Results of a cloud seeding project in the Snowy Mountains of Australia
Upper air soundings, routinely taken during the trial from a site on the western side of the mountain range, were used to drive a plume model to estimate the dispersion of seeding material. Cloud seeding events (experimental units, or EUs) were fixed at five hours duration, and EUs could only be commenced when specified criteria had been satisfied. These included a requirement that the SLW measured by a mircowave radiometer on the mountain ridge was above a threshold, and the dispersion model predicted that generator plumes would pass over the target area. Other criteria specified that the cloud top temperature must be colder than -7°C, and there must be at least 400 m of cloud above the -5°C level.
The conduct of the trial was enabled by special legislation, which amongst other things mandated certain operating and suspension criteria. The effect of this was an obligation to immediately terminate an experiment if the freezing level was found to exceed 1600 metres ASL, or if rainfall was reported at 1400 metres ASL.
The trial employed a fixed target-control design, with a randomised seeding ratio of 2:1. A total of 107 EUs were undertaken over the five year experimental period, with 71 of these EUs being seeded and the remaining 36 unseeded. Considerable care was taken to ensure that the seeding sequence was not disclosed to any personnel involved with decision-making or analysis during the course of the entire trial. Moreover, independent referees were appointed to ensure transparency in the implementation and analysis. To avoid the issue of multiplicity, an Evaluation Plan setting out in detail the primary tests for seeding impacts was defined and published in the scientific literature before evaluation of the trial commenced.
A passive tracer (indium (III) oxide) was released during every EU, while silver iodide was released only during seeded EUs. The generating infrastructure was designed to disperse silver and indium particles with similar particle size distributions. Snow samples were collected during and following each experimental campaign, and ultra-trace analysis was undertaken to determine the concentrations of silver and indium in each of these samples. It was proposed that a comparison of the ratio of silver to indium in these snow samples could be used to provide an indication of seeding effectiveness. This was based on the assumption that while silver iodide acts as an ice nucleating agent, indium (III) oxide does not. Therefore, in snow samples where seeding was effective it was expected that the ratio of silver to indium would be greater than 1:1.
On further investigation, it was found that substantial background levels of silver over the study area complicated this assessment, and so the primary test for targeting was taken conservatively to be that the maximum concentration of silver in the target area should be significantly greater in seeded than in unseeded EUs. This test is readily satisfied, and further analysis shows that the ratio of silver to indium was greater in seeded than unseeded EUs.
The evaluation plan prescribed the primary test for the impact of seeding on precipitation in the primary target area (above 1560 m), and it required precipitation to be increased at a significance level of at least 10%. While an increase of 7% is found, the significance level is low.
Further analysis shows that the main source of uncertainty is the inclusion of EUs where the seeding generators operated for relatively few hours (for example, by inclusion of EUs that were terminated because an environmental suspension criterion came into effect). When EUs with low generator-hours are removed from the analysis, precipitation in the primary target is found to be increased by 14% at the 8% significance level. Importantly, when this analysis is repeated for the entire target area, the outcome is found to be a 14% increase in precipitation at the 3% significance level.
Further analyses support the seeding hypothesis by finding statistically significant relationships between the seeding impact and a range of physical variables (including SLW).