|
<br />'-'':--'::~-,.~..
<br />
<br />"-~,,,",,",'"-
<br />
<br />flueck: field Experimentation in Weather Modification
<br />
<br />_ Scott Elizabeth L., and Wells, Marcella A. (1969), "St&-
<br />tisti~ in Meteorology," RevUw of the International 8tatiatica In-
<br />stitute, 37, llIH48.
<br />
<br />R. D. ELLlOTT*
<br />
<br />77
<br />
<br />--, Scott, Elizabeth L., and Wells, Marcella A. (1973), "Down-
<br />wind and Upwind Effects in the Arizona Cloucl-Seeding Experi-
<br />ment," Proceedi1lfl3 of the National Academy of 8cit:na3, 70, 357-360.
<br />
<br />Comment
<br />
<br />More collaboration between meteorologists and statis-
<br />ticians in the design and evaluation of weather modifica-
<br />tion experiments is a continuing need. This article is
<br />welcome in that it exposes statisticians to some examples
<br />of why this need exists, and to some of the difficulties of
<br />experimenting with the atmosphere. The case for well-
<br />designed and carefully conducted field experiments in
<br />cloud seeding cannot be overe~phasized. The potential
<br />for broad and effective application of weather modifica-
<br />tion asa water management tool is great and requires
<br />development. Present applications are small in scope and
<br />involve a considerable risk of failure through lack of
<br />understanding.
<br />In Section 2, Braham refers to the two strategies of
<br />"qptimum nucleus strategy" and "dynamic seeding." A
<br />more frequently used term for the first strategy is "static
<br />seeding." Even though Braham's term may be more
<br />accura.te, the re~der should be aware of the other usage.
<br />In connection with the discussion of Project Whitetop,
<br />the reader should know that the seeding rate of 2.7
<br />kg. AgI/hr. is sufficiently great that; in the light of what
<br />we now know, some "dynamic seeding" response to the
<br />treatment could occur. Also, the "patrol mode" of
<br />seeding upwind of the target area is not now favored for
<br />the seeding of summer cumulus clouds. In discussing
<br />Project Whitetop radar echo top categories, it should be
<br />mentioned tha.t if some dynamic response had occurred,
<br />then cloud tops would have been extended upward, shift-
<br />
<br />· R.D. Elliott is Chief Scientist, North American Weather Con-
<br />sultants,600 Norman Firestone Road, Goleta, CA 93017.
<br />
<br />JOHN A. FLUECK*
<br />
<br />ing some cases to the higher category; i.e., the seeding
<br />trea,tment would have affected the categorization, a fact
<br />to be considered carefully in the interpretation of the
<br />table.
<br />Readers should understand that application of the
<br />Simpson- Wiggert model of cumulus cloud dynamics in-
<br />dics~tes that there was a natural bias against vertical
<br />cloud development in the seeded-day sample air masses.
<br />But the evidence shows that the dynamic effect of seeding
<br />overcame this, and even added further vertical develop-
<br />ment to the cloud for a total lift of 6,000 feet.
<br />In the discussion of the San Juan project (Section 4.2),
<br />a sEmtence from the original text was deleted. It read,
<br />"Suitable clouds frequently were not forecast; and when
<br />forE:cast, frequently did (not) coincide with the 24-hour
<br />time blocks specified in the design." (The "not" should
<br />have appeared in the original.) This statement should
<br />not have been dropped since the 24-hour cloud-top tem-
<br />pen!l.ture forecast constraint accounts more than any-
<br />thing else for the difficulties of that project. It ended up
<br />as more of a test of forecasting skill than of cloud seeding
<br />suc1cess, and this fact should be noted by those charged
<br />with the design of future projects.
<br />Reanalyses certainly introduce an element of multi-
<br />plicity no matter how carefully they are carried out. In
<br />any mechanism as complex as the atmosphere, one can,
<br />after a prolonged search, "discover" some strange and
<br />contradictory pseudo facts. Still, there is a need for re-
<br />ans.lyses when experiments go awry. Whitetop broke all
<br />records for the number of reanalyses.
<br />
<br />Comment
<br />
<br />1. INTRODUCTION
<br />
<br />It can be said that statistics in the broad sense is the
<br />"science of doing science," and as such Professor Braham
<br />
<br />. · John A. Flueck is Professor of Statistics, Department of Statis-
<br />tiCl!' and Director, Data Analysis Laboratory, Temple University,
<br />Philadelphia, PA 19122.
<br />
<br />
<br />
<br />has raised a number of interesting and stimulating issues
<br />which penetrate t,o the core of proper science. The tempta-
<br />tion is strong to comment in some depth on all of these
<br />important issues (e.g., how to properly assess statistical
<br />significance in the face of "multiple looks" at the data,
<br />how to handle the status and related problems of "ap-
<br />
|