The Snowy Precipitation Enhancement Research Project: A description and preliminary results

Abstract

A gradual reduction in water from snow-melt over the past century has motivated Snowy Hydro Ltd. to pursue a wintertime cloud seeding project in the Snowy Mountains of south-eastern Australia. The Snowy Precipitation Enhancement Research Project is one of only a few cloud seeding experiments in the last two decades to employ a randomized design, and the first such randomized experiment to incorporate dual-trace chemistry analysis of snowfall as part of the pro-ject evaluation. The project design, seeding criteria, ground-seeding network, and measurement infrastructure are described, as are the general components of the statistical evaluation plan. Some initial results from analysis of physical and trace chemical measurements are presented for an extended storm period in 2006 that included five randomized experimental units. The trace chemistry results were found to validate several of the components of the seeding conceptual model, and a unique time series of tracer element concentrations appears to indicate when seed-ing and tracer materials were released. Progress during the first four seasons of the project is de-scribed, as are various findings that could affect the outcome of the project.

Full text

THE SNOWY PRECIPITATION ENHANCEMENT RESEARCH PROJECT:
A DESCRIPTION AND PRELIMINARY RESULTS
A. W. Huggins1, S. L. Kenyon2, L. Warren2, A. D. Peace2, S. P. Bilish2 and M. J. Manton3
[1] Desert Research Institute, Reno, Nevada, USA,
[2] Snowy Hydro Ltd., Cooma, New South Wales, Australia, [3] Monash University, Victoria, Australia
Abstract. A gradual reduction in water from snow-melt over the past century has motivated Snowy
Hydro Ltd. to pursue a wintertime cloud seeding project in the Snowy Mountains of south-eastern
Australia. The Snowy Precipitation Enhancement Research Project is one of only a few cloud
seeding experiments in the last two decades to employ a randomized design, and the first such
randomized experiment to incorporate dual-trace chemistry analysis of snowfall as part of the pro-
ject evaluation. The project design, seeding criteria, ground-seeding network, and measurement
infrastructure are described, as are the general components of the statistical evaluation plan.
Some initial results from analysis of physical and trace chemical measurements are presented for
an extended storm period in 2006 that included five randomized experimental units. The trace
chemistry results were found to validate several of the components of the seeding conceptual
model, and a unique time series of tracer element concentrations appears to indicate when seed-
ing and tracer materials were released. Progress during the first four seasons of the project is de-
scribed, as are various findings that could affect the outcome of the project.
1. INTRODUCTION AND BACKGROUND
The Snowy Mountains are one of the few
places in Australia with regular yearly snow
accumulation, enough to support a viable ski
industry. The spring snow melt is collected in
reservoirs that are used to provide water for
hydro-electricity and irrigation in southeastern
Australia. The snow cover and duration at some
sites in the Snowy Mountains has decreased
significantly over the past century (Hughes,
2003 and references therein) resulting in a
reduction of available water.
Cloud seeding is viewed as one means of
increasing snowfall in the Snowy Mountains
region. Australia has a long history of cloud
seeding research and operations, with initial
investigations occurring about 60 years ago.
Between 1955 and 1959, the Snowy Mountains
were the focus of an aircraft-based cloud
seeding experiment run jointly by the Com-
monwealth Scientific and Industrial Research
Organization (CSIRO) and the Snowy Mountains
Hydro-Electricity Authority (SMHEA). From this
experiment, Smith et al. (1963) reported a 19%
precipitation increase in seeded events;
however, despite these encouraging results
cloud seeding over the Snowy Mountains was
not pursued.
Recurring drought conditions in the early
1980s rekindled interest in cloud seeding in the
area. A feasibility study by Shaw and King
(1986) assessed the potential for cloud seeding
over the Snowy Mountains as positive. This
study considered meteorological and cloud
physics data over the region as well as the
ecological, community and wider-area effects of
cloud seeding. Further evaluation of the physical
and chemical characteristics of the clouds and
snowfall over the region during the winters of
1988–1989 (Warburton and Wetzel, 1992)
supported the findings of Shaw and King (1986).
In 1993 SMHEA drafted an Environmental
Impact Statement (EIS) proposing a six-year
cloud seeding experiment over a 2000 km2 area
of the Snowy Mountains. This experiment did
not proceed because of objections from key
stakeholders. An independent expert panel
report, addressing the principal objections that
had been raised in 1993, was submitted to the
New South Wales (NSW) government in 2003.
The Snowy Mountains Cloud Seeding Trial Act
(NSW) was passed in 2004 allowing Snowy
Hydro Limited (SHL) to undertake a six year
winter cloud seeding trial — the Snowy Precipi-
tation Enhancement Research Project (SPERP).
The objectives of SPERP are to determine the
technical, economic and environmental
feasibility of precipitation enhancement over the
main range of the Snowy Mountains.

The SPERP commenced in June 2004 with
meteorological data collection, verification of
cloud seeding equipment, and the development
of an experimental design and operational
procedures. The formal randomized experiment
started in winter 2005. Physical and chemical
data collected over the duration of the
experiment are used to document the conditions
in which seeding has occurred and to verify the
seeding conceptual model employed by SPERP.
This paper describes the design and operational
aspects of SPERP and presents some
preliminary results from physical and snow
chemistry studies conducted during the 2006
field campaign.
2. PROJECT DESIGN
Winter precipitation over the Snowy
Mountains is largely from moist westerly weather
systems. As these systems approach the
mountain ranges (orientated along ~30-210
degrees), the air mass is lifted and condenses to
form orographic clouds, often characterized by
an excess of supercooled liquid water (SLW).
This excess SLW indicates that either there are
not enough ice nuclei in the clouds to promote
crystal formation, or the crystal formation occurs
too slowly or late for precipitation-sized crystals
to grow before passing across the mountain
range and sublimating, or melting and
evaporating on the lee side. The premise or
conceptual model of the SPERP is to promote
the excess SLW to form ice crystals earlier than
expected naturally by introducing additional ice
nuclei into suitable clouds over the entire target,
producing a significant enhancement in the ice
crystal concentration, and allowing sufficient
time for the crystals to grow and fall out,
resulting in an enhancement of snowfall on the
ground. The target area for the project covers
approximately 1000 km2 of the main range of the
Snowy Mountains; this area is located entirely
within the Kosciuszko National Park (KNP). A
map of the area is shown in Figure 1.
2.1.
Legislative Restrictions
The Snowy Mountains Cloud Seeding Trial
Act (2004) prescribes a number of restrictions
on the SPERP to satisfy the concerns of various
stakeholders; briefly:
• SPERP operations can only occur when
precipitation is likely to fall as snow over the
primary target area;
• Silver iodide (AgI) must be used as the
ice-nucleating agent, and indium sesquioxide
(In2O3) as the tracing agent; these agents must
be dispersed from ground-based generators,
which cannot be placed within the Jagungal
Wilderness Area (north of the target area);
• SPERP must have a comprehensive
monitoring system to evaluate the effects of
cloud seeding operations.
In addition, SHL had to prepare and implement
an Environmental Management Plan prior to
commencing operations and is required to report
environmental sampling results annually.
2.2.
Background Studies and the Randomization
Scheme
The project design was initially based on an
Expert Panel Assessment (Environ, 2003) that
drew from the climatological feasibility study
(Shaw and King, 1986), the 1993 Environmental
Impact Statement and the results of the Snowy
Mountains Atmospheric Research Program
(SMARP) conducted in 1988–1989. SMARP
focused on the cloud seeding potential over the
Snowy Mountains using observations of wind,
cloud liquid water and temperature structure
(Warburton and Wetzel, 1992). The results of
SMARP also showed that the background
concentrations of silver (Ag) and indium (In)
were ≤ 3 PPT in the Snowy Mountain snowpack.
Based on these results, a snow chemistry
evaluation similar to that of Chai et al. (1993)
and Warburton et al. (1995, 1996) was incor-
porated into the project design.
An additional climatological analysis of
surface precipitation and temperature records at
a site immediately to the north of the target area
was conducted in 2004 to refine the details of
the design. Using the results, an Experimental
Unit (EU) duration of five hours was selected as
practical for evaluation of the impact of seeding.
The climatological analysis, combined with
“opportunity recognition” studies using atmos-
pheric soundings, microwave radiometer mea-
surements and plume dispersion modelling, was
used to predict the number of suitable storm
periods and EUs expected over the five year
experiment, and to assess the number required
for statistically significant results, taking into
consideration the impact of operational con-
straints and cessation criteria. The outcome of
this was similar to the number of suitable seed-

