WMOD00509

• Title: Physics of Winter Orographic Precipitation and it's Modification - Summary of Presentations
• Date: 10/1/1985
• Weather Modification - Doc Type: Report

not found to be significant at the 5% level. The explanation
presented by Elliott et al. (1978) was that the negative
trend was caused by the inclusion of unsuitable cases, cases
with cold cloud tops, deep convection, .and blocked airflow at
low levels. They highlighted the problem that forecasting
appropriate conditions 24 hrs in advance was not possible in
the context of the operational program. However, they
suggested that a positive 10% effect on precipitation would
have been achieved if seeding had been restricted to
appropriate conditions.
Their argument was obtained from stratification of the
experiment and division into 6 hr blocks. When they
characterized the storms according to lifted cloud top
temperature and an index based on conditional instability,
they found that there were positive effects (+54%) in stable
cases but weak negative (-6%) in unstable cases. (Cf. their
table 3, reproduced with changes below.) Cases with >-290C
cloud tops that were stable showed a 166% increase due to
seeding. From this, they argued that a positive effect could
have been achieved if only the warm stable cases were seeded.
There is an obvious problem with this analysis. If all
the categories in Table 3 are combined (and weighted by their
precipitation amounts), the result suggests that the
experiment produced a net increase in precipitation of +20%.
Table 3 from Elliott et al. (1978) is reproduced below, with
addition of the two boldface lines. The result using the
full data set is inconsistent with the net negative effect on
precipitation obtained from the 24 hr evaluation.
The source of this spurious positive effect is the
elimination of a set of control cases without elimination of
the corresponding seeded cases. The bias is illustrated by
the contrast between their Figure 3 (which shows the initial
data) and Figure 4 (which shows the seed/no-seed ratios after
exclusion of control cases on which either a pooled mode of
seeding from an earlier seeded period was possible or high
ice nucleus concentrations were observed in the target area).
Although Fig. 3 shows predominantly negative effects, almost
all effects in Fig. 4 are positive. A similar test was not
applied to the seeded cases. To treat the data fairly, the
same exclusion should be applied to seeded and unseeded
cases, because otherwise a departure from mean precipitation
in the excluded cases will cause a spurious seeding effect to
appear. In this case, the 22 control days excluded from the
analysis were cases following blocked airflow (likely to have
greater than average precipitation because the blocking only
ended in later stages of the storms when precipitation also
increased) or control cases with detected nuclei in the
target area (also likely to be biased toward heavy
precipitation by virtue of their stronger low-level wind
speed and associated lifting over the barrier).

10