Laserfiche WebLink
<br />i <br /> <br />11 <br /> 19-Mar <br />12 <br /> 20-Mar <br />13 <br /> 22-Mar <br /> 23-Mar <br /> <br />area. <br />West-northwest to east-southeast (light-moderate) maximum axis along <br />San Juan divide, with absolute maximum just east of eastern target area. <br />Uncom ah re d . <br />Two maxima axes - one along San Juan divide from central part of <br />western target area to eastern extremity of eastern target, and another <br />along Uncompahgre range including control area there. Light-to-moderate <br />dSWE. Western tar et is in relative minimum. <br />Light-to-moderate maximum axis along San Juan divide and Uncompahgre <br />ran e. <br />Large variation in dSWE across region, with maximum axis along San Juan <br />divide and minima on southern eri he of the tar et area. <br /> <br />b. Recommendations <br />Given these results, what other methods are available for evaluation of operational seeding <br />programs? The most common method is historical target-control regression. The area with the <br />longest record of seeding in Colorado, and therefore the most promising for such an analysis, is <br />at the Vail ski area. Even with nearly 30 years of data at Vail, however, there are limitations as <br />described in Section II a. An alternative to the historical regression is ratio statistics4. These <br />were recently applied to several long-term operational seeding programs in the Sierra Nevada of <br />California2 (Appendix A of Silverman). The result was a statistically robust indication that the <br />seeding there has produced significant increases in streamflow since the 1950s. A regression <br />using a similar data set failed to produce a statistically valid o~tcome. The statistical analysis <br />technique most widely recognized as robust is the randomized seeding experiment. This <br />experiment is more costly and requires that a portion of storms go unseeded. Therefore <br />randomization has largely been attempted only with research projects. <br /> <br />It is unclear why the current study using SNODAS did not show a seeding signal (Section IV a). <br />It is also unknown whether the system could diagnose an actual seeding signal in future <br />applications; if not, SNODAS may at least prove useful in providing snowpack information <br />relative to the meteorological conditions that influence seeding. The discussion of <br />meteorological conditions and resulting dSWE shown by Fig. 5 (Section III) is an example of <br />how snowfall patterns vary with the meteorological situation. If many such examples are <br />analyzed, patterns might emerge that will assist in adjusting seeding strategies (e.g., which <br />generators to operate and when, optimizing generator placement, etc.). <br /> <br />The foregoing two paragraphs describe statistical and modeling forms of evaluation. Another <br />major form is physicaP. Physical approaches involve either remote or in situ measurements of <br />seeded plumes, their effects on precipitation, or other atmospheric parameters related to cloud <br />seeding. To measure seeded plumes and their effects on precipitation, aircraft sampling and <br />trace chemical analyses of snow have been used. Examples include single and dual tracer <br />techniques, and combined physical and chemical methods. An example of single tracer use <br />would be measuring silver in snow from seeding). Durations and concentrations of silver iodide <br />in the target area can be measured by ice nucleus counters. All the foregoing techniques assess <br />whether seeding materials are reaching the target areas and/or how those materials are involved <br />in precipitation processes. To assess actual precipitation augmentation from seeding, high <br />resolution snow gauges are required, both in the target area and suitable control areas. A recent <br /> <br />11 <br />