Figure 1. Map of the Snowy Mountain region. Legend at the lower right identifies the various SPERP
instrument sites. Project soundings are launched from the Khancoban weather station site. The
primary SPERP target area is outlined in black and the blue circle represents the Blue Calf remote
field observing site with radiometer, laser imaging probe, icing sensor and standard meteorological
instruments. Icing sensors were also located near Thredbo AWS and Cabramurra AWS. Post-storm
snow profile samples in 2006 were also collected at all sites with wind fences, except Jardines, Mt.
Hudson and Waterfall Farm. Pluviometer sites Cabramurra, Weemalah, Lighthouse, Merricks,
Waterfall, Murray River and Youngal are part of the set of control sites for the statistical evaluation.
ing days estimated by Shaw and King (1986) for
a 5-6 year experiment. Plume dispersion
modelling and atmospheric soundings were also
used to decide the optimum placement of the
ground-based generators.
SPERP is a single target area experiment
that, when various seeding criteria are satisfied,
randomizes the decision to seed. The
randomization scheme is based on a 2:1 seed to
no seed ratio within every group of six
sequential EUs. For “seeded” EUs, silver iodide
(seeder) and indium sesquioxide (tracer)
aerosols are released simultaneously from
collocated generators; in a “no seed” EU only
the tracer aerosol is released. The random-
ization scheme was developed by an independ-
ent statistician from Charles Sturt University and
is a blind draw known only by the personnel
operating the generators and the maintenance
personnel, who are independent of the SPERP
scientific personnel. The seed draw will be made
available to the scientific personnel at the
conclusion of the trial period. There is a purge
time of at least one hour between EUs to ensure
that the target area is clear of the seeding and
tracer agents. The purge time may be increased
as determined by a simple model that uses wind
data collected from atmospheric soundings to
predict the transport of aerosol plumes over the
target area. Time series of trace chemical data
indicate the purge time is adequate.

2.3. EU Operating Criteria
The SPERP is subject to a number of
criteria that must be met in order for an EU and
cloud seeding operations to begin and continue.
The first set of criteria is associated with
environmental concerns:
• The high reservoir storage of the
Scheme’s largest dam, the snow water content
and the snowpack accumulation must be below
set threshold levels. These levels were
described in the SPERP Environmental
Management Plan and are measured weekly.
• There must be no severe weather
threats that precipitation enhancement could
exacerbate (for example, flooding).
• The height of the freezing level in the
atmosphere must be lower than 1600 m (MSL)
to ensure that precipitation falls as snow, rather
than rain, over the higher elevations. If the
freezing level is between 1550 and 1600 m then
remote cameras and external observers are
consulted to ensure that precipitation is not
falling as rain.
A number of criteria rely on the
meteorological conditions. The temperature of
the cloud top must be ≤ -7°C and there must be
at least 400 m of cloud above the height of the
-5°C level (the temperature at which silver iodide
activates as a nucleating agent). These param-
eters and the freezing level are measured, or
estimated, using upper air soundings. In
addition, a radiometer located on Mount Blue
Calf in the primary target area (Figure 1) must
show SLW in the cloud. The threshold value is
0.005 cm, slightly above the radiometer’s
detection limit for liquid water.
A plume dispersion and cloud physics
diagnostic scheme called GUIDE (Rauber et al.,
1988) was adapted to the terrain and wind
characteristics of the Snowy Mountains and is
used to predict the transport and dispersion of
the plumes from the generators to the target
area. Tests using an atmospheric dispersion
model with MM5 wind fields were used to adapt
GUIDE’s highly parameterized wind fields to the
Snowy Mountain region. Data, such as wind
speed and direction and temperature, from
atmospheric soundings are used to initialize the
GUIDE model. For seeding to take place, the
GUIDE output must show one or more generator
plumes passing over the target area, and these
generators must be available for operation.
Finally, the controller in charge of the event must
be confident that the event will last at least
three hours (the time between routine sonde
releases).
2.4.
Evaluation Plan
The SPERP has a comprehensive measure-
ment network comprised of a variety of instru-
ments collecting physical and chemical data
across the study area. The evaluation plan for
the project includes a primary analysis of these
data using statistical techniques to compare
precipitation in the target and control area
(which comprises sites generally upwind of the
target area), and ultra-trace chemistry analysis
of snowfall to ensure that any impact is
consistent with the seeding hypothesis. This
primary analysis aims to identify and quantify the
impacts of seeding in the target area. A
secondary analysis encompasses physical
evaluation of the cloud seeding processes and
further statistical tests to confirm the primary
analysis and investigate the physical processes
involved. The evaluation plan was developed
from the experimental design by an independent
and impartial research group from Monash
University; this group will also conduct the
statistical analysis at the conclusion of the 5-
year research period.
The primary analysis uses regression
methods to estimate the precipitation in the
target area based on observations in the control
area during unseeded EUs. These estimates are
then applied to estimate the 'natural' precipi-
tation in the target area during seeded EUs.
Statistical tests are used to compare the esti-
mated 'natural' precipitation with the observed
precipitation in the target area during EUs. While
area-averaged values of precipitation can be
used to quantify the impact of seeding, more
sensitive tests involving principal components
and canonical correlations can be applied to the
data in order to detect an effect of seeding
against the noisy background of natural
variability. This approach provides a robust and
sensitive method for detecting the presence of a
seeding effect. Further, based on the successful
use of ultra-trace chemistry techniques in
evaluating winter cloud seeding projects in the
western United States (e.g., Chai et al., 1992;
Warburton et al., 1996 and McGurty, 1999), a

snow chemistry analysis will be used in parallel
with the statistical methods to identify the
presence of a seeding effect in the target area.
One specific, and new, method for quantifying
the effect of seeding using trace chemical
analysis of snow profiles collocated with
recording precipitation gauges, and comparisons
between sites with and without chemical seeding
signatures, is in progress, but will not be
addressed in this paper.
The secondary analysis considers additional
statistical tests, such as the classic double ratio
of the average precipitation in the target to the
control area, and a single area analysis,
comparing the total precipitation in the target
area in seeded versus unseeded EUs. In
addition, the probability distribution of precipi-
tation in seeded and unseeded EUs will be
compared using the Kolmogorov-Smirnov test
(Press et al., 1986). A set of sensitivity analyses
will also be performed and these will include a
repetition of the primary statistical analysis with
more predictor variables, which should help
ascertain the sources of the seeding impact, or
identify how it was masked or suppressed.
3. SPERP INFRASTRUCTURE
The SPERP is centrally managed from the
Cloud Seeding Control Center (CSCC) based in
Cooma, NSW. Throughout the winter season
CSCC personnel continually monitor meteor-
ological conditions and forecast for potential
cloud seeding opportunities. Six hours prior to
an expected event, atmospheric soundings
commence and data from instruments located in
the target area are monitored. Data from all the
critical instrument systems are telemetered to
the CSCC in real time and displayed on
monitors in the CSCC control room. Cloud
seeding operations begin once the operating
criteria (see section 2.3) are met. The Cooma
facilities also include a clean room for the
preparation of snow sampling equipment and
the management of snow chemistry samples.
The infrastructure of the SPERP project
includes a radiosonde launch site near
Khancoban, a remote observing facility located
on Mount Blue Calf, 13 ground generator sites,
11 snow profiling sites and 50 surface
meteorological stations. Figure 1 shows the
locations of these sites. Since a large portion of
the project infrastructure is located within the
KNP there were limitations on the selection of
sites for generators and weather monitoring
associated with environmental and visual
impact, and access.
3.1.
Project Soundings and the Remote Observ-
ing Facility
Radiosondes are released from Khancoban,
NSW, about 20 km west and upwind of the
target area. The first sonde is released six hours
prior to the expected start of an EU and then
every three hours during operations. The data
received from the sonde are used to address
various operating criteria, such as the height of
the freezing level, cloud top temperature and
cloud depth. The SPERP uses Vaisala Digital
Radiosondes (Model RS92-SGP), the DigiCORA
III Sounding Software package and the GC25
Ground Check Set (used for pre-testing the
sonde and setting the radio frequency of the
transmitter). Data from the lower levels of the
sounding are telemetered to the CSCC and
input to GUIDE, which is then typically run about
15 - 30 min after a sounding launch.
The remote facility on Mount Blue Calf
(located close to the center of the primary target
area) hosts a number of key instruments
include-ing a microwave radiometer, an icing
rate detector, a 2D laser imaging probe, a heat-
ed wind vane (NRG IceFree3 2450) and ane-
mometer (NRG IceFree3 2448), and tempera-
ture (Vaisala HMT337) and humidity (Vaisala
Humicap 180) sensors. Real-time snow sampl-
ing also takes place at the site during operations
using a unique swivelling snow collector to allow
sample collection in high winds. Additional
equipment may be added to the facility in the
future to further strengthen the physical
measurements.
The dual-channel radiometer (Radiometrics,
Inc, Model WVR-1100) measures the liquid
water and water vapour in the atmosphere
integrated along the zenith. For SPERP
operations, a 30-min running average of liquid
water is calculated in real-time and used as part
of the operating criteria. A heavy-duty heated
blower system is directed towards the
radiometer to keep the window clear of water
and icing. This system is used in lieu of the
Radiometrics blower which was found to be
inadequate in keeping the window clear in the
very harsh environment on Blue Calf. The
SPERP also uses three icing rate detectors
(Goodrich Model 0871LH1) that are located at

