drainage,
differences between April 1 and January 1 snowpack SWE were examined. Such data would
be expected to show more variability than April 1 observations for the entire accumulation
season. However, any seeding signal should be less diluted by using the difference data for
a period that was entirely seeded.

Although all eight target sites in table 1 had January 1 observations, control sites 4 and 44
did not. Figure 6 is based on the average of control sites 5, 15,47, and 48. The R-value was
not markedly reduced by using the April 1 minus January 1 difference data, resulting in R
= 0.949. The 1993 observation is well above the estimated value based on the regression line.
In fact, it is almost 2 standard errors of estimate above the line. The 1993 measurement
represents a 26.7 percent (4.0 inches SWE) increase above the regression line. However,
1965 and 1986 had larger positive departures, and 1972 also had a large departure, so the
1993 observation may represent nothing but a similar natural variation. Nevertheless, the
1993 data point is certainly in the right direction if seeding increased the target area
snowfall. The statistical analysis is simply not sensitive enough to determine if seeding was
partially responsible for the 1993 positive departures seen in figures 3 through 6.
Observations from additional winters are needed to provide more definition.

The four sites immediately downwind from the Boise target, noted in table 1, were compared
with control sites 4, 5, 15, 44, 47, and 48. These control sites were all judged unlikely to be
contaminated by seeding. These sites are the same control sites used in figures 4 and 5. The
resulting R-value was somewhat lower at 0.932, likely because of the greater distance
between the downwind and control sites. Figure 7 shows the 1993 data point is slightly less
than 1 standard error of estimate above the regression line. The 1993 point represents a 12.7
percent (2.4 inches SWE) positive departure from the estimate provided by the regression
line. Although the 1993 data point again has no statistical significance, no indication exists
of any downwind decrease in snowfall associated with the seeding.

16