|
The point estimates and confidence intervals for the additional exploratory
<br />analyses, for which the P values appear in italics in Table 3, are quite similar for all the
<br />season-total comparisons to the primary-analysis values. For the monthly comparisons
<br />the confidence intervals are 3 – 5 times wider, and all straddle 1.0. In two cases (August,
<br />Target vs. N Control; June, N Control vs. S Control) the point estimate exceeded 1.1, and
<br />in one case (August, N Control vs. S Control) it was less than 0.9. In view of the
<br />associated P values, discussed earlier, and the wide confidence intervals, no significance
<br />can be attached to those values.
<br />6. Summary and Conclusions
<br />This analysis of the climatic rain gage data from the NDCMP target area and
<br />upwind control areas in eastern Montana has yielded no significant evidence of an effect
<br />of the NDCMP seeding on the summer-season rainfall in the target area. While there may
<br />in fact be no such effect, a small effect might not show up in this analysis. For example,
<br />an analysis of wheat yield data (Smith et al. 1992) suggested an increase of about 6% in
<br />the NDCMP target areas that could be attributed to the seeding activity. With the 0.1
<br />confidence interval indicated in Table 3, even a 6% effect on the rainfall (and part of the
<br />wheat-yield effect would reflect reductions in hail losses) would be difficult to find in the
<br />climatic rain gage data. Furthermore, the limited number of climate gages available for
<br />use in this analysis, the limited usable period of record, and the fact that seeding was
<br />taking place in the target area prior to 1976, all may have diluted any indication of a
<br />seeding effect under this analysis procedure.
<br />References
<br />List, R., K.R. Gabriel, B.A. Silverman, Z. Levin and T. Karacostas, 1999: The rain
<br />enhancement experiment in Puglia, Italy: Statistical evaluation. J. Appl. Meteor.,
<br />38
<br />, 281-289.
<br />Mielke, P.W., Jr. and K.J. Berry, 2001: Permutation Methods: A Distance Function
<br />Approach. Springer, 352 pp.
<br />Mielke, P.W., Jr. and K.J. Berry, 2002: Multivariate multiple regression analyses: A
<br />91,
<br />permutation method for linear models. Psychological Reports, 3-9 (Erratum:
<br />91,
<br />2).
<br />Smith, P. L., L. R. Johnson, D. L. Priegnitz, and P.W. Mielke, Jr., 1992: A target-control
<br />analysis of wheat yield data for the ND Cloud Modification Project region. J.
<br />24,
<br />Wea. Modif., 98-105.
<br />Smith, P. L., L. R. Johnson, D. L. Priegnitz, B. A. Boe, and P. W. Mielke, Jr., 1997: An
<br />exploratory analysis of crop-hail insurance data for evidence of cloud-seeding
<br />36,
<br />effects in North Dakota. J. Appl. Meteor., 463-473.
<br />10
<br />
|