Laserfiche WebLink
<br />DECEMBER 1983 <br /> <br />ARLIN. B. SUPER AND JAMES A. HEIMBACH, JR. <br /> <br />2009 <br /> <br />respectively) and both depart in the direction expected <br />if seeding increased snowfall at S 1 and S2. Both the <br />1971 and 1972 data points are 18% above the values <br />predicted by the regression line. However, the non- <br />seeded 1974 data point has an even greater percentage <br />departure from the regression line at 22%. <br />An effort was made to improve the predictability of <br />the target-control relationship by including additional <br />courses. Correlation coefficients were calculated for all <br />combinations of 2, 3, . . . , 6 of the courses with <br />individual correlation coefficients ~ 0.88 in Table 12. <br />Of the 57 possible combinations, the highest coefficient <br />achieved was 0.965 resulting from the means of Sa- <br />cajawea, New World, Shower Falls and South Fork <br />Shields correlated with the means of the two target <br />courses. These seasonal means are plotted in Fig. 6 <br />together with a least-squares regression line fitted to <br />the 13 nonseeded winters and the :t 1.96 standard errors <br />of estimate (dashed) lines. <br />The data points for the 1968-69 and 1969-70 win- <br />ters are near the regression line, suggesting little net <br />change in snowfall due to seeding during those winters. <br />However, the 1 March data points for the 1970-71 <br />and 1971-72 seeded winters are 4.1 and 4.7 inches <br />above the regression line, respectively, corresponding <br />to percentage departures of 17 and 25%. These de- <br />partures represent standard errors of 2.8 and 3.1, im- <br />plying a significant positive deviation from.the non- <br />seeded winters. <br />The mean and median double ratios were calculated <br />between the three control gages and the 12 gages of <br />Zone 1, as well as for the single gage nearest the Bridger <br />Bowl and Maynard Creek snow courses, for all ex- <br />perimental days of the 1970-71 and 1971-72 winters <br />respectively. The same was done for the two winters <br />combined. These values are given in Table 13. <br />It can be seen that the mean double ratio for the <br />gage nearest the snow courses suggests twice as great <br />an increase for the 1970-71 winter, and is in close <br />agreement for the 1971-72 winter, when compared <br />with the snow course estimates. The median double <br />ratio estimates for the single gage depart markedly <br />from the snow course estimates. Agreement is better <br /> <br />TABLE 13. Single and two-winter double ratios. <br /> <br />Double ratio between 3 <br />control gages and: <br /> <br /> 12 gages of Gage 1.5 km NE <br /> Experi- Zone I of Maynard Creek <br /> mental <br />Winter days Mean Median Mean Median <br />1970-71 86 1.10 1.63 1.35 2.33 <br />1971-72 99 1.26 1.19 1.24 0.69 <br />Both winters <br />combined 185 1.15 1.24 1.25 1.14 <br /> <br />~',f:.~_"_ <br /> <br />when median double ratios are based on all 12 gages <br />of Zone 1, but the 1970-71 value of 1.63, implying a <br />63% increase, is still far above the snow course estimate <br />of a 17% increase. <br />When the large data pool of 185 experimental days <br />from both winters is considered, the zone and single <br />gage double ratios, both mean and median, all fall in <br />a range suggesting 15-25% increases. This should be <br />considered somewhat of an overestimate because some <br />storm days were not included in the randomized pro- <br />gram (e.g., days with "busted" forecasts). These esti- <br />mates are'in good agreement with the snow-course <br />estimates. Thus, both precipitation gage and snow <br />course data suggest that seeding may have increased <br />seasonal snowfall in Zone 1 by ~ 15%. <br /> <br />9. Summary and recommendations <br /> <br />The Bridger Range winter orographic cloud seeding <br />experiment has been analyzed using upwind and <br />crosswind precipitation gages as controls and recently <br />developed statistical approaches. Because this was an <br />exploratory experiment with post hoc analysis, the re- <br />sults cannot be considered scientifically conclusive. The <br />probability values presented should be interpreted with <br />caution due to potential multiplicity problems. At best, <br />the results can be considered strongly suggestive. <br />The statistical analysis suggests that seeding increased <br />snowfall in the intended target area and sometimes <br />further downwind as well, when the temperature near <br />the top of the Main Ridge was colder than about -90C. <br />This apparent increase was found in partitions using <br />the 700 mb temperature from radiosondes, or the tem- <br />perature measured by thermograph at 2595 m ele- <br />vation. <br />The apparent increase in target area precipitation <br />was found for the 100 experimental days with mean <br />Ridge temperatures ~ -90C during the 1970-71 and <br />1971-72 winters combined. It was also suggested in <br />the same data but with each winter. examined sepa- <br />rately. Further, the same suggestion was found in the <br />1969-70 winter dataset which was not totally com- <br />patible with the two later winters (e.g., one rather than <br />two seeding sites was used). <br />Single partitioning by estimated cloud top temper- <br />ature, 500 mb temperature, or the component of the <br />700 mb wind perpendicular to the Main Ridge did <br />not suggest that changes in precipitation were asso- <br />ciated with seeding. A dual partitioning by cloud top <br />temperature ~ -9.0oC suggested increased precipi- <br />tation tended to be associated with colder cloud tops. <br />Double ratios suggest that, on a seasonal basis, ~ 15 <br />percent more precipitation fell during the seeded days <br />in the intended target than predicted by the control <br />gage data. When the colder days were considered alone, <br />double ratios suggested seeding-associated snowfall in- <br />creases of 50 percent or more. <br />