WMOD00444

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

-

1. INTRODUCTI!ON

The State (State ofIdaho), in cooperation with local water interests, sponsored a precipitation
management program during the winter of 1992-19:93. This "cloud seeding" program was
designed and conducted by a private firm, NAWC (North American Weather Consultants),
of Salt Lake City, Utah.

The State requested that Reclamation (Bureau of Reclamation) perform an independent
evaluation of the seedling program. Reclamation was asked to statistically evaluate the
program by comparing target and control areas. Redamation was also asked to physically
evaluate the program in terms ofthe reasonableness of seeding generator placement, overall
output of effective IN (i.ce nuclei), general conduct of the program, and other factors. Neither
the statistical nor the physical approach was expected to provide definitive answers
concerning program effectiveness because of the limited duration of the seeding, the lack of
randomization, and the lack of supporting physical measurements. Nevertheless, a "first
look" evaluation was judged useful for planning future precipitation management activities
within the State. The State also requested that Reclamation make recommendations for
improving the operational seeding program in the future.

2. LIMITATIONS OF STATISTICAL EVALUATION

Statistical evaluation of cloud seeding projects is a challenging process which has significant
potential for serious bias and error. The main problem is that natural weather has
considerable variability. In particular, precipitation varies markedly with time and space.
(The terms "precipitation" and "snowfall" will be used synonymously herein because this
report deals with Idaho wintertime precipitation at high elevations). The amount of seasonal
snowfall change expected from successful seeding of winter orographic (mountain-induced)
clouds is on the order of 10 percent (American Meteorological Society, 1992). But 4,00-percent
variations in seasonal mountain snowpack accumulation are commonly found in the
intermountain West over periods of a few decades. The challenge is to detect the relatively
weak "seeding signal" within the large natural background "noise."

Elaborate schemes ha.ve been developed to minimize statistical errors III cloud seeding
experiments. The Weather Modification Advisory Board (1978) devoted an entire volume to
statistical issues. Statistical experiments are designed to minimize errors in the evaluation
process, employing such features as randomization ~md "blindness." Randomization refers
to the process of selecting only some of the experimental units (e.g., storms, days, entire
winters) for treatment (seeding) by random decision. Other units are left untreated, but have
the same measurements taken to serve as the "controls" or basis for comparison with treated
units. Blindness refers to the experimenters being unaware of which experimental units are
actually seeded and which receive a placebo treatment until after analyses are complete.
Statistical experiments also require precise definition of experimental units.