WMOD00534

• Title: Reanalysis of "Generalized Criteria for Seeding Winter Orographic Clouds"
• Date: 5/5/1980
• Weather Modification - Doc Type: Report

Reanalysis of "Generalized Criteria for Seeding Winter Orographic Clouds"

DONALD RaTTNER, LARRY V ARDIMAN AND JAMES A. MOORE

Water and Power Resources Service, Department of the Interior, Denver CO 80225
10 December 1979 and 5 February 1980

ABSTRACT

A serious systematic bias of the data set used in the original Vardiman and Moore (1978) work was
detected while performing a continuing investigation of seeding "windows." This bias occurred because
a large number of no-seed cases were used in the Climax I and Climax II data sets that were not within
the original strict randomization. Removal of this systematic bias produced a significant change in both
the overall and stratified results. The nature of the bias and its impact on the original results
are discussed.

1. Introduction

Vardiman and Moore (1978) presented the results
of a study to develop' 'Generalized Criteria for Seed-
ing Winter Orographic Clouds. " A number of physi-
cal stratifications were discussed in that paper which
indicated the bounds of those meteorological and
topographical conditions that were likely to produce
positive and negative seeding results (seeding
windows). The "windows" were developed by
stratifying cases according to four general physical
categories for precipitation to develop and fall on
the mountain barrier: time available, water avail-
able, nuclei available and mixing available. The
study showed that positive seeding effects occurred
at the mountain crest under stable or unstable con-
ditions when a "crest" trajectory was present,
moderate-to-high moisture was present, and the
cloud-top temperature was between -10 and - 30oC.
Decreases occurred at the crest for unstable clouds
with a "blow-over" trajectory, low cloud moisture
and cloud-top temperatures lower than -30oC. The
precipitation for upwind and downwind regions of
a barrier was also increased or decreased depend-
ing on stability, trajectory, cloud moisture and
cloud-top temperature. A number of other conclu-
sions and implications were discussed.

Vardiman and Moore assembled the data for their
generalized seeding criteria study with physical
reasoning foremost in their minds. In doing so they
took into consideration 1) contamination of un-
seeded days by preceding seeded days, 2) the actual
operation of seeding generators and not just seed
decision logs, 3) the existence of cloud as deter-
mined by rawinsonde, and 4) the presence of winds
with an adequate normal component to the barrier
to cause lifting over the barrier. All this physical
reasoning was sound and would not be expected to
have caused bias in their results. However, fol-
lowing publication of that paper additional studies
using the same data set in a continuing investiga-
tion of seeding windows revealed that a serious prob-
lem did, in fact, exist. The problem arose because
the seed and no-seed labels were not assigned to
soundings according to strict project randomization.
Comparison of published results with the results
yielded by adherence to the strict randomization
showed that stratifications based on contamination
and actual seeding generator operation produced
only minor differences; whereas the inclusion of
non-experimental days in the non-seeded population
had an overriding effect on the results. The problem
was found in their analysis of the Climax Experi-
ment. An examination of the other projects showed

-"

j
)
I
I
I

-]

"

~!