Blue Calf and two other locations on the main
ridge of the target area (Eagles Nest near the
Thredbo AWS and Cabramurra in Fig. 1). These
detectors have a probe that resonates at about
40 kHz. When icing on the probe damps the
resonance by an amount equivalent to 0.020 ±
0.005 inches of ice accumulation, a pulse is
registered as a de-icing cycle begins. For three
field campaigns a two-dimensional (2D) laser
imaging probe has also been operated at Blue
Calf. In 2005 the probe was a Droplet Measure-
ment Technologies cloud imaging probe (CIP),
and in 2006 and 2007 a precipitation imaging
probe (PIP) was used. The CIP measures par-
ticles over a range of 25-1600+ microns at 25-
micron resolution and the PIP has a range of
100-6400+ microns at 100-micron resolution.
The sampling rate of the instrument is controlled
by a heated wind speed sensor mounted near
the probe tips.
3.2.
The Ground Seeding Generator Network
The SPERP uses thirteen pairs of seeding
generators, located along the western perimeter
of the target area at altitudes ranging from 439
to 1662 m (five above 1000 m), to disperse the
seeder and tracer aerosols into the atmosphere.
(The seeding aerosol is specifically
AgCl0.22I0.78·0.5NaCl, reported on by Feng and
Finnegan (1989), with an activity of 1.2 x 1014
nuclei gm-1 at -10o C and approximately 1012
nuclei gm-1 at -6o C.) The generators were all
placed on sites that had already been disturbed
to minimize the environmental impact of
installing the structures. Three of the sites have
burners elevated on towers 15–26 m above the
ground to ensure the aerosol plume is released
above the forest canopy. At the remaining sites
the burners are located on trailers and are 5 m
above the ground (Fig. 2). The generators are
remotely operated from the SHL Snowy
Mountains Control Center, thus ensuring that all
CSCC personnel are blind to whether the EU is
a “seed” or “no seed”. Many controls (including
environmental controls) have been implemented
for the operation of the generators.
For interpretation of the ultra-trace chemical
results, the seeder and tracer particle sizes and
releases rates must be known (see for example
Warburton et al., 1995). In SPERP the solution
flow rate of all generators is set at 1250 mL h-1,
and is automatically regulated during operations
via computer-managed control valves. The
particle size distributions of both aerosols have
been measured and found to have average
diameters of about 0.07 μm. The mass release
rates of silver iodide and indium sesquioxide are
20.4 g h-1 and 11.4 g h-1, respectively. The
elemental mass release rates of Ag and In are
both 9.4 g h-1, which leads to an expected mass
ratio of Ag/In in snowfall of one, if both are
removed only by scavenging processes.
Figure 2. The two types of ground-based generator
mountings used by the SPERP. Three generator pairs
are elevated on towers (right) and the remainder are
mounted on trailers (top).

3.3. Meteorological Stations
Fifty meteorological stations are used to
monitor parameters such as precipitation,
temperature, relative humidity, wind speed and
direction, and atmospheric pressure in the
upwind, target and downwind areas. Sixty-one
pluviometers are positioned at 47 of the stations
to monitor snow and rainfall; twenty-four are
0.01 in (0.25 mm) resolution total precipitation
gauges supplied by ETI Instrument Systems
(Model NOAH II). These are used to measure
precipitation at sites above the height of the
typical snowline. Australian Hydrological Ser-
vices tipping bucket gauges (Model TB3) are
used at sites below the snowline. The gauges
are outfitted with low-power heaters (Model
TB323LP) at sites that occasionally experience
snowfall events. The majority of the tipping
buckets have a resolution of 0.2 mm and the
remainder are resolved to 0.5 mm.
Many of the precipitation gauge sites on the
Snowy Mountains are exposed to high wind
speeds with limited or no shelter from
topographic features or vegetation. In these
conditions precipitation gauges are known to
“under-catch” snow because it is blown across
the gauge orifice. The World Meteorological
snow under these conditions is to use Double
Fence Inter-comparison Reference (DFIR)
structures around the gauges. These structures
consist of two concentric-circle (4 and 12 m
diameters) lath-fences around the gauge
(Goodison et al., 1998). The SPERP was not
able to install these full-sized structures at all the
gauge sites because of the limitations
associated with placing infrastructure in the
KNP. As a solution, SHL developed half-sized
structures (½ DFIR), which were collocated with
unfenced gauges at four key sites (Perisher,
Guthega Dam, the Kerries and Grey Hill) prior to
the 2006 cloud seeding season. A full-size DFIR
was also constructed at Guthega Dam to
provide a comparison with a gauge shielded with
a ½ DFIR and an unfenced gauge. Figure 3
shows a photo of the Guthega Dam site and a
graph of the cumulative precipitation recorded
by the three gauges over a two-day period; the
precipitation under-catch by the ½ DFIR and
unfenced gauge is clearly shown. An additional
seven ½ DFIRs were installed at exposed sites
prior to the 2007 season. To account for
differences in gauge shielding among years
unshielded gauge data will be adjusted based
on regression analysis between shielded and
unshielded gauges prior to the formal statistical
evaluation.
Organisation (WMO) standard for collecting
Figure 3. Left: The pluviometer comparison site at Guthega Dam, consisting of an unfenced gauge
(a) and identical gauges fenced with a full DFIR (b) and a half DFIR (c). Right: The cumulative
precipitation measured by the gauges over a two day period in July 2006.
(b) (a) (c)

