WMOD00534 - Laserfiche WebLink

Home Browse Search

624 JOURN,~L OF APPLIED M.ETEOROLOGY TABLE 5. As in Table 4 except for unstable clouds. TABLE 3. Tabulated statistics for all projects except CENSARE with no stratifications. Projects All less CENSARE Precipitation group Crest Stratifications None M + N = 1130/885 Seeded Nonseeded M = 428/408 N = 702/477 Zeros = 48/47 Zeros = 137/55 Mean = 2,26/2.26 Mean = 2.00/2.43 Ratio (seeded/nonseeded) = 1. 13/0.93 Test statistic = 3.051/-0_064 P value = <0.001/0.952 days were declared for the Climax experiment, it becomes obvious that there was a serious systematic bias caused by the inclusion of these renegade cases. Experimental days were declared by a forecast of 0.01 inch of precipitation or greater for the Lead- ville, Colorado area. Therefore, all the renegade cases used by Vardiman and Moore were rejected as experimental days because they were forecast to be minor precipitation events. Obviously, if the rene- gade cases were rejects and were categorized as non-seeded, they would be expected to cause a seri- ous bias in favor of the seeded population. An examination of the other six projects used in Vardiman and Moore (1978) showed no additional bias from renegade cases or other physical reason- ing. The results shown in the Vardiman and Moore stratifications were changed when a reanalysis was accomplished excluding the renegade cases and assigning seed and no-seed categories by strict project randomization. 3. Reanalysis of the data The following tables are identical to those shown in the original Vardiman and MQore results but with TABLE 4. Statistics for a stable cloud, with crest trajectory, mod- erate water content and LCTTI between --10 and - 30oC. Projects All less CENSARE Precipitation group Stratifications Crest EPOS < 0.005 J g-l - 3600 < BTI < 3600 s 3 < CBWS < 8 g kg-' - 300C < LCTTI < - lOoC M + N = 178/149 Seeded Nonseeded M = 57/61 N = 121/88 Zeros = 2/2 Zeros = 22/9 Mean = 2.90/2.92 Mean = 2.40/2.94 Ratio (seeded/nonseeded) = 1.18/0.99 Test statistic = 2.357/1.431 P value = 0.018/0.152 VOLUME 19 Projects All less CENSARE Precipitation group Stratifications Crest EPOS > 0.005 J g-' - 3600 < BTI < 3600 s CBWS > 3 g kg-' - 300C < LCTTI < - 100C M + N = 230/187 Seeded Nonseeded M = 98/96 N = 132/91 Zeros = 7/7 Zeros = 26/12 Mean = 2.08/2.10 Mean = 1.70/2.10 Ratio (seeded/nonseeded) = 1.22/1.00 Test statistic = 2.321/1.008 P value = 0.020/0.312 an absolute adherence to the original randomiza- tion (i.e., removal of all renegade cases). Note that in all the results presented here the Santa Barbara Project data were eliminated due to the concern that the cloud-top temperatures were unusually warm as discussed in Vardiman and Moore (1978). The num- bers appearing first in each line of the following tables are those from the original Vardiman and Moore analysis. The numbers that appear following the slash are numbers that result from the new analysis (i.e., original/new). Table 1 shows the re- sults of seeding without regard to meteorological variations from storm to storm. The new results in- dicate a 12% increase in -precipitation by seeding. This decrease from the earlier results is due to the removal of a number of low or zero precipitation events (non-experimental renegade cases) from the no-seed sample. The number of zero events in the no-seed sample decreased from 145 to 63 cases, and the zero events in the no-seed events is now similar between the seeded and non-seeded samples. Tables 2a, 2b and 2c provide statistics on stratifi- cation by cloud-top temperature (LCTTI) for all the projects. Table 2b retained its significance but the TABLE 6. As in Table 4 except for moderately unstable cloud. Projects All less CENSARE Precipitation group Crest Stratifications 0.005 < EPOS < 0.05 J g-l - 3600 < BTI < 3600 s CBWS > 3 g kg-I - 300C < LCTTI < - lOoC M + N = 104/84 Seeded Nonseeded M = 41/41 N = 63/43 Zeros = 3/3 Zeros = 14/7 Mean = 2.28/2.19 Mean = 1.50/1.92 Ratio (seeded/non seeded) = 1.52/1.14 Test statistic = 2.004/0.896 P value = 0.046/0.368 ~