Laserfiche WebLink
<br />Even with a single seedline, natural variability can mask seeding effects. It is very important to <br />monitor both the temporal and spatial changes over and near a the target site. One useful approach <br />is to operate surface measurement stations in addition to the target. These stations should be <br />located crosswind of the area expected to be affected by seeding so as to provide a record of natural <br />variations. Radar observations of the region can also be very helpful in this regard, as noted earlier. <br />It is essential that enough observations be collected in both space and time to determine which <br />perturbations are real seeding effects, which are natural variations, and which may be seeding <br />effects masked by natural variations. <br /> <br />Some past weather modification experiments seriously underestimated the resources required to <br />analyze their field observations, or delayed detailed analysis until several field seasons were <br />completed. In either case the very valuable feedback from analysis to improving field design was <br />lost. In the worst case, programs were canceled after several years of expensive data collection, but <br />before adequate analysis and reporting. Such programs were very wasteful of time and resources. <br />It is strongly recommended that the Arizona Program provide substantial funding for analysis, and <br />that analyses of each field season's data be reasonably complete before finalization of the next <br />season's design, Conducting field expeditions every second or third year is one approach to <br />accomplishing this. Another is to isolate the analyses group from any field involvement beyond that <br />required for their familiarity of general field techniques and procedures. In general, it is preferred <br />to have the same scientists collect and analyze the data. This ensures careful collection and, even <br />more important, expands the scientists' comprehension of the overall project, which can significantly <br />improve both design and analysis. However, use of the same group in both roles does require <br />considerably more time for completion of the experimental program. <br /> <br />Any physical experiment requires some targeting scheme to decide when and where to release the <br />seeding material in the case of a fixed target, or when and where to operate the "mobile target" <br />(usually sampling aircraft) in the case of fixed generator locations. The scheme may be no more <br />complicated than using a typical wind velocity for the altitude range in question to estimate <br />transport time, and typical growth rates and fall speeds for the type(s) of ice particles expected. <br />Such approaches are sometimes referred to as "back of the envelope" calculations. On the other <br />extreme, a highly sophisticated three-dimensional time-dependent numerical model may be run on <br />a supercomputer to simulate the entire airflow pattern around the barrier and all important <br />microphysical processes for expected ranges of conditions. Given the uncertainties in certain key <br />processes (e.g., aggregational growth rates) and impracticality of making detailed measurements <br />around mountains, in particular concerning the winds and the spatial distribution of SL W, it is <br />probably more reasonable to use a targeting model of modest sophistication that can be run on a <br />small computer in the field using real-time input data. The approach used by Rauber et al. (1988) <br />for the SCPP is a good compromise. If resources permit, a more sophisticated model should be run <br />for various combinations of atmospheric conditions believed typical of winter storms in the region <br />of interest. The resulting predictions should be in general agreement with the operationally used <br />scheme or the latter might require some modification. For the purposes of the Arizona Program, <br />it is advised that an existing numerical model of moderate sophistication be adapted for the <br />Mogollon Rim, compared with real observations, and modified if necessary for real-time decision <br />making during physical experiments. Because of the uncertainties mentioned, any model should <br />only be used for general guidance as its predictions are no substitute for actual measurements of <br />reality. <br /> <br />An important lesson concerning physical cloud seeding experiments is that they should be as <br />complex as required to document and understand the: important phenomena, but no more so. <br /> <br />9 <br />