3.4. Snow Sampling Sites
Vertical snow profiles are collected as soon
as
fter some campaigns there was not
eno
he snow samples are collected in ultra-
clea
possible following each field campaign of
sufficient snowfall. Each profile is divided into 2-
cm segments that are each analyzed for the
presence of the seeder (Ag) and tracer (In)
elements. In the 2004 season samples were
collected from 26 sites, which were usually
accessed by over-snow vehicles. Often the
sampling would extend over three days following
the end of a field campaign because weather
conditions prevented access. The number of
sampling sites was reduced to eleven for the
remaining years of the project. These sites were
selected to represent the primary target and
control areas, to have reliable access, and to be
collocated with ETI pluviometers.
A
ugh snow to warrant sampling because rain
or wind had depleted the snow, or the storm
event has not produced much precipitation. In
particular, during the 2006 season the storm
events were very short and produced only
limited snowfall.
T
n polycarbonate vials, each identified with a
unique barcode. The samples are maintained in
a frozen state from collection to delivery to
Melbourne University (Australia) for ultra-trace
analysis using an Inductively Coupled Plasma
Mass Spectrometer (ICPMS). Results from snow
chemistry analysis for each season are used to
confirm that the primary target and upwind
mountain ranges are being effectively targeted
by seeding material during EUs. The chemical
results are also used in combination with high
resolution precipitation measurements in a new
technique to estimate the quantitative effect of
cloud seeding.
4. BRIEF SUMMARY OF THE FIRST FOUR
SEASONS OF SPERP
The SPERP began operations on 18 June
2004. Since 2004 was the start-up year, the
generators were operated whenever the
conditions were suitable and were not subject to
the randomized five-hour EUs. Table 1
documents the number of campaigns (single
storm events), EUs, hours of operation, and the
seeding operation dates for each year. The
winter of 2006 was a particularly dry season and
did not offer many cloud seeding opportunities.
In addition, two of 12 EUs were suspended in
2006 because changing weather conditions
during the EU period led to a rise in the freezing
level above the restricted height of 1600 m.
Preliminary results from the last four years are
encouraging and annual assessment of snow
chemistry has shown strong seeding signatures
and effective targeting of the study area.
4.1.
Snowfall in the SPERP Target Area
Snow accumulation in the Snowy Mountains
is naturally variable from year to year. Figure 4a
shows the highest snow depth measured at
Spencers Creek (located within the primary
target area) each year since 1954. As shown in
this figure, only one of the first four SPERP
seasons had snow accumulation above the
long-term average. The 2006 season stands out
as having the least snowfall during the entire 54
year period. Despite the below average snowfall
during the three randomized seeding years, only
2006 produced fewer EUs than the number
estimated to be needed for the statistical
evaluation. Figure 4b shows the seasonal snow
depth for 2004 to 2007 where the peak snow
levels generally occurred in August
Table 1. Summary of 2004-2007 SPERP cloud seeding seasons.
Year Campaigns Experimental UnitsaHours of operation Operation dates
2004 12 N/A 187 18 Jun-13 Sept
b
2005 13 31 (3) 164 15 Jun-29 Sept
2006 8 10 (2) 55 7 May – 24 Sept
2007 10 22 (1) 112 21 May – 29 Sept
a The number of suspended EUs is shown in brackets.
ents occurred.
b 2004 was the start-up year and no randomized experim

Figure 4. The snow depth measured at
Spencers Creek in the Snowy Mountains.
(a) The maximum snow depth each year
from 1954 to 2007 and the long term
average (horizontal line). (b) The temporal
trend in snowfall accumulation for SPERP
winters 2004–2007. Data courtesy of
Water Resources (SHL).
4.2. Targeting Effectiveness Using Ag/In Ratios
The use of the Ag/In mass ratio to evaluate
cloud seeding targeting effectiveness and to
demonstrate that Ag arrived in the snowpack by
the ice nucleation process rather than by cloud
and precipitation scavenging processes was
documented by Warburton et al. (1995). The
hypothesis indicated that both aerosols would be
removed by scavenging at the same rate and,
with known particle sizes and release rates, the
expected mass ratio in snowfall due to
scavenging could be computed. SPERP uses a
similar dual-tracer technique, with the main
difference in approach being that the AgI
nucleant is expected to operate by a fast-acting
condensation-freezing mechanism, rather than
by contact nucleation as in Warburton et al.
(1995). If ice nucleation occurs more rapidly
over or upwind of the SPERP target, then the
peak Ag/In ratio could be displaced further
upwind, but evidence of ice nucleation (versus
scavenging) should still be demonstrated by an
Ag/In ratio greater than one as predicted by the
release rates noted in Sec. 3.2.
The snow samples analyzed from the 2004
and 2005 SPERP seasons indicated that
relatively effective targeting was occurring, and
also that some improvement in targeting
occurred after generators were repositioned or
added following the 2004 testing season.
Figure 5 (left) displays the percent of snow
samples at each sampling site in 2004 that were
found to have an Ag/In ratio greater than the
expected 1.0 (from scavenging alone), and
Figure 5 (right) displays the same data for the
2005 season. The percentages in 2005 were
higher at all sites but one, with less variability
over the target. The 2004 data tend to show a
deficiency in targeting across the center and
some eastern portions of the target. Figure 6
compares 2004 and 2005 snow chemistry data
for sampling sites that were the same for both
years. The mean Ag concentrations were 2-10
PPT higher in 2005 and the percentages of
samples with Ag/In ratios greater than 1.0 were
at least 10 percentage points higher in 2005 for
all but one site. This apparent improvement in
targeting is noted even though all storms in 2004
were seeded, while, as a consequence of the
randomization, only about 2/3 of the qualifying
storm periods in 2005 were treated with AgI.

Figure 5. Plots showing the percent occurrence of Ag/In ratios greater than 1.0 (circles) for snow
samples collected at the SPERP sites shown in 2004 (left) and 2005 (right). Circle size is proportional
to percentage as indicated by examples in each plot. Note the two small dots above the 3% circle in
the left plot represent 0%. White squares represent seeding generator sites for each season.
Figure 6. Comparison of snow chemistry results at similar sampling locations for 2004 and 2005
SPERP seasons. Top: Mean Ag concentration from all samples collected at each site with bar
indicating the standard deviation. Bottom: Percentage of Ag/In ratios that exceeded 1.0 at each site.
Note that North Bulls Peak was intended to be a control site.

5. RESULTS FROM AN EXTENDED
OPERATIONAL PERIOD
An extended experimental campaign,
consisting of two storms occurring between 30
July 2006 and the early hours of 1 August 2006,
is examined in detail in this section. Five
experimental units (EUs 38–42) were conducted
during these events; however, EU 39 was
suspended after three hours because the
freezing level rose above 1600 m. The first
storm (EU 38) started with the arrival of a cool
moist westerly airstream over the Snowy
Mountains. This was followed by a short period
of warmer dry air ahead of a cold front. A return
to cool moist conditions with the passage of the
front allowed recommencement of seeding
operations (EUs 39-42). An extensive snow
sampling campaign followed the entire
operational period. In this section we discuss the
atmospheric conditions as measured by
soundings, the GUIDE plumes generated from
the sounding data and the data collected from
Blue Calf and the surrounding area.
5.1. oundings
S
Thirteen radiosondes were launched during
this op ational period and Fig. 7 summarizes
some of the data pertinent to seeding
operations. Figure 7(a) shows the wind speed
and direction at the height of the -5°C level. At
the start of the first storm and continuing through
EU 40 the winds were predominately from the
west, providing potentially good coverage of the
target ea from several generator sites. The
winds b
EU 41
provide
heights
are sh
indicate
(i.e., wi
the low
near or
the targ
er
ar
acked to the southwest halfway through
indicating fewer generators were likely to
effective coverage of the target. The
of the freezing levels and cloud layers
own in Fig. 7(b). The hatched areas
the cloud regions suitable for seeding
th temperature less than -5°C). Note that
er part of these cloud layers was often
below the height of the highest point in
et (Mount Kosciuszko, 2228 m).
Figure 7. A time series showing data
from the thirteen radiosondes that were
launched during the operational period.
The grey shading shows the duration of
each EU which is numbered on the top.
(a) The wind speed and direction at the
-5° C level. (b) The heights of the
freezing levels (white dots) and cloud
layers (black and hatched bars), where
the hatching shows the regions of the
cloud layers with temperature less than
-5 C. The dashed line shows the 1600
m level and the dash-dot line shows the
height of Mt Kosciuszko (the highest
peak in the range). Time is local
Eastern Standard Time.

5.2. GUIDE Plume Plots and Blue Calf Targeting
turned
on if the predicted centerline crosses the target
area. As indicated in Fig. 8, all generators
except Spring Creek and Grassy Flat were
these two generators were not turned on at the
GUIDE is used to predict the trajectory and
spread of the aerosol plume from each
generator and the locations where ice nucleation
and fallout first occur. It is the method that
SPERP uses to select cases in which seeding
plumes should reach at least the -5o C level, and
far enough upwind to permit crystal nucleation,
growth and fallout in the target. GUIDE uses
data, such as temperature, and wind speed and
direction, from the three-hourly atmospheric
soundings for these predictions. Two plan view
plots are produced from the model; one shows
the predicted centerline and the other the
boundaries of the plume from each generator.
Figures 8a and 8b show these plots for the
model run preceding EU 38. An EU can only be
declared if GUIDE predicts the crystal fallout
from at least one generator in the vicinity of the
target area, and a generator can only be
predicted to produce plumes over the target, so
Figure 8: GUIDE model output based on the sounding released before EU 38. Top: Plan view plots of
the expected plume boundaries (a) and centerline (b) from each generator. The predicted point of
nucleation (cross) and fallout (star) for each plume, and the locations of Guthega Dam, Mt. Blue Calf,
Perisher and Thredbo are also shown. Bottom: the vertical trajectory of the aerosol plume and ice
crystals from the Youngal (c) and Pinnacle Mountain (d) generators.

beginning of EU 38. Spring Creek was turned on
part way through the EU following a later balloon
release. Figure 9 shows the centerline plots for
the GUIDE runs preceding EUs 39 to 42. The
winds clearly shifted from northwesterly to
southwesterly as the storm progressed. GUIDE
also predicts the vertical trajectory of the aerosol
plumes, where an ice crystal is expected to form
and the expected ice crystal habit (plate,
column, dendrite, needle or small crystal). Once
an ice crystal is formed GUIDE then estimates
its growth and fallout trajectory. The parameteri-
zations for ice crystal nucleation, growth and fall
speed can be found in Rauber et al. (1988).
Figures 8c and 8d show examples of the vertical
trajectory plots.
Figure 9. GUIDE centerline plots of the aerosol plume from each generator based on the sonde
released before each EU. See Figure 7 for further details.

