WMOD00545

• Project Name: Colorado River Basin Pilot Project
• Title: Test of Runoff Increase Due to Precipitation Management for the Colorado River Basin Pilot Project
• Date: 3/3/1973
• Weather Modification - Doc Type: Report

MARCH 1973

480

440

400

360
~
~ 320
0<
0
)280
~ 240
l
~ 200
l 160
~
'"
0
t-
120
80
40
0
0 20
Conlral

. Seeded Season
. Unseeded Season

328

.
11."

H. J. MOREL-SEYTOUX AND l'. SAHELI

the general combination denoted Q/ and defined by

"31
.

'9.47

1946
.

Target: San Juan River 01 Pogoso Springs, Colorado
Conlrol: Animas River 01 Howordsville. Colorado
Regression Lme Based on 29 Years of Record
SignifIcance Level: 86"/0 (one -toiled lest)
Apporent Relaflve Increase: 15%
Apparent Mean Seasonal Increase: 33,000 acre-fl.

40

60

80

100

120

140

FIG, 4. Regression line between target seasonal runoff
and control seasonal runoff.

Seasonal (April-Auqusl) Flow In Thousands of Acre-Feel

1

7ii
..
...
,
e 300
u
<(
"0
on
'0
c:
0
on
:I
0
~
I-

,5 200
it
o
G:

'5
c:
o
on
o
..
V'l

Q; 100
Cl>
~

400

. Seeded Season
· Unseeded Season

o

n

Qt* = L XiQi,
i=l

(3)

where the Xi are fixed but unknown pammeters and 11-
is the total number of target LV. Similarly, a combina-
tion can be defined for the control area, denoted Qc*,
and defined by

n+m
Qc* = L XiQi,
i=n+l

(4)

160

where m is the total number of control LV. The introduc-
tion of the variables X permits a much more thorough
minimization of N if one realizes that this problem is
the minimization of a conditional variance of a linear
combination of several LV., given a linear combination
of other LV. In multivariate analysis it is traditional
(Graybill, 1961) to denote a set of LV. as a random
vector; we thus use Q in this study of runoff. Let 1: be
the covariance matrix of Q. Suppose the vector Q is
partitioned into two vectors Qt and Qc. The covariance
matrix can be partitioned similarly into four matrices
1:tt, 1:tc, 1:ct and 1:c". Let Q/ denote the linear combina-
tion Xt Qt, where the prime indicates the transpose
operation, and Qc* denote Xc'Qc, where [X,', Xc'J is a
partition of X', a vector of parameters, fixed but
unknown weight factors. Then it is a well-known
result (Graybill, 1961) that the variance of Qt* condi-

Target: Son Juan River at Pagasa Springs, Colorado
Control: Piedra River near Piedra, Colorado
Regression Line Based on 27 Years of Record
Significance Level: 92% (one- toiled test)
Apparent Relative Increase: B %
Apparent Mean Seasonal Increase: 18,500 acre-fl.

100 200 300
Control Seasonal (April-August) Flow in Thousands of Acre-Feet
FIG. 5. Regression line between target seasonal runoff and control seasonal runoff.

400