Laserfiche WebLink
<br />1~33 <br /> <br />six, plus three, plus one minus two or what- <br />ever the case might be. Simply a study of <br />what was the amount of precipitation received <br />over the area in relation to the target area <br />.as such. <br /> <br />I <br /> <br />This, of course, doesn't prove anything. <br />The amount might be above and it might be <br />below and of course we can't prove anything <br />and say that unless a one percent change has <br />happened this indicates above normal or be- <br />low normal precipitation. In order to get <br />that we would have to go through a target <br />control type of analysis which is shown on <br />the next chart. In general this amounts <br />simply to this - that we would compare the <br />target area with an adjacent control area <br />where we would be reasonably sure that seed- <br />ings would not have taken place during this <br />summer. Based on past records which would <br />have to go back ten or fifteen years, we <br />would compare the monthly amounts of preci- <br />pitation received in the target area with <br />those received in the control area. Based <br />on that information we wouldaaw then a re- <br />gression line between the control area and <br />the target area as shown in the center line <br />of this particular chart. This then would <br />give the average relation between precipi- <br />tation received in the control area and <br />the target area. In addition to that aver- <br />age relation we could also draw confidence <br />limits on this particular line. The upper <br />limit would represent a line at which we <br />would say that unless a one in twenty chance <br />has occurred if precipitation falls above <br />this line, there in fact has been an in- <br />crease; if it falls below this line, again <br />unless a one in twenty chance has occurred, <br />there would have been a decrease. So we <br />hope that by the time 1959 data is plotted <br />on this, we would have an indication of <br />whether the precipitation was above normal <br />or below normal in relation to historical <br />records and in particular if it would fall <br />enough above or enough below then we could <br />apply a confidence statement to it and say <br />unless a one in twenty chance occurred, <br />there was in fact an effect due to the <br />treatment. <br /> <br />I <br /> <br />The next one involved is essentially <br />the same thing. By comparing what has <br />