OCR Text |
Show categories, each corresponding to a three-hour period. p The quantity where r P s 3 = 3S 100o/o P5 X = 3 hour seeded period precipitation total, and = 3 hour seeded + 3 hour unseeded precipitation total was computed and plotted for each station, for each six hour block, and the resulting field of r-value was contoured. In the absence of seeding effect, or with insufficient data measurement capability, the mean value of r should trend to 50% as the number of cases become large. Statistical testing of the differences in mean r-value for data points within the calculated area of seeding effect compared to the mean value for those outside the area of effect yielded inconclusive results. A sum of squared ranks test (Mielke, 1967) was made for each station by pooling 3 hour seeded and unseeded precipitation totals together in a mixed sample, and stratifying by cloud top temperature and 550 mb temperatur~This yielded interesting patterns of indi_cated seeding effect, with Rabbit Ears Pass (the most frequently manned and serviced station) showing statistically significant positive _seeding effects for cloud top temperature -20°C, and negative effects for colder cloud tops. (This statistical routine and the results are described in more detail in Section 5. 4. ) The good results_obtained by this method of testing on only one season of data are quite encouraging, especially in view of the existence of three large sources of data and stratification error. These are: ( 1) a large error band in precipitation rate measurements, (2) a large natural variance in precipitation rate$ over the si.x hour operation time increment, due to a preferred frequency of convective band occurrenc~ at 2. 5 t _o 4 hour intervals, and (3) the assumption of a stationary, accurately computed area of seeding effect for the entire period designated as the seeded three hours. From the detailed case study type of analysis positive seeding effects on precipitation rate were frequently clearly evident for individual cases. Statistical testing of silver content in precipitation for seeded versus unseeded samples showed the difference in means significant at the 0. 1 % level. The seeded sample mean was twice as h~gh as that for the unseeded sample. Also, from replica series made during five operational days, and analyzed for AgI particle participation as active ice nuclei,· there is strong evidence that many small hexagonal plates were initiated by Agl particles during the 292 |