Figure 10. Plot showing which GUIDE generator plumes existed over Blue Calf for each atmospheric
sounding in the two-day storm period. The grey shading shows the duration of each EU, the black
shading shows the generators that were targeting Blue Calf and the hatching shows the generators
that had crystal fallout predicted upwind of Blue Calf. Sounding dates and times are noted on the
bottom axis (local Eastern Standard Time).
Figure 10 indicates which generators
produced GUIDE plumes over the Blue Calf
observing facility during each EU. Ice crystal
fallout was also predicted in the vicinity of Blue
Calf from a number of generators during EUs
38, 39 and 41. For EU 40 the fallout was
predicted downwind of Blue Calf and for EU 42
nucleation was not predicted for the plumes
passing over Blue Calf. Given these predictions
that the observing facility was potentially being
rgeted during almost every EU, the data from
lue Calf are examined for evidence of seeding
ta
B
effects.
5.3. Description of Data from Blue Calf and the
Surrounding Area
In this section a detailed description of
chemical and physical data recorded at Blue
Calf and the nearby sites of Perisher and
Guthega Dam (see Fig. 1) is given. The wind,
temperature, radiometer, icing sensor and
precipitation data were collected continuously
be at Blue Calf only occurred when
observers were at the site, generally from a few
hours prior to the start of an EU until the end of
throughout the winter. Real time snow sample
collection and operation of the laser imaging
pro

the purge period from the last EU of a sequence.
The pertinent wind data were shown in Fig. 7.
The depth of the liquid water (LW) in the
atmosphere above Blue Calf is shown in Fig.
11b as a 30-minute running average. Also
shown is the threshold depth of liquid water
required as part of the SPERP starting criteria
for an EU. It is obvious that EUs 39, 40 and 41
easily satisfied the radiometer LW criterion, but
that LW during EU 38 and 42 was marginal,
after the EU began. It is also interesting to note
that radiometer LW and the icing rate are not
particularly well correlated. There was steady
icing during EU 38 (Fig. 11c), but the radiometer
LW was barely above background, suggesting
the liquid cloud was surface-based and quite
shallow. In contrast, radiometer LW between
EU 38 and EU 39 and during EU 40 and EU 41
was substantial while the icing rate was quite
slow. This lack of correspondence indicates a
relatively deep LW cloud existed over the site
during these periods, but that the cloud base
was likely above the surface much of the time.
Cloud liquid and precipitation also showed
two interesting relationships. In the period
between EU 38 and EU 39 there was very little
precipitation (Fig. 11d), but a considerable
amount of LW and icing indicating that the
precipitation process was inefficient; the
“classic” definition of good cloud seeding
potential (the period did not qualify for EU
consideration due to insufficient cloud depth and
increasing temperatures). However, during
EU 40 there was a large amount of liquid water
available in the clouds, and at the same time the
highest precipitation rate of the 2-day period was
observed at Perisher (Fig. 11d) a short distance
away. High ice crystal concentrations were also
measured with the PIP (see Sec. 5.5). This
suggests a significant seeding potential can
Figure 11. A time series of physical and chemica
Guthega Dam (see Figs. 1, 8 and 9 for locations).
Blue Calf real time snow samples (the length of th
The 30-
l data
(a) Th
e lines
minute running average liquid water depth (LW
(ITCrecorded at Blue Calf. (d) The cumulative precipita
Dam and Perisher. Figure 3 shows the precipitation measured by the fenced gauges at Guthega over
the same time period. (e) The temperature (T) measured at each site.
collected from Mount Blue Calf, Perisher and
e concentration (C) of Ag and In detected in the
shows the collection period for the sample). (b)
) and (c) the cumulative number of icing trips
tion (p
) C) from the unfenced gauges at Guthega

exist even in the presence of moderate precip-
itation or high crystal concentrations. Finally, th
similar to EUs 38 and 40, indicating EU 41 was
likely a “no seed” decision. In the one sample
collected after the start of EU 42 both Ag and I
e
liquid water dropped to the background level and
the icing (Fig. 11c) decreased at Blue Calf soon
after the start of the last EU, corresponding
temporally to the drop in temperature seen at all
sites (Fig. 11e) and likely to an advection of drier
air into the region after the passage of the upper
level trough.
Snow samples were collected at Blue Calf
whenever an adequate amount of snow for trace
chemical analysis was present in the snow
collector, so the temporal extent of each sample
varied considerably. Figure 11a shows the
collection period for each of the 15 separate
snow samples from the 2-day period. The
concentrations of Ag and In detected in the real
time snow samples are also shown. (Note that
background levels of 3 PPT and 1 PPT have
been subtracted from the Ag and In
n
were sl htly enhanced and the Ag/In ratio was
0.6.
ig
As with EU 39 it is difficult to draw
conclusions from one sample, but the indication
is that the EU was seeded, but that ice
nucleation did not contribute to the presence of
Ag in the snow. Overall, from the three EUs that
were sampled reasonably well at Blue Calf, the
snow chemistry data are consistent in
suggesting what the seeding decision was and
that the GUIDE results were at least qualitatively
verified in their predictions of plume and ice
crystal targeting.
5.4.
Spatial Results from Snow Chemistry
Profiles
On 1 August, following the two-day storm
event, snow profiles were collected at nine sites
within and to the north of the SPERP target. The
condition of the snow at the bases of the profiles
indicated that the profiles only contained snow
from the period of interest. Although the 2-cm
resolution of the profile segments did not permit
as clear a delineation of EUs as the real time
snow data from Blue Calf, there was still an
indication of the EU seeding decision based on
the Ag and In concentrations within the profiles.
Two profiles collected north of the target (Tooma
and Bulls Peak) showed no evidence of
enhanced Ag or In. All seven of the other
profiles showed evidence of enhanced Ag or In
in one or more segments. Recalling the
precipitation data from Fig. 11, the Perisher
trace indicates that 80-90% of the precipitation
came between the start of EU 38 and the end of
EU 40. If snow in the profiles was deposited
proportionally, then the seeding effects from
EU 38 and EU 40, noted at Blue Calf, should
have been present in the lower 75% of the target
profiles. Likewise the “no-seed” indium signature
from EU 41 should have been present in the
upper 25% of the profile segments (assuming
temporal targeting as at Blue Calf). Five of the
seven profiles showed evidence of a seeding
effect in the lower 75% of the profile and three of
seven showed evidence of a no-seed event in
the top 25%. Three other profiles showed
evidence of enhanced Ag and Ag/In ratio in the
top segments, possibly from the last EU which
was not well observed at Blue Calf.
The spatial pattern of enhanced Ag/In ratio
from all profiles is shown in Fig. 12 (as in Fig. 5).
concentrations, respectively.) A snow sample
bag was lost in high wind conditions between
the first and second sample, accounting for the
gap in the data.
Based on the GUIDE predictions of
successful plume and ice crystal targeting of
Blue Calf the following trace chemical results
might be expected. If an EU was seeded, then
enhanced levels of bothAg and In would be
expected. If ice nucleation by AgI contributed to
the Ag concentration, then the ratio of Ag/In
would be expected to exceed 1.0. If an EU was
unseeded, then the Ag concentration would be
near the background level and the In
concentration would be enhanced. The Ag/In
ratio would also be less than 1.0, or would not
be able to be calculated if Ag minus the
background was zero. Examining the chemistry
data in Fig. 11a it is obvious that Ag sample
concentrations were enhanced during EUs 38
and 40, and that the Ag/In ratio was also greater
than one, ranging from about three to five. This
indicates that EUs 38 and 40 were likely seeded
and that ice nucleation contributed to the
enhanced Ag at Blue Calf.
In the suspended EU 39 there was only
enough snow for the collection of one sample in
which no In and approximately background Ag
were detected. The result from this one sample
is not very conclusive, but suggests Blue Calf
was not targeted during the brief seeding period.
During EU 41 no Ag above background was
detected, but In was detected at concentrations

