Laserfiche WebLink
<br />. <br /> <br />5. Sample-size estim~ltes. - Estimates of the sample size needed <br />to reject the null hypotheses at specified error probabilities of <br />a = e = 0.10 were madE~ for three of the primary n~sponse variables: <br />ice-crystal concentration 5 minutes after treatment - CICS, the <br />time of first precipitating ice - TFPI, and the average number <br />concentration of precipitating ice 8 minutes after treatment - <br />PIC8. The.seeded and unseeded data for these calculations were <br />obtained as part of the randomized calibration seeding trials <br />conducted under HIPLEX. A five-part simul ation program was used, <br />involving MRPP. Each of the three variables was treated indi- <br />vidually and then jointly as a multiresponse variate. Results <br />indicate that, for the seeding rate to be used in HIPLEX-1, about <br />50 to 100 samples will be required. Based on the frequency of <br />opportunities at MilE!S City of 30 to 45 workable experimental <br />units per year, the duration of the HIPLEX-1 experiment will be <br />about 2 to 3 years. <br /> <br />HIPLEX-l is intended to 'Iast at least two seasons. Subject to this <br />restriction, it will be tenninated when one or the other of the <br />following conditions is fulfilled: <br /> <br />1. The evidence supports rejection of the null hypotheses (HOl <br />and HOII) stated in sE!ction VI (page 10) and the null hypotheses <br />for all or most of the primary response variables (page 12), and <br />the results show consistency from season to season. <br /> <br />2. The combined expE!rimental results do not pennit rejection of <br />the null hypotheses ~tated on page 10 (HOl and HOII) and <br />revised simul ation experiments based on the data from the actual <br />experiment indicate that no significant improvements can be gained <br />during two additiona"j experimental seasons. <br /> <br />B. Physical Evaluation <br /> <br />PhYSical eval uation is jl central effort in HIPLEX-l and contributes <br />directly to objectives 1, 2, and 5, mentioned in section III. <br />The hypotheses specify the expected behavior of natural and modified <br />cumul us congestus cl OUd:i, and extensive phYSical measurements wi 11 be <br />available from test events. One important use of these measurements <br />will be to carry out detailed case-study analyses for the individual <br />clouds. These analyses can produce three important contributions to <br />the experiment: (1) a rather complete physical picture of precipita- <br />tion fonnation will be provided, (2) additional covariates or strati- <br />fications could be identified, and (3) deviations from expectations <br />specified by the physica.l hypotheses, while not anticipated, may be <br />detectable first in the case-study analyses, allowing suitable <br />refi nements to the hypotheses to be introduced. <br /> <br />31 <br />