Laserfiche WebLink
<br />Table 3 from Elliott et al. (1978), with changes <br />(LCTT = lifted cloud top temperature; BTI = barrier index) <br /> <br />A. All stable cases fi:!g Ill! l2Q diff <br /> a. -29<LCTT<-9, BTI,O 12 21 2.21 0.84 +163% <br /> b. -29<LCTT<-9, BTI~O 22 19 1.30 0.36 +261% <br /> c. -46<LCTTS-29, BTI<O 11 11 2.47 3.49 -29% <br /> d. -46<LCTTS-29, BTI~O 8 3 2.56 1. 71 +50% <br /> e. -29<LCTT<-9, all BTI 34 40 1. 62 0.61 +166% <br /> f. Composite, a+b+d 42 43 1.80 0.69 +161% <br /> Composite, all LCTT/BTI 53 54 1.94 1.26 +54% <br />B. All unstable cases <br /> g. All LCTT, BTI<O 39 28 2.77 3.18 -13% <br /> h. All LCTT, BTI~O 27 15 2.19 1. 78 +23% <br /> i. All LCTT, all BTI 66 43 2.53 2.69 -6% <br /> ALL STABLE + ALL UNSTABLE 119 97 2.27 1.89 +20% <br />ns, nc = number of seeded and control cases; <br />ps, pc = average precipitation (mm/6 hr), seeded and control; <br />diff = difference between seeded and unseeded precipitation. <br /> <br />This bias was carried into the analysis based on 6 hr <br />periods as well. Elliott et ale (1978) state that 45 control <br />cases were removed because of contamination. other cases <br />(presumably seeded and unseeded) were removed because seeding <br />could not have had an effect, or no clouds were present. <br />Some cases were removed because seeding was not yet started; <br />this is again a potential bias since no similar test was <br />applied to the control cases. Finally, cases (seeded and <br />unseeded) for which all generators were in stable air were <br />excluded. After this data selection the ratios in Table 3 <br />were obtained. <br />The reason for discussing this in such detail is to argue <br />that the statistical evidence for a positive effect in the <br />CRBPP is flawed. The data selection procedures produced a <br />net positive effect from a data set initially having a net <br />negative effect. Although significance was not claimed for <br />either, the positive data set was then stratified in ways <br />that apparently yielded statistical significance for some of <br />the stratifications. However, this result is not defensible <br />in view of the failure to treat seeded and control cases on <br />an equal footing. I suggest therefore that the indication of <br />seeding potential cannot be taken seriously, and that the <br />negative result from the CRBPP is an important result to <br />understand if another program is to be designed in a nearby <br />(or the same) area. <br /> <br />11 <br />