WMOD00063

• Applicant: Steven M. Hunter
• Sponsor Name: California Energy Commission
• Project Name: Optimizing Cloud Seeding for Water and Energy in California
• Title: Optimizing Cloud Seeding for Water and Energy in California
• Prepared For: California Energy Commission
• Prepared By: Steven M. Hunter
• Date: 3/31/2006
• State: CA
• Country: United States
• Weather Modification - Doc Type: Report

due to seeding.

Merced River at Pohono Bridge

1.2E+06
~ 1,OE+06
:!. 8,OE+05
~
0 6,OE+05
;:
E
~ 4,OE+05
~
~
-
(f) 2.0E+05

.

.

..i

. .~.
, .
,...-.-.~. - t.
, . .
...... , ......
-. . . ....~ .
, . .
. . +-
.
.

.

'.

.

.:+=

. I
.' .
.

1930

1940

1950

1960

1970

...
I I'
.
. .
'.,
'T'
1990 2000

O.OE+OO
1920

r--"
I ...
..' .
---+-'1
"
!

1980

Water Year

Figure 2, Streamflow at the Mcrccd River near Pohono Bridgc control site, Mil!>, as a function
of\vatcr year.

Beeausc of thesc othcr non-random clTccts on streamflow, the attempt to predict seeding
cffccts on an annual basis in this evaluation yielded misleading rcsults. Because of the lack of an
cxact model that dcscribes the behavior of thcse non-random cffects, the variability of
streamflow cannot be taken into account propcrly in predicting annual strcamflows. If one were
to actually believe in the annual evaluation of effects, onc would have to conclude that sceding
frcquently decrcases as well as incrcascs thc streamflow and by fairly si:l..able amounts. It is
unlikely that seeding acts in this manncr to produce these kinds of highly divergent effects. Such
results arc inconsistcnt with the seeding conceptual modcl upon which the operational sceding
program is based or any other knO\\l1 seeding conceptual model.
The fact that these highly divergcnt cffccts have occurred should be rcason enough to
doubt the appropriateness of using the historical regression mcthod on the data in this evaluation.
In addition, there is reason to doubt that the assumption that the relationship between thc target
and control station dcrived from the historical period is valid during the operational period for
these data. Brownlee (1960) has shown that this analysis method requires homoscedasticity
between the data in thc historical and operational periods and this is not the case here. A test of
the equality of variances betwcen the historical and operational streamt10ws indicatcs that they
arc statistically differcnt. Dennis (1980) warned that the most serious difliculty with the
historical regression method has to do with the lack of stability of the target-control relationship,
as is the case with the data in this evaluation,

5. E,'alllation of all targets by ratio statistics

38