WMOD00284

• Applicant: Thompson, J.R., G.W. Wilderson, and D.A. Griffith, North American Weather Consultants
• Sponsor Name: USBR
• Title: Cloud Seeding Data Collection, and Analysis Assoc. with the Colo River Augmentation Demonstration Program
• Prepared For: USBR, Divison of Atmospheric Research
• Prepared By: Thompson, Wilderson, Griffith
• Date: 12/1/1987
• State: AZ
• Weather Modification - Doc Type: Scientific Study

gauges to rank first more than once were Happy Jack (3 times)
and Baker Lake (2 times) and neither of these ranked last
more than once. Conversely, Flagstaff Arboretum ranked last
three times and Blue Ridge R.S. ranked last twice, but neither
of these gauges ranked first more than once.

The Mingus Mountain gauge had a remarkably high correlation
coefficient (about 0.9) with the six other gauges, both individ-
ually and collectively, but this appears to be largely due
to the inclusion of the 20 year storm (storm no. 9) in the
11 pairs of data points used to develop the linear regression.
Inspection of the data used in Table 3-5 in the linear regression
calculation reveals that over a third of the data point pairs
are near the origin (about 0.10 inch precipitation or less).
Conversely, the data point in storm no. 9 is far from the
origin and this "outlier" significantly improves the correlation
coefficient. When storm no. 9 is eliminated from the calculations
the correlation coefficients drop to about 0.7. It is unlikely
that a storm of the magnitude of storm no. 9 will occur very
frequently. This suggests that using the Mingus Mountain
gauge as a control to predict the expected results in the
main target area could be useful, since about half the target
variance is accounted for. However, controls with a higher
association with the target could shorten the duration of
a statistical experiment.

3-31

I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I