WMOD00293 - Laserfiche WebLink

Home Browse Search

suggested a 56% of snowfall increase for the seeded storms when the temperatures on the main ridge were -9 oC or colder. Later physical observations (Super and Heimbach, 1988) lent credibility to the statistical findings. Further exploratory analysis by Super (1986), limited to 6 h periods with main ridge temperatures of -9 oC and colder suggested that, "the seeding method used was highly effective during a small portion of all seeded episodes, but had little or no effect for the other periods". It is noteworthy that a double ratio of 2.22 (122% increase) was calculated by Super (1986) for the warm cloud top partition. This suggests that large increases may be realistic for particularly seedable cases and these are responsible for most of the seasonal increase detected. If the above statistical indications approximate reality, the apparent seasonal snowfall increases of about 15% in the BRE were due to large percentage increases from a small fraction of all units, and little or no increases ( or decreases) from all the rest. No evidence of decreased snowfall due to seeding was found. . . SUMMARY AND CONCLUSIONS The purpose of this paper is to examine how the duration of a statistical winter orographic cloud seeding experiment might vary with different assumed responses. Several previous investigations using Monte Carlo simulations have assumed that each treated experimental unit (case) has the same percentage response. There is increasing evidence that this simple model is unrealistic. It seems more likely that seeding results in large percentage increases from a fraction of cases that are particularly amenable to treatment, but has little or no effect on the remainder. Monte Carlo techniques were applied to nonseeded 6h data blocks from the Bridger Range Experiment in the simulated experiments. For simplicity, only the probability levels' a= 0.05 and p= 0.1 were run. Following the results of the exploratory statistical analysis of Super (1986) for the cold partition, a net precipitation increase of 66% was approximated in each experiment. However, this increase was achieved by six different assumed responses to seeding. First, it was assumed that each "seeded" case resulted in a 66% precipitation increase. Second, only half the cases randomly chosen to be "seeded" had their precipitation increased, but by 132% (2 x 66%). Similar procedures were followed until the final simulation provided increases for only 1/6 ofthe "seeded" cases, but each was given an increased of396% (6 x 66%). These simulations were calculated for both a "cold" partition (all available cases with main ridge temperatures of -9 oC and colder) and a "warm" partition (all cases colder tha~ 0 OC). It was found that the number of units required to achieve statistical significance varied considerably with the different seeding effects. For the cold partition only about one winter season would be required if each "seeded" case resulted in the assumed 66% increase. However, over 5 winters would be required if the same net increase resulted from 396% increases applied to 1/6 of the "seeded" cases. As might be expected, less time would be required if both the warm and cold cases could be successfully treated, because of the large increase in available cases each winter. However, this was partially offset by the increaseq variability associated with the warmer cases. The implications of these results are sobering. It may be that the number of experimental units required to achieve given a and ~ levels is far larger than estimated for a number of past experiments. This could partially explain the frequent finding of inconclusive results. A number of points need to be considered in designing future statistical experiments. First, strong predictor-covariate relationships with target area precipitation are necessary to reduce the cases needed to detect a seeding signal. Similarly, partitioning based on a good physical understanding is necessary to reduce the number of experimental units that have minimal response to seeding. Third, improved 36