The percentage of samples with enhanced Ag/In
ratio from Blue Calf is also plotted. For all profile
segments and temporal samples the percent-
tages ranged from about 29% to 80%. The
highest percentages were actually from sites in
the Grey Mare range upwind of the primary
target, but within the secondary target and in
close proximity to two seeding sites. The
percentages for this 2006 event were 30-50
points higher than the seasonal values from the
same sites in 2004 and 20-40 points higher than
2005 sites shown in Fig. 6. In all cases the
intended control site at Bulls Peak showed no
chemical evidence of being targeted. Additional
trace chemical statistics from the 2006 case
study profiles are shown in Fig. 12. All sites
except the Bulls Peak control had average val-
ues of Ag, In and Ag/In ratio that exceeded the
natural background, or values to be expected if
the target was not being affected by the SPERP
cloud seeding operations.
5.5.
Ice Particle Characteristics Measured with
the PIP at Blue Calf
Several past research experiments related
to ground-based cloud seeding have docu-
mented ice crystal enhancement within seeding
plumes. Super (1999) summarizes oro-graphic
seeding experiments in the mountains of Utah,
and notes ice crystal enhancement and particle
habit changes induced by ground seeding with
AgI and liquid propane. In other case studies
docu-
Holroyd et al. (1995) and Huggins (2007)
Figure 12. rrence
Circl
not sam
ach s
e sam
tio > 1 computed
ercent of
entrati
Left: Plot showing percentage occu
collected at the sites indicated on 1 August 2006.
the target near the west edge with no circle was
Graphs showing trace chemical statistics from e
Average Ag and In concentrations from all profil
Second panel: Percent of samples with Ag/In ra
from all samples in a profile. Bottom panel: P
concentrations greater than the background conc
of Ag/In ratios > 1.0 in snow profile samples
e size is proportional to percentage. One site in
pled. White squares are seeding sites. Right:
ampling site on the map to the left. Top panel:
ples. Triangles indicate the heights of the sites.
.0. Third panel: Average Ag/In ratio
samples in each profile with Ag and In
ons in unseeded snowfall.

mented microphysical responses to seeding in
the data from ground-based and aircraft particle
imaging probes (CIP model). These experiments
were aided by terrain which allowed (by special
waiver) aircraft flights within 300 m of the
surface and roads through the target region that
permitted penetration of seeding plumes by in-
strumented vehicles. Also, in a specific experi-
ment there was typically only one source of
seeding material and therefore no ambiguity in
estimating times and distances to plume inter-
section points. Several experiments documented
seeding-induced ice particles that were quite
uniform in size and habit, and easily different-
tiated from natural ice crystals.
The SPERP target has no convenient high
altitude roadways to enable use of mobile
ground instrumentation, and to date instru-
mented aircraft have not attempted relatively low
altitude flights over the main range of the Snowy
Mountains. Detailed physical observations to
document the effects of seeding have been
limited to the remote field site at Blue Calf,
where high temporal resolution real-time sample
collection began in 2006. The Utah studies
made use of mobile and stationary NCAR ice
nucleus counters (Langer, 1973) and other
tracer gas releases and detection equipment to
document seeding plume locations in space and
time. This permitted detailed comparisons of
seeded and unseeded cloud regions. The snow
chemistry measurements are the only datasets
that permit any type of seed-no seed
comparisons in SPERP. This section presents
some preliminary analyses of the microphysical
data in two of the apparently seeded EUs.
5.5.1. Microphysics Data from EU 38
The PIP at Blue Calf collected data
continuously from 0000 EST through 1600 EST
on 30 July and therefore sampled the period
before, during and after EU 38. The data were
processed to obtain 30-s averages of ice particle
concentration, 5-min averages of particle size,
size distributions at 10-min intervals, and a
v
(I
re
m
d
a
s
(2
c
temporal history of
everal PIP variables together with the snow
-
trati ue Calf for the period
surrounding EU 38. Also indicated on the time
axis
of aerosols were likely below the -8 C level. The
range of transit times and nucleation temper-
atures could have produced a variety of particle
sizes and habits at Blue Calf, but “warmer”
habits like needles and columns would be the
most likely in the -5o to -8o C temperature range.
In the low-level SLW layer above the mountain
these crystals could also potentially rime quickly
into small graupel.
The plume arrival times in Fig. 13 bracket a
sharp increase in particle concentration to >100
L-1 (0455 EST) following a one hour period
where concentrations were generally < 15 L-1.
Snow from this initial peak and a second peak at
about 0540 EST would have been collected in
the first snow sample in EU 38 that showed
enhanced Ag and In concentrations. A broad
concentration maximum which exceeded 100 L-1
(0720-0815) was contained in the second and
third snow samples where Ag concentration was
about 24 and 32 PPT, respectively, and the
Ag/In ratio was >4 in both samples. The third
snow sample covered about three hours of
snowfall, but the enhanced Ag was likely
contributed by the ice particles collected at
0755-0820 EST. Particle concentrations in the
two to three hours following the last plume had
peaks about 40 L-1 less than the peaks between
plume arrival and departure, where trace
,
These preliminary results focus on only a limited
set of the measured and derived variables.
Figure 13 shows the
s
sampling periods and trace chemical concen
ons measured at Bl
are the earliest and latest times (36–75 min
transit times) of seeding plume arrival at, and
departure from Blue Calf based on GUIDE
estimates from four ground generators (see Fig.
8). From GUIDE vertical plume trajectory
estimates and sounding temperatures the seed-
ing plumes potentially interacted with cloud in
the -5° to -11°C temperature range (suitable for
nucleation by AgI), but the higher concentrations
o
ariety of variables including ice water content
WC), ice mass concentration and radar
flectivity that were derived from empirical size-
ass relationships. A complete set of the two-
imensional images from the PIP was also
rchived. In addition, a particle habit recognition
cheme similar to that of Korolev and Sussman
000) was employed to obtain the fractional
ontribution of habits to concentration and mass.
chemistry verified that ice crystals contained the
seeding and tracer elements. The average
concentration in the seeded period was 31.1 L-1
compared to 28.9 L-1 in the two-hour period prior
to seeding, and 25.7 L-1 in the 2.5 hour period
following the plume period. The standard
deviations were similar to the means in all
periods, so the differences noted were not very
meaningful.

Ice crystals in the period between plume
arrival and departure were dominated by single
crystals with habits fluctuating among needles
and columns, and small rimed columns or
graupel. At times, uniformly-sized particles of
similar habit, that could have initiated from a
seeding point source, were noted, as in Fig. 14
from 0527-0537, but this was not observed in all
the concentration maxima. Overall there was
some evidence of microphysical changes that
d seeding
er Blue
Cal
could have been induced by clou
during passage of seeding plumes ov
f, but none that could be definitively termed
seeding signatures. The variability noted could
have resulted from a mixture of natural and
seeding-induced ice and also from ice in
seeding plumes originating at various times and
temperatures.
5.5.2. Microphysics Data from EU 40
The time series of PIP data and snow
chemistry for EU 40 is shown in Fig. 15. For this
case the range of plume arrival times (0535-
0556) at Blue Calf was based on GUIDE
predictions for three generators, with transit
times to Blue Calf between 39 and 60 min. Wind
direction shifted somewhat during the EU so the
plume departure time (1046) was based on the
GUIDE prediction from just a single generator.
The estimated plume arrival corresponded quite
well to a marked increase in Ag and In
concentrations (from ~6 to >15 PPT for Ag). The
enhanced trace chemistry concentrations contin-
ued throughout the period of effect at Blue Calf
with Ag/In ratios ranging from 2.8 to 4.2. In the
sample taken after 1300 (not shown) Ag and In
concentrations had dropped to values of about
0.3 and 2 PPT, respectively. Although the plume
arrival corresponded to an increase in PIP
particle concentration from ~20 L-1 to >100 L-1,
the average particle concentration over the
plume period at Blue Calf was only 8-10%
greaterthan the averages over the 2-3 hours
before and after the plume period, with the
averages being well within the standard devia-
tions of all three periods. Strong evidence of ice
particle enhancement in seeding plumes has
been previously documented (e.g., Holroyd et
al., 1994 and Huggins, 2007) but in these case
studies the background concentrations in natural
precipitation were quite low, often less than 5
L-1, so the seeding signal stood out well above
the background. In the current case with a
relatively high natural ice particle background (~
20-30 L-1 before and after EU 40) a statistical
method will be required to test for differences
Figure 13. Time series plot of Blue Calf precipit
chemistry data for the period encompassing EU 3
difference between the total bar concentration a
time axis represent the initial aerosol plume arrival
one or more generators. Dashed black lines are s
reach Blue Calf once generators are turned off.
ation
8 on
nd th
tim
imilar
imaging probe (PIP) data and real time snow
30 July 2006. Actual indium concentration is the
e silver concentration. Solid black lines on the
es at the start of an EU predicted by GUIDE for
departure times for the back edge of plumes to

