WMOD00523 - Laserfiche WebLink

Home Browse Search

increase in IPC for the seeded distribution, with most of the seeded minus nonseeded differences in the range of 10 to 20 crystals per liter. These differences are very similar to TAR 2D-C probe results from previously published propane seeding case study experiments which had lower propane rates released from the HAS. The probability that the seeded and nonseeded differences were caused by seeding, as suspected, cannot be stated because of the limited sample size and lack of crosswind control observations. All natural IPCs, averaged over the 40 minute EU periods, were between lO to 90 per liter whatever the SWE observed by nearby gauge GTR. This is in good agreement with a previously published much larger data set which showed that all but a small portion of natural TAR IPCs were between 10 to 200 per liter when precipitation was 0.0 I inch per hour or more. These findings demonstrate that natural IPCs are seldom low enough to prevent masking of seeding effects in case studies. Another important implication is that natural snowfall of detectable (and meaningful) rates requires a minimum IPC of at least lOpeI' liter on the plateau top, A minimum of 20 per liter is probably a more realistic threshold, Seeding attempts which cannot achieve higher IPCs are unlikely to produce useful snowfall, especially since seeded crystals tend to have more limited growth times and, therefore, smaller masses than natural crystals. Summary of Randomized Experiment To summarize, the essential features of the Utah randomized experiment were: . Used propane relea,ses to treat SL W clouds which were often too warm to be seeded with AgI. . Conducted using a seeding site already demonstrated to provide routine targeting. . Used a fully automated operation with seeding or placebo treatment triggered by SL W detection. . Experimental Units (EUs) blocked into pairs with one EU randomly seeded. . Target precipitatiort gauges operated along the expected seeding plume trajectory. . Control gauge operated crosswind to monitor natural precipitation variations. . Supporting wind and other observations collected for partitioning EUs into similar subpopulations. . Applied three sophisticated statistical testing procedures, all using the control gauge as a covariate to substantially reduce the variance caused by natural snowfall fluctuations. Summary of Statistical Conclusions Results of statistical testing of the available ElJ sample population were considered significant and "strongly suggestive" if the probability of indicated seeding-caused differences between seeded and nonseeded EUs not being by chance was at least 95 out of toO (a-level = 0.05). Results were considered "somewhat suggestive" if the same probability was from 90 to 95 out of 100 (a-level = 0,06 to 0.10). Results to follow discuss the four gauges consisting of (1) GSC midway between the seeding location and (2) GTR, the primary intended target gauge on the west edge of the plateau top, (3) GSO located I km due south of GTR also on the plateau top, and (4) GDN, the only gauge downwind ofGTR, about 6.5 kIn plume travel distance from the seeding site. Because results vary among the statistical tests, each having different strengths and weaknesses, the predominant findings of two or more tests will be noted below. The results of statistical testing of the Utah randomized experiment can be briefly summarized as follows: . Tests ofthe entire sample of98 EUs without any partitioning were strongly suggestive of a real seeding effect of increased snowfall at gauges GSC, GTR and GSO. . Results for the 98 EUs were inconclusive at the downwind gauge GDN meaning no valid conclusion can be made whether seeding was effective there or not. . Results for the 69 ElJs partitioned by southwest quadrant seeding site winds, known to transport seeding plumes by the primary target, were also strongly suggestive of snowfall increases at the GTR and GSO gauges on the plateau top. Resuhs were somewhat suggestive at the GSC gauge closest to the propane dispensers, and also at the downwind gauge GDN furthest from the seeding location. . Comparisons of the wind direction- partitioned 69 EUs with all available 98 EUs n~v'ealed 11 4iILu-.i-. .