WMOD00478 - Laserfiche WebLink

Home Browse Search

"""",-. - -;;;.. Mielke: Field Experimentation in Weather Modification PAUL W. MIELKE, JR.* - 87 Comment A major point made by Professor Braham is that the acceptance of weather modification concepts after an experiment involves both support by adequate .statistical evidence and compatibility with established meteoro- logical physics. A very recent example involving the Climax I and II experiments (Grant and Mielke 1967, Mielke, Grant, and Chappell 1971) dramatizes this and other points of concern in weather modification experi- ments. This example will be discussed shortly, but first, some comments are needed to indicate both the similari- ties and the differences between the Climax I and II experiments and the Colorado River Basin Pilot Project (CRBPP) in the San Juan Mountains, and their rela- tionship to the statements made by Professor Braham. In Professor Braham's article, the statement is made that the CRBPP verified the scientific hypotheses from the Climax I and II experiments. In response, it is im- portant to reemphasize that the results from Elliott et a1. (1976) which prompted this statement involved meteoro- logically defined six-hour periods nested within the 24- hour experimental units of the CRBPP. Also, excellent upper air measurements were obtained in the immediate vicinity of the CRBPP. In the Climax experiments, budgetary limitations prohibited such detailed spatial or temporal resolution in upper air information. In contrast to the CRBPP, the upper air data were based on inter- polated measurements from available National Weather Service upper air data taken at a considerable distance from the experimental vicinity and taken only at twelve- hour intervals. These interpolated results are adequate for such measurements as upper air temperatures and also are useful in describing large-scale patterns of moisture and airflow. However, they are poor in de- scribing cloud structure and other fine details in the im- mediate experimental vicinity. Also they offer no informa- tion on rapid fluctuations of any quantities. Thus (in retrospect), considering the extreme short-period vari- ability of cloud and moisture conditions as found in the CRBPP, the upper air data for the Climax experiments could not reliably index the detailed character of the cloud conditions on which the physical seeding hypotheses were based. Also, Elliott et a1. (1978) indicated a poor correlation between 500 mb temperature and cloud-top temperature (computed from available upper air mea- surements of temperature and relative humidity). Thus, · Paul W. Mielke, Jr., is Professor, Department of Statistics, Colorado State UniveI'llity, Fort Collins, CO 80523. The author ap- preciates suggestions given him by G.W. Brier, L.O. Grant, G.T. Meltesen, and J.O. Rhea of the Department of Atmospheric Science at Colorado State UniveI'llity. whiile the physical seeding hypotheses from the Climax experiments do tend to be supported by the CRBPP ans.lyses of six-hour periods, an appropriate replication of '~he Climax I and II experimental results should be based on 24-hour experimental units and 500 mb tem- perature partitions (as will be seen, such a replication is not, advocated). Also, while the CRBPP six-hour-period ans,lyses associated with stable-topped orographic cloud partitions are very encouraging, the feasibility of con- duc:ting operational selective seeding of these apparently favorable periods without contaminating periods having unfavorable conditions must be established before much optimism is warranted. Furthermore, replications of the CR.BPP results are urgently needed. As Professor Braham states, the. replication of the Climax I experimental results in the Climax II experi- ment did strengthen both the statistical evidence and the compatibility with meteorological physical reasoning. However, the following paragraphs discuss a substantial caUlae for concern about these two experiments, and also illustrate problems that can arise in making statistical inferences with precipitation data (even when proper randomization and reasonably large samples are involved). For the past several years, meteorologists at Colorado Ste,te University have advocated planned research to study possible extended-area effects from intentional we:ll.ther modification. Very recently, in connection with del:lign studies for a possible experiment of this type in the central and northern Colorado mountains, station-by- station precipita.tion analyses of the Climax I and II experimental units were made for all available hourly stations in Colorado. The resulting maps of seeded to nOlil-seeded mean precipitation amount ratios and non- parametric test statistic values plotted over the western half of Colorado indicated (for meteorological partitions such as warm 500 mb temperatures) that the Climax experimental results were part of a region-wide pattern rather than an isolated anomaly produced by seeding. In particular, these recent results cast serious doubts on the consistency of apparent effects associated with the replicated five-YE'Ar winter periods of the Climax I and II experiments. This present problem drastically outweighs previous concerns regarding the Climax experiments (cf. Neyman 19'n, Mielke 1978). These latter results indicate that at least part of the previously reported increases during seeded experimental units of the Climax I and II experi- ments (cf. Mielke et aI. 1971) were due to a type I .sta-