between seeded and unseeded periods. Such a
comparison is part of the SPERP evaluation
plan.
Changes in ice crystal habits during EU 40
can be seen in the bottom panel of Fig. 15. Prior
to EU 40 the fraction of needle habits shows at
least three prominent peaks (0305, 0415 and
0455) of about 0.4. During these needle peaks
the fraction of aggregates or irregular particles
generally decreased from 0.8 to 0.4 or less.
Within the plume period at Blue Calf there were
several interesting habit transitions. The first
habit fluctuation came between 0700 and 0800
when needles first peaked to about 0.4, then
declined to a relative minimum of <0.1, while
graupel and then aggregate fractions increased
markedly to 0.6-0.8. Another needle maximum
Figure 14. Ice particle images recorded by t P at Blue Calf between 0527 and 0540 EST
during EU 38, and corresponding to the sec ncentration maximum in Fig. 13 following the
initial seeding plume arrival. The images app to be rimed columns, plates, or small graupel.
The height of the vertical bars is 6.4 mm.
he PI
ond co
eared

(fraction >0.2) occurred from 0930-1000 with a
corresponding drop in aggregates. The final
hour within the plume period, and most of the
sub quent hour, had very steady fractions of
the
se
four habit types with the bulk, about 0.6 of
the total, being aggregates. The period from just
after 0800 to 1100 had characteristics that were
most consistent with a seeding effect; an abrupt
particle concentration increase that stayed
between 60-80 L-1, the highest Ag concentration
of the EU (~23 PPT), and PIP images that
indicated a predominance of small graupel and
rimed particles ≤ 2 mm in size. Images from this
period and from a later period outside the plume,
which was dominated by large crystals and
aggregates, are shown in Fig. 16.
Figure 15. Top: As in Fig. 13 for a time period encompassing EU 40. Bottom: The fractional
artitioning of particle habits classified by the PIP processing program. The partition above the
irregular trace represents the unclassified particle fraction. For reference, between 08 and 09, the
aces from bottom to top are needle, graupel, aggregate and irregular.
p
tr

Figure 16. Images as in Fig. 14, except recorded during and after EU 40. Images in top three strips
were recorded at about 0810, the next three strips at 0900, and the next three strips at 1050, during
the late plume period of EU 40. Images in bottom three strips were recorded at 1150, about one hour
after the plume period ended.

6. SUMMARY
The SPERP design is based on several cli-
The randomized seeding project also
incorporates additional physical and trace
chemical measurements to verify certain
aspects of the conceptual model. The simul-
taneous release of an ice nucleating aerosol and
habits and sizes that were consistent with the
time required for nucleation, growth and fallout
matological, physical and environmental studies
in the Snowy Mountains of Australia. The
conceptual model and plan for the randomized
portion of the experiment are similar to
orographic ground-based cloud seeding ex-
periments of the past, but there are also char-
acteristics that are unique to the SPERP. The
experimental unit has a relatively short duration
of five hours, which was chosen in order to
collect enough samples to detect a statistically
significant difference in both treated and
untreated EUs, and between the target and
control areas, over the planned 5-year duration
of the experiment. Legislative restrictions have
limited the seeding methodology to a specific
seeding agent (AgI compound) and release
method (ground-based), and introduced some
environmental seeding criteria; the overall effect
of which could be a reduction in the seasonal
number of EUs.
The SPERP infrastructure and data
collection plan provide for the measurement of
precipitation in the target and surrounding
region, and at several control sites that are
required for the statistical analysis that will be
conducted at the end of the project. A much
improved precipitation network, comprised of
gauges in the primary target that are shielded by
half-size DFIRs, has been installed for this
purpose. Other SPERP measurements provide
continuous monitoring, with real time data
access, of meteorological variables (winds, tem-
perature, and SLW) needed to assess project
seeding criteria. A simple plume dispersion and
ice crystal trajectory prediction scheme (GUIDE)
has been adapted to the Snowy Mountains to
aid in the objective selection of EUs.
Sixty nine EUs have been completed to date
within the first three formal randomized
experimental seasons of SPERP, which puts the
project essentially on track to reach the
estimated required target of 110 EUs in five
seasons. This has occurred even with the large
variability in storm frequency and snowfall during
the three seasons, particularly during the 2006
El-Nino season, which experienced very low
snowfall and resulted in only 12 EUs, with two of
these suspended.
an inert tracer from all ground-based generator
sites, and the subsequent ultra trace chemical
analysis of snow samples collected at sites
spread across the target permit an assessment
of targeting frequency. Enhanced ratios of
seeding element (Ag) mass to tracer (In) mass
also indicates whether the seeding agent was
deposited in the snow as a result of ice
nucleation, and not just removed by scavenging.
Results from the first two seasons revealed
enhanced Ag concentrations over most of the
target and enhanced Ag/In ratios were found in
up to 50% of individual profile samples. Profiles
from an extended storm period in the third
season (2006) showed even better coverage of
the target through enhanced Ag concentrations,
and enhanced Ag/In ratios in up to 80% of
samples in some profiles, with a range of 29-
80%. Profiles are typically collected after every
storm event, so this technique permits a
targeting assessment on a storm-by-storm
basis.
Analysis of atmospheric soundings, SLW
measurements, real time snow samples and
laser imaging probe data from a 2006 extended
storm period produced some insights into
probable seeding processes within a storm. A
time series of chemical results from the center of
the target showed a pattern that was consistent
with the apparent seeding decision for several
EUs conducted during the storm, even though
the seeding decision has not yet been revealed.
The results also showed that seeding or tracer
material was reaching the center of the target
throughout each EU, that the GUIDE prediction
of plume and ice crystal trajectories for the
target were verified (not specific generator
origin), and that ice nucleation contributed to the
Ag in the snowfall at the observing site. As noted
above, the snow profiles for this storm period
showed similar results for the bulk of the target,
and also verified that sites north of the target
area were not affected by seeding operations.
The trace chemical data delineated periods
when the aerosol and ice crystal plumes were
present at the Blue Calf observing site. One
period with a markedly higher Ag concentration
near the end of EU 40 also showed a coincident
particle concentration increase and crystal

