Laserfiche WebLink
<br />might be expected that a real seeding signal would be <br />more evident during light to moderate natural <br />snowfalls than during heavy snowfalls, these results <br /> <br />are remarkable in view of considerably weaker <br />associations (correlations) with the control gauge. <br /> <br />Table 3. HAS direction partition <= 270 deg combined with GNO SWE<= 0.05 inch: <br />24 pairs, 30 seeded and 25 non seeded <br /> <br /> Pooled Re- <br /> Target - randomized I-tailed 95% <br /> GNO Sample I-tailed Confidence <br />Gauge Correlation MDR P-value Interval <br />GSC 0.52 1.30 0.10 1.16 - 00 <br />GTR 0,67 1.32 0.04 1.09 - 00 <br />GSO 0,58 1.45 0,01 1.21-00 <br />GUN 0.60 1.51 0,02 1.20 - 00 <br /> <br />Somewhat encouraging suggestions resulted from <br />other single and dual partitions mentioned above but <br />not presented in detail here, There were suggestions <br />that effective seeding might have been associated <br />with the presence of SL W as detected by icing <br />sensors, with warmer than median HAS temperatures <br />and with less than median HAS wind speeds. But <br />none of these tested partitions provided results as <br />convincing as those already noted. It is suspected <br />that lack of a significant seeding signal for SL W <br />(icing) partitions was partial due to limitations in the <br />icing rate sensors discussed in the references, Most <br />ofthe resulting statistical sllggestions should be <br />regarded more as tantalizing than as conclusive. The <br />main underlying problem is very likely the limited <br />population sizes tested when any two partitions were <br />applied. If any actual seeding effects existed, the <br />available sample size was too limited to detect them <br />to an acceptable level of statistical significance, Such <br />results should be considereq inconclusive, neither <br />demonstrating an actual seeding effect nor the lack of <br />one. Only testing a larger population than available <br />would offer the possibility of providing more <br />definitive results, <br /> <br />The most solid and important results of the <br />statistical testing can be summarized as follows. It is <br />highly likely that a real seeding effect resulted from <br />propane seeding when transport winds carried the <br />seeding plume over the target gauges. Overall, <br />seeded units provided at least 20% more snowfall <br />than nonseeded units. Based on those results, it was <br />hypothetically estimated that if all hours containing <br />the 69 EUs had been operationally seeded, the overall <br />precipitation increase would have been 8% in the <br />"seasonal" precipitation averaged for gauges GSC <br />and GTR for the 3.5 month experimental period, <br />This result was provided both by considering the <br /> <br />mean residuals from the pairs testing and from the <br />mean double ratio testing emphasized above. <br /> <br />Case Study Analysis <br /> <br />A case study analysis was performed for three <br />adjoining EUs on 21 December 2003, with two of the <br />EUs nonseeded and the middle EU seeded. By <br />chance, this particular seeded EU coincided with a <br />time period having minimal natural snowfall, selldom <br />observed during other EUs. The adjoining nonsl~eded <br />EUs had considerably more natural snowfall than the <br />seeded EU, both with highest accumulations at the <br />control gauge. But that gauge received only 0,009 <br />inch during the seeded period while target gauges had <br />2 to 4 times as much SWE, increasing with distance <br />downwind of the seeding site. The most convincing <br />evidence that seeding produced much of the increase <br />was provided by examination of 2D-C probe i<:e <br />particle images from TAR. These were very similar <br />to those published from pulsed seeding experiments <br />conducted during earlier winters. As seen in Figure <br />5, based on the 2D-C probe observations of <br />individual ice crystals, the seeded images were <br />relatively small and generally uniform in size and <br />shape. They were in much higher concentrations <br />than observed during the adjoining nonseeded <br />periods which had much greater natural snowfall <br />rates. The first nonseeded period would be classified <br />as having heavy snowfall rates and the last as having <br />moderate rates. Even with conservative assumptions <br />detailed in the references, it was estimated that <br />seeded snowfall rates exceeded 0.010 inch per hour at <br />the core target gauges and likely approached twke <br />that rate at the downwind gauge, Such seeding- <br />produced precipitation rates are in good agreement <br />with earlier cases study experiments in Utah and <br />elsewhere in the Intermountain West. <br /> <br />9 <br /> <br />