WMOD00473 - Laserfiche WebLink

Home Browse Search

1996 JOURN AL OF CLIMATE AND APPLIED METEOROLOGY VOLUME 22 However, due to late delivery of the generators, and subsequent operational problems, it was decided by early December to commence seeding with manual generators from a single location. Limited resources did not permit manual operation of the planned second seeding site during the winter of 1969-70, and it was recognized that coverage by a single generator site might often be inadequate. Nevertheless, as all other equipment and personnel were operational, it was de- cided to proceed so that at least a complete operational "shakedown" of all equipment and procedures might be achieved. Additional resources and improved technology made several significant advances possible prior to the 1970- 71 winter. These were also carried through the 1971- 72 winter season. By November 1970 both high ele- vation seeding sites were operational. Each was equipped with an improved AgI generator system compared with the commercial versions of the Col- orado State University (CSU) modified Skyfire gen- erator previously used. Another important change was that NH41 was used to complex the AgI rather than NaI. This improved the ice nucleation in the laboratory at higher temperatures (Donnan et aI., 1970). Precip- itation gages were installed at each seeding site. Also, while radiosonde observations were made during 1969- 70, rawinsonde capability was added before the 1970- 71 winter, providing information on local winds aloft. For these reasons, the 1970-71 and 1971-72 dataset is considered superior to al).d not totally compatible with the 1969-70 data. .Therefore, except in Section 7f, only the 1970-71 and 1971-72 dataset is analyzed herein. This consists of 185 experimental days. The BRE used the 24 h day beginning at local noon, a known minimum in Bridger Range winter precipi- tation, as the basic experimental unit. Experimental days were declared whenever a special forecast, issued each morning by the National Weather Service Office at Helena, Montana, indicated the probability of pre- cipitation at the Bozeman Airport exceeded 20% during the 24 h beginning at noon. Forecasts were made for the airport located 17 km west of the Bridger Range rather than the Range itself because the forecasters routinely predicted the weather for this location, and had verification observations available from the Federal Aviation Administration office there. The forecasters were not aware of the consequences of their forecast on each day's activity for the BRE. The purpose of the forecast was to attempt to exclude days with little likelihood of precipitation from the experiment in order to give field personnel time off. Before the AgI plume tracing and pibal investigations discussed in Super (1974) and Section 4 were com- pleted, it was considered important to provide a buffer period between seeded and nonseeded days to reduce possible contamination by the AgI. During the winters of 1969-70 and 1970-71, a buffer period of at least 3 h was provided in the following manner. Cloud and snowfall observations, obtained prior to the BRE, in- dicated that snowfall was unlikely for at least 3 h fol- lowing an observation of cloud base higher than 3000 m and cloud cover less than 6/10. These cloud base and cover criteria were designated Potential Storm Con- ditions (PSC). It is noteworthy that the highest peak in the Bridger Range is just under 3000 m, providing a convenient visual reference for cloud base. On each experimental day, whether or not seeding had taken place, PSC observations were made at 0900 (all times MST), 3 h prior to the end of the experimental unit. IfPSC did not then exist, a new random decision was used for the next experimental day and if clouds had been seeded during the present day, AgI generators were immediately turned off. If PSC existed at 0900, another observation was made at 1200. If PSC still existed, the prior random decision was carried over to the new 24 h period starting at that time, which was considered an experimental day regardless of the NWS special forecast. IfPSC existed at 0900 but not at 1200, a new random decision was made for the next exper- imental day, that is, the next day with a greater than 20% precipitation probability forecast. Thus, in this case, the next experimental day would start immedi- ately at 1200 if, and only if, the forecast probability exceeded 20%. These procedures were identically ap- plied for nonseeded and seeded experimental days. Whatever the period of actual treatment, the sole response variable in the analysis herein was the 24 h noon-to-noon precipitation amount as measured by the gage network shown in Fig. I. Random decisions were obtained by the opening of sealed envelopes which had been prepared under the supervision of Dr. Paul Mielke ofCSU, statistical con- sultant to the BRE. A simple block randomization scheme was used during the 1969-70 and 1970~7-1 winters in which no more than four seed or nonseed decisions were allowed to occur in sequence. To summarizefor the 1969-70 and 1970-71 winters, experimental units and the precipitation amount re- sponse variable were always for 24 h, beginning at local noon. The treatment period, however, was some- times shorter if PSC didn't exist throughout the 24 h experimental unit. Individual random decisions were sometimes carried over to an additional day or even days ifPSC continued to exist during the last 3 h (0900 and 1200 observations) of an experimental unit. This procedure was intended to minimize possible AgI con- tamination of nonseeded days following seeded days by providing a minimum 3 h buffer period without treatment between all experimental units based on a new random decision. Appendix A of Part I contains a complete listing of hours of AgI generator opera- tion and the basis for each day's treatment decision, whether new random decision, or carry-over due to continued PSc.