WMOD00528 - Laserfiche WebLink

Home Browse Search

missed. A much better approach is to ~ontinually monitor the weather and declare an experimental unit when preselected objective criteria are met for some minimum time, say 30 min. Such criteria could include radiometer-observed SL W exceeding a certain value, crestline winds greater than a speed threshold and within certain direction limits, cloud base lowering to near the mountain crest and base temperature cold enough for ice nucleation by the seeding agent. Thus, the design would attempt to exclude periods when seeding is expected to be ineffective. Some undesirable periods would still be selected for experimental units; for example, when SLW appears for awhile and then disappears. However, this objective approach based on observations would have significantly fewer "false alanns" than a scheme based on forecasts. 7.3 Physical Observations Identical observations would be made during all seeded and nonseeded (control) experimental units for later statistical evaluation. The primary response variable likely would be total precipitation per experimental unit, averaged over all gauges in the prespecified target area. Secondary response variables might include precipitation rate and duration. Some supporting observations would be used to partition the experimental units into meteorologically similar populations. The limits and objective procedures for partitioning would be developed prior to the randomized experiment. The most important observations likely would be seeding plume temperature and those that indicate the amount of excess SL W passing over the barrier unconverted to snowfall. This would include radiometer observed SL W, or SL W flux, in a region not affected by seeding. Measurements could be made in the buffer zone between target areas if a randomized crossover design is used. Past statistical experiments were unable to routinely monitor excess SLW availability. As a consequence, many experimental units likely were included that had no seeding potential because of lack of SLW. Inclusion of numerous units without seeding potential can seriously weaken the power of a statistical experiment. Additional supporting observations likely to be valuable for partitioning the experimental units are indicators of natural stonn efficiency of conversion of SL W to precipitation. These include measurements of the IPC, upwind or crosswind of the seeded zone, which is thought to be related to cloud top temperature. Precipitation rate observations crosswind of the target indicate the natural conversion of SL W to snowfall but do not show how much excess SL W is left for seeding to affect. A stonn period producing moderate or even heavy snowfall rates may still have excess SL W available if strong winds result in substantial condensate production (Super and Holroyd, 1989). However, it is likely that most heavy precipitation rate periods result from efficient stonn processes. Conversely, light snowfall periods .with SLW available would tend to be inefficient. Consequently, snowfall rates observed crosswind of the target may be useful in partitioning. Observations of key physical processes would continue to be made frequently throughout the course of the statistical program to test for flaws in both design and conduct of the experiment, and to further physical understanding. Specifically, comprehensive physical experiments and more limited physical experiments (e.g., of transport and dispersion) as described in section 6 would be continued throughout the statistical/physical experiment phase. However, it would not be practical to conduct such measurements throughout the course of each statistical experimental unit. But suitable physical measurements for a subset of the experimental units should indicate whether the seeding hypothesis was usually being satisfied. 54