from a seeding site. In general the PIP
microph sical data did not show consistent and
obv
n background was 20-30 L ,
with occasional peaks > 80 L-1. Both situations
tend
S
emistry data have verified
that seeding and tracer material reach the target
are
ment.
y
ious ice particle enhancement when com-
pared to periods before and after the seeding
period of effect. There were times when ice
particle increases were found to be coincident
with plume arrival times, and average particle
concentrations were slightly higher than average
concentrations from surrounding periods, but the
plume average was well within the background
variability. Likewise ice particle habits that were
consistent with the cloud temperature range
where nucleation and growth likely occurred in
the seeding plume were observed, but similar
habit occurrences were also noted outside
seeding plume periods.
In one case the PIP data showed periodic
occurrences of relative high ice particle
concentrations (peaks to 100 L-1), with a low
background (≤ 10 L-1) between peaks. In the
second case during the 2-day storm period, the
ice concentratio -1
ed to mask any obvious ice particle
enhancement signatures. Compared to results
from experiments where obvious seeding effects
were noted, such as Huggins (2007) in Utah, the
SPERP cloud temperatures were markedly
higher, -5o to about -10o C, versus temperatures
in the Huggins case study where the seeding
plume interacted with cloud at -12o to -17oC. The
nucleation activity of AgI is about two orders of
magnitude greater in the latter case. The ice
concentration background in the Utah case was
often 5 L-1 or less (so seeding enhancement of
30+ L-1 was obvious), while the SPERP case
had natural averages of 10-30 L-1 and peaks of
80-100 L-1.
7. DISCUSSION AND CONCLUSION
A number of conclusions can be drawn
regarding the design and operation of SPERP
and the preliminary results that have been
presented. Based on the seeding hypothesis
that ground-based seeding with an AgI-type
seeding agent can increase precipitation in the
mountainous target area, the design of the
generator network appears sufficient to transport
seeding plumes across the target in the typical
wind directions that accompany snow-producing
storms. The snow ch
a a relatively high percentage of the time
during experimental seeding periods. However,
for the case study analyzed, the current
instrumentation did not fully assess real time
seeding plume aerosol concentrations over the
target, and siting an ice nucleus counter at the
Blue Calf site would be a good addition to the
physical measurements. This would also allow
other measurements such as SLW, precipitation
and ice particle measurements to be analyzed
and stratified with respect to seeding plume
presence, and increase the likelihood of
identifying seeding effects in the target.
Additional physical measurements, including
aircraft measurements, and cloud modelling
studies are being considered by the SPERP and
these could contribute significantly to the
understanding of the natural cloud processes
and to changes induced by seeding. In particular
microphysical measurements made in the
upwind portions of the orographic cloud over the
main range of the Snowy Mountains combined
with Blue Calf radiometer and microphysics data
could provide a much clearer picture of both
natural and seeding-induced precipitation
develop
The real time meteorological measurements
collected and monitored by the SPERP have
been shown to be well suited for the declaration
of EUs. The GUIDE scheme, although simple by
current model standards, was found to correctly
predict plume presence and arrival time over the
Blue Calf site within the time resolution of the
real time snow samples which verified when Ag
and In were present in snowfall at the site. The
addition of ice nuclei measurements would
provide another means of verifying plume
transport to Blue Calf. Also, another way of
verifying the source generator would be the
addition of a third tracer element to one or more
generators in the network. The SLW
measurements are critical to the success of
SPERP and, although there is considerable
variability, the case study suggests excess SLW
is relatively common, and at times present
coincident with moderate precipitation and
relatively high ice particle concentrations.
Although the main range of the Snowy
Mountains provides good orographic lift, with the
terrain ascending approximately 1800 m over a
horizontal distance of about 15 km, the overall
height of the ranges may occasionally be a
detriment to ground-based seeding in that the
typical layer where the seeding plumes are
expected to be found (~300-600m above the

terrain) represents temperatures where AgI is
less effective as an ice nucleant. This situation
might be improved by aircraft delivery of AgI, or
with the use of materials such as liquid propane
which can create ice at temperatures lower than
about -2o C. However, because of aircraft terrain
avoidance restrictions, it may not be feasible to
fly where seeding would be most beneficial, and
additional equipment for ground-release of
propane would likely require new and time
consuming environmental studies and
approvals.
The SPERP precipitation gauge network,
particularly in the primary target, has been
significantly improved and the spatial coverage
and data quality from this network should permit
an assessment of seeding impact using the
proposed statistical evaluation. At least two
factors could impact the success of detecting a
positive seeding impact; a lower than expected
number of EUs in the five-year experimental
period and a seeding effect that is less than that
used to project the number of EUs required to
attain a significant result. On the positive side
several of the current data sets, including SLW,
temperature, microphysics observations, and the
trace chemical data can be used as either
secondary response variables or as a means of
stratifying EUs to statistically evaluate those
cases that have the best potential to show an
enhancement in precipitation.
Acknowledgments
The SPERP is strongly supported by an
AusIndustry Renewable Energy Development
Initiative (REDI) grant. The authors wish to thank
John Denholm, Phil Boreham, Mike Mathews,
Amanda Johnson, Barry Dunn and Mark Heggli
nnovative Hydrology) for the initial
the
SPERP. We are grateful to all the SHL
per
Env
st functioning nucleating agent –
AgI*AgCl-4NaCl. J. Weather Mod., 21, 41-
ice
nuc us counter. Part I: Basic operation. J.
(I
development and ongoing support of
sonnel who are involved in the project, in
particular with instrument maintenance, balloon
launching, snow sampling activities, and
operation of the Blue Calf facility.
We also thank Stephen Siems, Jingru Dai
and Thomas Chubb from Monash University,
School of Mathematical Sciences for their
contributions to the project.
In addition, thanks are given for all the
further support provided to the SPERP, in
particular, Stephen Chai (Desert Research
Institute), Brian Williams, Alan Greig (Melbourne
University), Kevin Wilkins (Charles Sturt
University) and Doug Shaw (CSIRO
Mathematical and Information Sciences).
8. REFERENCES
Chai, S. K., W. G. Finnegan, and R. L. Pitter,
1993: An interpretation of the mechanisms
of ice-crystal formation operative in the Lake
Almanor cloud-seeding program. J. Appl.
Meteor., 32, 1726-1732.
iron, 2003: Snowy Precipitation
Enhancement Trial. Expert Panel
Assessment, Snowy Hydro Ltd., Cooma,
NSW, Australia.
Goodison, B.E., P.Y.T. Louie and D. Yang,
1998: WMO Solid Precipitation
Measurement Intercomparison Final Report,
Instruments and Observing Methods, Report
No. 67, WMO/TD No. 872.
Feng, D. and W. G. Finnegan, 1989: An
efficient, fa
45.
Holroyd, E. W., J. A. Heimbach and A. B. Super,
1995: Observations and model simulation of
AgI seeding within a winter storm over
Utah’s Wasatch Plateau. J. Weather Mod.,
27, 35-56.
Huggins, A. W., 2007: Another wintertime cloud
seeding case study with strong evidence of
seeding effects. J. Weather Mod., 39, 9-36.
Hughes, L., 2003: Climate change and Australia:
trends, projections and impacts, Austral
Ecology, 28, 423–443.
Korolev, A. and B. Sussman, 2000: A technique
for habit classification of cloud particles. J.
Atmos. Oceanic Technol., 17, 1048-1057.
Langer, G., 1973: Evaluation of NCAR
le
Appl. Meteor., 12, 1000-1011.
McGurty, B. M., 1999: Turning silver into gold:
Measuring the benefits of cloud seeding.
Hydro Review, 18, 2-6.

Press, W.H., B.P. Flannery, S.A. Teukolsky, and
W.T. Vetterling, 1986: Numerical Recipes,
Cambridge University Press, 818 pp.
Rauber R. M., R. D. Elliot, J. O. Rhea, A. W.
Huggins and D. W. Reynolds, 1988: A
diagnostic technique for targeting during
airborne seeding experiments in wintertime
storms over the Sierra Nevada. J. Appl.
Meteor., 27, 811-828.
Shaw, D. E. and W. D. King, 1986: Feasibility
study to assess the potential of a cloud-
nd M. A. Wetzel, 1992:
Field study of the potential for winter
Wa g and R. H. Stone,
1995: Assessment of seeding effects in
arburton, J. A., S. K. Chai, and L. G. Young,
seeding experiment over the catchment of
the Snowy Mountains Scheme. SIROMATH
Pty Ltd. report.
Smith, E. J., E. E. Adderley and D. T. Walsh,
1963: A Cloud Seeding Experiment in the
Snowy Mountains, Australia, J. Appl. Met.,
2, 342–332.
Super, A. B., 1999: Summary of the NOAA/Utah
Atmospheric Modification Program: 1990-
1998. J. Weather Mod., 31, 51-75.
Warburton, W.J. A. a
precipitation enhancement in the Australian
Snowy Mountains. Atmos. Res., 28, 327-
363.
rburton, J. A., L. G. Youn
snowpack augmentation programs: Ice
nucleation and scavenging of seeding
aerosols. J. Appl. Meteor., 34, 121-130.
W
1996: A new method for assessing snowfall
enhancement by silver iodide seeding using
physical and chemical techniques. J. Appl.
Meteor., 35, 1569-1573.