Laserfiche WebLink
<br />of the study area, and all but the northeast comer of the target area, are statistically significant at the <br />5-percent level (see figure 3.7). In other words, there is a high probability that the differences <br />between seed-period rainfall and no-seed-period rainfall indicate a real shift in rainfall patterns. <br /> <br />3.4.2 Monthly Patterns <br /> <br />~ <br /> <br />There is considerable variation in seed/no-seed ratios from month to month. For April and May, <br />much of the study area has ratios greater than 1.1 (all of the target area in April). In June and July, <br />most ratios throughout the study area are less than 0.9. August and September have somewhat <br />larger areas with ratios greater than 1.1, but the target-area ratios are generally less than 1. <br />Generally speaking, the target area has patterns of seed/no-seed ratios not much different than <br />those of the entire study area. Figure 3.8 shows the ratios for July as an example. <br /> <br />The Wilcoxon rank-sum test results also vary from month to month. For April, the ratios in <br />approximately the western half of the study area (and all but a few cells in the eastern portion of the <br />target area) are shown to be statistically significant (5-percent level). In July, much of the study <br />area and the target area shows significance at no better than the to-percent level. May, June, <br />August, and September have smaller areas of significance (5-percent level), with no distinct pattern <br />to the probabilities. The July values are presented in figure 3.9. <br /> <br />It should be pointed out that, although the April statistics indicate more rainfall in the 1975-85 <br />period, the difference cannot be ascribed to seeding. There was no seeding in April in 1979 <br />through 1983 and little seeding the other 6 years. <br /> <br />3.5 Discussion <br /> <br />Examination of the mean seasonal rainfall values calculated from the National Weather Service <br />cooperative observer network shows that the mean rainfall in the target area was greater in the <br />no-seed period than in the seed period. The rainfall in the three control areas was also greater in . <br />the no-seed period, though the difference was not as large as in the target area. <br /> <br />The pattern of observed rainfall could be explained in either of two ways: (1) seeding acted to <br />decrease the rainfall in the target area or (2) the seed period was, for some reason, naturally drier <br />than the no-seed period, and seeding had little or no effect. <br /> <br />Analysis of the gridded rainfall data also indicates, in general, less rainfall in the target area during <br />the seed period. The seed/no-seed ratios indicate statistically significant decreases in rainfall in <br />some of the study area, including parts of the target area. However, the patterns of both the ratios <br />and the significance levels in the target area are similar to those for the entire study area. There is <br />little about the target area to distinguish it from the rest of the study area. <br /> <br />It is interesting to note that the band of strongest rainfall decreases that runs north-northeastward <br />from the southwestern corner of Kansas (see figure 3.6) is broken in the target area. It is possible <br />that this apparent lessening of the seed-year rainfall deficit is related to the seeding. The Wilcoxon <br />probabilities for this same area are at best marginally significant. <br /> <br />~ <br /> <br />When the locations of release of seeding material are compared with the rainfall patterns, one sees <br />little, if any, relationship. If the mean wind at cloud level is assumed to be about 10 m/s from a <br />generally westerly direction, and if it is assumed that the time required for the seeding <br /> <br />21 <br />