Laserfiche WebLink
<br />MARCH 1973 <br /> <br />H. J, MOREL-SEYTOUX AND F, SAHELI <br /> <br />324 <br /> <br /> <br />. Runoff Slatlons Utilized as Conlrols <br />lor Runoff Stallons in Tan~el <br />@ Clly <br /> <br />,.. <br /> <br />fIG, 2, Control area (Maroon Peak and Grand Mesa Region). <br /> <br />last column of Table 1 divided by 4, since N is inversely <br />proportional to the square of the increase). Based on the <br />current state of the art (Grant, 1969) it is doubtful that <br />an increase much greater than 20% could be achieved <br />in the Colorado River Basin Pilot Project. Under these <br />conditions the two-sample test is inadequate. <br />More sophisticated techniques are needed. The target <br />control concept was introduced, and different tests were <br />devised, including a chi-square test and a Student I-test. <br />In a previous study (Morel-Seytoux, 1968), a target- <br />control chi-square test was applied to the mean annual <br />or mean seasonal flows of some rivers and it was shown <br />that the number of years iV necessary to detect a given <br />percentage increase in population mean h, at the 95% <br />level of significance (two-tailed) and 50% power <br />(Brownlee, 1961), was <br /> <br />N = (1.96)2(1-p2)C}/h2=4(1-p2)CHh2, (2) <br /> <br />where p is the correlation coefficient between the <br />target and the control watersheds, and Cv the coefficient <br />of variation of the target watershed. <br />It is important to know before actually conducting <br /> <br />" <br />Iii? <br /> <br />the experiments how many years are needed to have an <br />even chance to achieve significance. Therefore, calCltla- <br />tions were performed for a few stations in the Upper <br />Colorado River Basin to get an idea of what could be <br />expected if seeding operations were conducted in the <br />area. In particular, the number of years (needed) to <br />detect a 10% (population) increase (at the 95% level <br />and 50% power) was calculated. This number of years <br />can be viewed as an average number of years needed <br />because the power was chosen for the calculations at <br />the 50% level. The assumed 10% increase is a conserva- <br />tive estimate for the region (Grant, 1969). However, <br />because the operations of seeding are randomized on a <br />fifty-fifty basis (Bureau of Reclamation, 1968), half <br />the opportunities to increase the snowpack are lost. <br />Furthermore, the operations are stopped when the <br />snowpack exceeds 150% of the "normal" conditions. <br />Again opportunities are lost. Under these conditions <br />the probability that the runoff be increased by more <br />than 10% seems rather dim. For this (plausible) <br />assumed 10% increase the number of years needed for <br />significance was calculated, based on Eq. (2) for basins <br />