WMOD00061 - Laserfiche WebLink

Home Browse Search

midway from (he liAS to the TAR on Ihe west edge of the plateau top. GTR near Ihe TAR. and GSO I kIn due south (crosswind) ofGTR. These three are referred to as the "core" gauges, located wilhin 2 km or less of one another. The downwind gauge GO!" had a P-value of 0.09. somewhat suggestive of se~ing effectiveness. but not meeting the 0.05 critenon. These test results, from the entire EU populalion, provide a strong statistical suggestion that propane seeding increased the "seasonal" 3.5 month experimental period snowfall on the windward slope and upwind portion of the plateau (Op. The "seeding signal" is less obvious althe gauge furthest downwind which had a lower likdihood ofroutine1y being under the seeding plume because of variations in wind direction. Table I. Summar)" for all a~'allable 91\ [Us wilh no partitioning: 47 pairs. 50 seeded and 411 nomcedoo [Us, Pooled R,- Target- randomized l-tailcd95% Target GNO Sample I-tailed Conlidence Gauge Correlation \1DR P-~alue Inh:rvll[ GSC 0.85 1.22 0.06 1.02.:r; GTR 0.88 1.22 O.O~ 1.04 - 'Y. GSa 0.89 1.23 O.oJ 1.06- or: GDN 0.84 1.21 0.09 0.98- x \\'hereas a significance test can implYlhet'e isa seeding effect, contidence intervals can quantify the possible range of efTect which seeding may cause. Rerandomization.the repetilive simulation of the experiment using the same data but with dilTerenl random seeding decisions. was required for the deri\ation of P-\alues and contidence limits shown in Table I. The same set of 1000 rerandomized :>.tDRs used to deri\'e the P.values was applied to estimation of confidence limits. For example. a test of significance using u = 0.05 might estimate from the MOR thatlreatmcnt provided an additional 10'% SWE. The upper limit ofthecontidence interval for the l.tailed application, which assumes sCt.-ding can increase but not decrease snowfall, cannot be specified. lienee, upper limits in the lables are given by the "-,:;" symbol meaning infinilY. The lower limits for the core gauges in Table I range from 1.02 to 1.06 corresponding to minimum precipilation increases of2. 4 and 6%. respectively. The lower limit for gauge GDl'\ is 0.98 meaning thai it is possible that decreased precipitation could haH resulted there, but by no more than 2%. That gauge also has Ihe highest P.value (lowest significance). All the confidence interval values are valid althe 95% level. It is reiterated that these are minimum possible \alues and actual increases may be substantially higher. The sample ~IDRs range bel\n-en ].21 and ].23 implying about 22'l'~ more precipitation for Ihe st.'eded EUs than for the nonsecded control EUs for Ihe enlire available sample populalion of98 EUs ",-ith no panitionmg. Paffitionin~ Applying stalisticallcsts to EU groups which meet one or more meteorological crileria can give insight to Ihe crt.odence of physical hypotheses. This is tenned "partitioning" or "stratitication." The most obvious and physically reasonable partitioning of the available EUs was by liAS wind direction. :-';0 real seeding elTect can exist if the wind did no! transpon the seeding plume owr the target gauges. Considerable prior plume lracing with tracer gas and AgI was conducted with the same liAS seeding site. primary intended targel, TAR. and its associalt.-d gauge. CiTR. These earlier dala sets were analyzed in delail for this experiment. They showed that plumes released at the liAS when winds there were trom about 170 10 270 deg routinely reached the GTRff A R area with Iypical .2 km across.the-wind plume widths by that downwind distance. Plumes released when liAS winds had any northerly component did not targelthat area but were transported funher south. The ob\ious wind panilion was tor liAS median wind direClions of270 deg true or less. as the lowest EU value wa.~ 169 deg. This wind direclion panition provided a lotal of 32 pairs. and a total of 35 seeded plus 34 nonseeded or 69 EUs. The target-control correlations of Table 2 are almost as high as those of Table I. Strongly suggesti\'e P-valuC'S of 0.03 and 0.05 are shown in Table 2 for gauges GSO and GTR. respectl\e1y. Oo"'-llwind gauge GDN had a respectable 0_06 and 7