WMOD00061

• Sponsor Name: Utah Department of Natural Resources
• Title: Propane Cloud Seeding Expiriment Overview: Wasatch Plateau, Utah, During Winter of 2003-2004
• Prepared For: Utah DNR
• Prepared By: Dr. Arlin Super and James A. Heimbach, Jr.
• Date: 3/1/2005
• State: UT
• Weather Modification - Doc Type: Scientific Study

GSC, nearest the HAS, had a somewhat suggestive P-
value of 0.09.

While not shown in the tables, the three core gauges
had mean residuals from the pairs testing equivalent
to 0.012 to O.oI5 inch per hour. This means that, on
average, the seeded EUs in this partition had
precipitation rates that much higher than the
nonseeded EUs with minor variation from gauge to
gauge. The stated values averaged 1.5 times as much
mean SWE per EU as suggested by testing the entire
population in Table 1. The 47 pairs of Table I
represent 1.5 times as many EUs as the 32 pairs of
Table 2. Consequently, these results indicate that the
entire seeding effect at the core gauges was contained

within the wind direction partition. This is a
physically reasonable result which would be expected
if targeting was limited to southwest quadrant winds
at the HAS, as demonstrated by prior plume tracing
studies. If the seeding effect was not concentrated in
this wind direction partition, doubts should be raised
about the reality of the statistical suggestions.
Conversely, finding the seeding effect concentrated
within this partition lends considerable physical
support to the statistical results. Also, if the seeding
effect was not concentrated in this partition, the lower
number ofEUs would have produced larger P-values
than the tests using the unpartitioned data set. The
fact that they were approximately the same gives
further confidence in these results.

Table 2. HAS wind direction partition <= 270 deg.
32 pairs, 35 seeded and 34 nonseeded

Pooled
Target -
GNO Sample
Target Correlation MDR
GSC 0.82 1.22
GTR 0.85 1.24
GSO 0.86 1.26
GDN 0.80 1.27

Mean double ratios increased somewhat from
Table 1 to the 22 to 27% increase range and the
confidence intervals show minimum increases of 5 or
7% for the three gauges with P-values between 0.03
and 0.06. The minimum increase for gauge GSC
nearest the HAS may have been as low as none by
the confidence interval but the MDR was a
respectable 1.22.

Two other single partitions were attempted, for
SL W presence and by HAS air temperature. A
number of dual partitions were tested with one
always rejecting HAS wind directions with a
northerly component (> 270 deg). That latter
requirement alone reduced the population size to 69
EUs as in Table 2. Therefore, any second partition
would reduce the sample size further so dual
partitioning attempts were recognized as challenging.
Detecting a real seeding effect at an acceptable level
of statistical significance becomes problematic with
small sample populations.

The most successful dual partition removed the
larger natural snowfall periods with the expectation
that a seeding signal would be more detectable during
lower natural snowfall rates. Experimental units in
the specified wind direction range were excluded

Re-
randomized I-tailed 95%
I-tailed Confidence
P-value Interval
0.09 1.00 - 00
0.05 1.05 - 00
0.03 1.07 - 00
0.06 1.05 - 00

when control gauge GNO accumulation exceeded
0.05 inch for the 40 minute EU period. That criterion
removed 20% of the EU population which met the
wind direction criterion of 270 deg or less, and had
the highest observed natural SWE amounts. In other
words, the 0.05 inch SWE value was the 80th
percentile for GNO ofthe ordered 69 EUs with wind
direction no greater than 270 deg. The remaining
sample population was still of reasonable size with
55 EUs of which 48 were paired. Some of the
excluded EUs might be considered outliers.

The MDR results for this dual partition are given in
Table 3. It is noted that correlations are much
reduced from Tables 1 and 2 as might be expected
with elimination of the higher values. The reduced
correlations make it more difficult to detect a seeding
signal with the available sample population, all else
being equal. Therefore, the results are especially
encouraging with P-values of 0.10, 0.04, O.oI and
0.02 ordered by distance from the HAS. The
associated confidence intervals indicate a 95%
certainty of a minimum increase of9% for GTR and
between 16 to 21 % minimum increases for the other
gauges. The MDRs ranged between 1.30 and 1.51
suggesting large percentage increases of 30 to 51 %
from seeding within this dual partition. While it

8