Two general approaches are guiding the evaluation of the WWMPP: a) a randomized experiment that builds distributions of seeded and control (unseeded) cases, and b) exploratory studies to investigate a wide variety of ideas on detecting seeding effects, including physical studies to document the precipitation formation events hypothesized to be important to snowfall production in orographic storms. Collaborations with other researchers, particularly those at the University of Wyoming, have led to "piggy–back" studies applicable to the assessment of seeding impacts on precipitation formation and eventually on streamflows.

Map of Wyoming with coarse representation of topography and land use. Red outlined areas denote the three mountain ranges selected for cloud seeding operations: Medicine Bow, Sierra Madre, and Wind River. The randomized seeding experiment involves only the two southern ranges, the Medicine Bows and Sierra Madres.

Initial Studies

The first two winter seasons of the program (2005–06, 2006–07) were needed for performing exploratory studies to develop the evaluation plan, permitting activities, installing equipment, and peer review of the scientific experimental design. Initial assessments were necessary to:

1. Determine the suitability of current snow measurements (i.e., the NRCS SNOTEL network) for detecting the potentially small effects of cloud seeding
2. Specifically identify target areas
3. Define adequate seeding periods
4. Optimally locate ground-based seeding generators
5. Determine the frequency and locations of potential seeding conditions
6. Investigate a number of other requirements for seeding operations (e.g., the role of numerical models, potential use of airborne seeding, deployment of instruments, etc.)

During the same time as the initial assessments were being performed, permitting requirements needed to progress so that the ground–based seeding generators and other instruments could be installed on federal lands without undue delay. Specific sites needed to be chosen and then surveyed for environmental impacts, and more generally, potential impacts of seeding on federal lands needed to be addressed. A baseline study on silver concentrations (determined through sampling watersheds and snowpack) found that even in the worst–case scenario, no significant rise in baseline silver concentrations would occur under the proposed seeding, due to the extremely low concentrations of silver iodide used for seeding and the high natural background levels of silver in soil and water.

At the end of 2007, the two southern ranges were established as targets for the randomized seeding experiment to be conducted using ground–based seeding (with updated generator locations) from 15 November through 15 April. High–resolution precipitation gauges were deployed for the snowfall measurement, along with auxiliary instruments to assist in determining seeding conditions. Operational–like seeding would take place in the Wind River Range, with somewhat less–than–ideal generator locations, and evaluated using current instrumentation (e.g., SNOTEL data) and numerical modeling output.

Design of the Randomized Experiment

The final experimental design includes the following elements:

1. Target areas near the crests of the Medicine Bows (near the Brooklyn Lake SNOTEL) and Sierra Madres (near the Old Battle SNOTEL)
2. A cross–over design in which one range is randomly chosen for seeding and the other becomes the control, resulting in paired cases
3. A seeding period of four hours with a "buffer" period of four hours to clear the areas of seeding material
4. high–resolution precipitation gauge data for determining the seeding response, averaged over two sites within each target area
5. Two control gauge sites (unaffected by seeding) in each range to help describe the natural variability in precipitation between target areas and between snowfall events
6. Seeding conditions occurring in each range simultaneously (i.e., cold enough temperatures for the seeding material to work, wind flow of the right direction for the seeding generators to affect snowfall in the target areas, and the presence of SLW)
7. Enough cases to provide statistical significance of the results

Resources used for carrying out the experimental design include 16 ground–based seeding generators (8 in each range), 20 precipitation gauges at 8 sites (with redundancy at each site and some experimental gauges), 12 weather stations (at each gauge site and four at generator sites), two microwave radiometers for detecting SLW, a radiosonde unit ("weather balloon" for measuring temperature and winds at cloud heights), and a numerical forecast model cycling every three hours with updated observations (WRF RT–FDDA). The seeding generators, radiosonde unit, and one radiometer are operated by WMI, and the precipitation gauge network, one radiometer, and forecast model are operated by RAL/NCAR.