Laserfiche WebLink
<br />wrong because the two years are simply extremes in the natural variation of 30 nonseeded <br />winters. These examples further illustrate the danger of attempting to conclude anything <br />about a single winter's seeding operation based on this type of target-control analysis. <br /> <br />Figure 4 is a plot very similar to figure 3, but with sites 3 and 8 removed from the <br />calculations. As previously discussed, those sites might have been contaminated by the <br />seeding. Assuming they were affected by seeding in the same direction as the target <br />(presumably increased snowfall), inclusion of sites 3 and 8 might reduce any seeding signal. <br />Figures 3 and 4 are very similar, and the R-value is not significantly reduced by excluding <br />sites 3 and 8. The figure 4 data point for the 1993 winter is slightly more above the <br />regression line than in figure 3, with a 12.6-percent (3.7 inches) departure. This departure <br />still falls well within the natural variation from winter to winter. <br /> <br />50 <br /> <br />o <br /> <br /> <br />~ <br />...c 40 40 <br />u <br />c <br />""--'" <br />w <br />~ 30 ,30 <br />(f) <br />-+- <br />Q) <br />(J) <br />L <br />o 20 20 <br />l- <br />e <br />0 <br />Q) <br />2 10 10 <br /> <br />o <br /> <br /> <br />o <br />50 <br /> <br />o <br /> <br />10 <br />Mean <br /> <br />20 <br />Control <br /> <br />30 <br />SWE <br /> <br />40 <br />(inch) <br /> <br />Figure 4. - Similar to figure 3, but without control sites 3 and 8. <br /> <br />14 <br />