Finance Committee Meeting 2009

• Board Meeting Date: 9/15/2009
• Description: Finance Sub-Committee Meeting September 15, 2009
• Board Meetings - Doc Type: Meeting

•
8 Journal of Weather Modification • Volume 41 •
evaluations of snowpack enhancement cloud major concepts about the bya.4us ed regres`the
seeding programs in the Colorado Roads*were all based measurements taken evaluation dfopetato & �pno-
over a period of several hours to several days as grams. See Gabriel (1999. 2002)tar a descrtp-
the response variable. lion of the ratio statistics methodology, and
Silverman(2001)for a more comptats description
Both statistical and physical evidence are re- of its application to operational(nos rarxbrrnized)
nuked to establish the emcees of any cloud cloud seeding programs.
seeding activity (AMS, 1998). Because the ex-
pected effects of cloud seeding are within natural The regression ratio(RR)is given by the refatiorh-
meteorological variabilly, statistical al met hods are ship, RR = SRI Wimp where tie single ratio
needed to detect a seeding egad with reason- (SR)is the ratio at the average target shear low
able certainty. Physical evidence is needed to during the operational period(TS0)to the average
establish plausibility that the effects suggested by st uamflow for the seeding target during the hie-
* the reatu0s of the statistical evaluation axed have torical period(TS„),i.e.,SR=TSolTS,,,and SR,
been caused by the seeding intervention. This is the ratio of TSo and TS„that are predicted by
study is primarily concerned with assessing the the target-control regression relationship for the
ststatic al evidence in support of the Vail opera- data over the entire period of analysis(including
Lionel cloud seeding program. The purpose of both the historical and operational periods). By
this study is to conduct an independent statistical dividing the SR by SRp,the SR is adjusted for
evaluation of the Vail operational cloud seeding effects due to natural ral dMerences in streamfiow
prewar torn water year 1971 through water between TSo and TS,,,and thereby improves the
yew 2006. The objectives of the evaluation are precision in the estimate of the target*eamiiow.
(1) to determine if cloud seeding enhanced
steamily"in the Vii Basin, (2)to provide infix- The RR resits are then adjusted for biases that
=bon on the strength of the seeding effect and can occur when operational data are compared
Ms !lortidenor Interval to slow informed judg- to ihisforiccal records in an a posteriori evaluation
•
MOM fief miff* abort its coat-effectiveness, of non-randomized seeding programs. An adjust-
fllow p physical studies that mert is made to the NI results based on multkily-
. suppdrt the plausibility of the i ng its computed P-value by an adjustment factor
(Gabriel and Peirondas. 1983).For this study,the
acVustment factor vas found be slightly fess
2. EVALUATION PROCEDURES than 2. However, an adjustment facer of 2 was
used so the calculated d values of the bias-adjusted
The bias-adjusted regression ratio was used to a regression ratio are conservative esthetes of the
aagel-control evaluation d the effect of seeding seeding effects. The results using the regression
on stoney" In the Vail River Basin.The water ratio that were &lusted for bias in this way are
yew(October-September)**endow expressed called RRA.
trt Acre-Feet (AF) served as into response vari-
able in the evaluations. Silverman (2007) de- The main emphasis in the presentation of the re-
scribed and demonstrated the capability and suits is on confidence intervals because they infer
merits of using ratio statistics and the bias- a range within which the true effect lies whereas
adjusted regression ratio,in particular,in*velvet- nut hypothesis significance tests only assess the
ing the effectiveness of operational (non- probabilly that an effect is due to chance(Gabriel,
randomised) cloud swing programs. He 2009: Nicholls, 2001). Confidence Intervale were
I!`
showed that the bias-adjusted region ratio is calculated as prescribed by Gabriel(2002).Use of
a more precise and more reliable method for confidence intervals provides Information on the
evaluating operational(non-randomized)seeding strength of the seeding effect to allow informed
programs than the badgered hlstorfdai regression judgments to be made about its wet-
methodology used heretofore. He also showed effectiveness and societal significance. In this
that the bles-adjusted regression ratio results for study, an evaluation result is considered to be
the Kings River operational cloud seeding pro- significant lifts 90 nascent confidence
gram(2007)and the Kern River operational clotid interval does not include the null hypothesis
seeding program (Silverman, 2008)were strati- value of RRA = 1 or zero percent change in
catty comparable to those from the re-randomi- strearnflow, i.e., It satisfies a 2-sided level of sig-
zatlon analysis. Following is a summary of the nilIcance of 0.10.
•
-Scientific Papers-