WSPC03885 - Laserfiche WebLink

Home Browse Search

003197 In addition to the direct detection experiments, a series of experiments with more limited objectives will investigate specific processes. One example is investigation of the T&D of seeding agents which can be accomplished with a specially instrumented aircraft and limited supporting observations. These T&D experiments can be conducted even when full stonn conditions do not exist, with cloud base above the terrain permitting very low level sampling. Another example of a limited experiment is study of the concentration of ice particles created by release of a seeding agent under particular temperature, moisture, and turbulence regimes. Again, such observations can be made primarily by an instrumented aircraft with no attempt to follow the seeding signature to the surface. Both the direct detection and more limited physical experiments will markedly increase understanding of the physics involved in winter orographic stonns. This knowledge is required before a statistical experiment can be properly designed The direct detection experimental approach is essential to understanding and verif'ying the chain of events from release of seeding material to resulting snowfall on a small, well- instrumented target. However, this approach does not permit accurate assessment of the precipitation changes that would result from routine seeding over the long-tenn (several winters) and over a significantly larger target area. Accurate assessment is impractical even if results are available from several comprehensive (but short duration) direct detection experiments conducted over a wide range of storm conditions. As stated by Hobbs (1975b), "The (direct detection) method should not be considered as an alternative to statistical evaluation but rather as a logical precursor which permits determination of whether a particular seeding teclmique can modify certain types of clouds and precipitation. Armed with such knowledge, the procedures and statistical design and evaluation of further cloud seeding experiments can be appreciably sharpened." The results of the direct detection experiments will be used to design and carry out statistical experiments to estimate average seasonal snowfall changes over two relatively large tarl!et ~. The statistical experiments will be designed for confirmatory analysis, but exploratory analysis will also be employed. Using the definitions of Gabriel (1981), "A confirmatory analysis presupposes a rigidly carried out sequence of experimentation-measurement-analysis. For valid confirmation or repudiation of the original hypothesis there has to be faithful adherence to the original design." He later notes that "A confirmatory analysis may, however, also simultaneously test a limited number of prestated hypotheses for confirmation, rather than only a single one." Exploratory analysis can (and should) be conducted with the same experimental data sets. Gabriel (1981) states that "An exploratory analysis . . . is one in which no predetermined course is charted. The experimenter and analyst look for leads from the data and follow them as far as they can." However, statistical inferences resulting from exploratory analysis should only be considered suggestive, as no precise probabilities of error can be assigned to them. Clearly, the confirmatory analysis is intended to provide "proof' that seeding can enhance the multiwinter snowfall over a large target area with a high degree of conclusiveness. Exploratory analysis is only intended to supplement the confirmatory analysis, pointing the way to improving understanding in the future. A final "definitive" statistical experiment for cloud seeding may be impossible to achieve. The path to knowledge never ends (consider the long-term search for a "cure" for cancer), and continued improvements in cloud seeding technology can be expected over the foreseeable 14