WMOD00464 - Laserfiche WebLink

Home Browse Search

A-8 One route that can lead to such covariates is gaining a better understanding of cloud phy- sics and cloud processes. The Board's report will, we are sure, stress this point. We can only echo their stress. Another route is one where we have a special responsibility for pointing out the possibili- ties. This is a matter of empirical study of the details of rainfall in the light of rather detailed up-wind measurements. It is possible that we could learn a useful amount about how clouds and rain behave in some detail, without waiting for deeper knowledge of cloud physics and cloud processes. We urge careful consideration of this possibility and of the possibilities of implementing appropriate studies. (The extent to which such studies could be cost-effectively built into weather modification studies that would otherwise be undertaken should also be given attention.) Attention to how rainfall from different storms is distributed over the rain gauges used may offer important clues to what kind of taxonomy of storms is important. Analysis of rainfall over small parts of . storms, when combined with records of changes in wind direction (with time and height) or with detailed micro barograph records may provide similarly valuable clues. * overwhelming instances * As discussed in somewhat greater detail in the Appendix, there are situations where one event outweighs, in importance, all the other events of a season. This seems to happen most often with hail. If we bite the bullet, and admit that the effect of seeding on such very massive events may not be what we can predict from the effect of seeding on less massive events, we are in a most difficult position. Twenty successive seasons of experimentation may not be enough to give us direct evidence as to what happens to the most massive events. (Even worse, we may not have an objective way to recognize the events that would have been massive in the absence of seeding.) Usually, we will be unable to do anything but assume that the effect of seeding on mas- sive or overwhelming events extrapolates smoothly from its effect on moderate, or merely large, events. All who do this owe an obligation to their readers -- or listeners -- to explain the existence and plausibility (high or low) of this assumption. The situation where 3 or 4 events make contributions of great importance (perhaps jointly of overwhelming importance) is not quite so serious. Steps can, and often should, be taken in the statistical analysis to anticipate and accommodate such contingencies. S. Critical Tactical Issues * blindness vs. blocking * In conducting a randomized experiment where the modification is "seeding", not all days will be judged suitable for seeding. The randomization will then determine which of these suit- able (seedable) days will in fact be seeded. If the results of the randomization for the next suitable day were known (perhaps by inference) to the expert making the judgment of suitabil- ity, he would not be "blind" to the consequences of his decision about suitability. As noticed above, there is real reason for concern -- probably deep concern -- about the possible impact, through unconscious mechanisms, of this knowledge on the judgment being made. One of the main problems of designi~ weather modification experiments is one of bal- ance between the importance of such blindness of those making subjective decisions and the desirability of tidiness in the arrangement of "se,eded" and "non-seeded" days or situations. (For the rest of our discussion of selected issues, we will use "days" to mean "days or