WMOD00519

• Project Name: Weather Modification Programme - Precipitation Enhancement Project
• Title: A Review of the Hydrological Aspects of Evaluation of Precipitation Enhancement
• Date: 5/1/1977
• Weather Modification - Doc Type: Report

8

(c) tho population parameters are known;

(d) the variances of annual rivGr.flows in thE: nonseeded and seeded
periods are equal.

An al terna te t-test can be used if the varj,ances are unequal (BroHnlce, 1961).

35. Target-control X2-test: The object of this test is to compare sets of
annual river flo..lS for nonseeded and seeded_ periods from the target Hatershed ..,ith
those from an unseeded control ..ratershed. The basic assumptions underlying this
test are:

(a) the tareet and control annual river flovs are not stochastically
independent but are highly correlated, with the population corre-
lation coefficient characterizing their dependence; and

(b) the target and contJ:'ol annual river flows are bi varia te normally
distributed.

Another assumption that should not be too easily taken for granted is that it is
possible to treat the target basin vithout inadvertently treating all or part of
the control. Particularly in the complex air circulations which characterize mountain
basins, the potential for contamination should be checked Hith diffusion studies, if
possible.

36. TurgE::J--control T2_t"~st: '['he object and aSSU1Tlptions of this test are
similar to the one above with the exception that the population parameters are con-
sidered unknown, requiring the use of estimators based on samples instead of the
true parameters. This implies substitution of the F-distribution for the- Chi-square
distribution employed in the previous test.

37. Condi tional ta.rget-control tl/t2-test: The basic idea of this test is to
develop the joint distribution of target and control annual rj.ver flo',rs in the seeded
period by usine; the sample statistics as best estimators of unkno..m populat:Lon para-
meters. Four Bets of river floH data are required: hro samples of annual floHs
from tlle target and two samples from control va tersheds.

38. Jllorel-SeytouY (1972) characterized the six tests studied by Markovic as
simple but insensitj,ve, requiring long periods of observation prior to seed:Lng and
during seeding if satisfactory results were to be obtained. He designed a target-
control conditional Student's i-test to be applied to a test variable which is a
linear combination of runoff variables. The pov,,,r of the test is maximized by the
proper choice of weights for the' various runoff variables, With this test, the
chances of detecting a 10 per cent increase in precipitation over a 5-year period
are computed to be slightly better than one in bllO, but if the average increase is
15 per cent, the chances of detection are 17 in 20.

VI

MEASUREl1EWl' OJ<' SNOHF1l.L1 AS AN AID TO EVAWATION OF PHECIPITATION IiJffiANCEr.JEN'J:

I
I

39. Although the site selection process for PEP hR,8 gone forward with a clear
preference for areas Hhere the pre(:j,pi to,tio::1 faLls as rain due to concern about the
imprecision of sno_dull measurements, a brief revi eH of the techniques of snOH
measurement may be useful to planners of other enhancement efforts. A comprehensive
discussion of the subject has been given by Go'rtska (1964).

40. Measurements of snoHpack water equivalent have proven to be a quite useful
basis for forecasting the streamfloH expected from snowmelt, In a study (Grant et al.
1969) relating 1 April snow _vater equivalent on Belected snOH COurses to April-
August streamflow for 70 representative staU.ons in the Upper Colorado R:i.ver Basi.n