Laserfiche WebLink
1. Introduction <br /> The primary objective of the North Dakota Cloud Modification Project <br />(NDCMP), which has operated in western North Dakota since 1976, is to reduce losses <br />due to hail damage. Stimulation of enhanced rainfall is a secondary objective, and in any <br />case knowledge of hail suppression effects on total precipitation would be important to <br />the project’s operation. This report summarizes an exploratory analysis of rainfall data <br />for the NDCMP target areas and an upwind control region in eastern Montana, intended <br />to elucidate any such effects. <br /> The planned approach was to use techniques similar to those employed in Smith <br />et al. (1997) to explore the effects of the NDCMP seeding on crop-hail insurance losses. <br />In brief, the approach uses data from the target and from upwind control areas, for both <br />the seeding period of interest and a prior historical period, in a statistical analysis based <br />on the multi-response permutation procedures (MRPP; Mielke and Berry 2001). <br />Techniques like those described in Smith et al. can provide estimates of the probability <br />that any observed differences could be due to random chance, along with point and <br />confidence-interval estimates for any difference that may be found statistically <br />significant. <br /> Two differences from the earlier hail insurance data analysis procedure were <br />initially considered useful. One is improvements in the statistical procedure to use <br />multivariate multiple least-absolute-deviation (LAD) regression techniques (Mielke and <br />Berry 2002). The other is to subdivide the control area, consisting of the same 12 <br />counties in eastern Montana used in the hail insurance analysis, into a north control and <br />south control. That would permit two separate target versus control analyses, yielding <br />some indication of the stability of the results, along with a north versus south analysis <br />which would give some sense of the possibility of finding a “seeding effect” in these <br />areas where no seeding in fact occurred. <br />2. Climatic raingage data <br /> The North Dakota Atmospheric Resource Board (NDARB) operates a statewide <br />network of some 900 reporting stations (rain and hail) during its summer – usually June <br />through August – operating season. This provides a mean gage density of about one gage <br />2 <br />per 200 km, and data which would have obvious value because of the close spacing. <br />Unfortunately, neither upwind control-area data from Montana nor historical data prior to <br />the inception of the NDCMP are available from this network. Consequently the NDARB <br />data do not lend themselves to the type of analysis carried out here. Scientists at the <br />University of North Dakota are carrying out a separate analysis of the NDARB data, and <br />it will be reported elsewhere. <br />Climate stations operated as part of the NOAA climatic network measure rainfall <br />along with other quantities of interest. A few of these stations in the NDCMP target area <br />have records dating back at least as far as 1894, but the earliest Montana data are from <br />1911. The data from this network were chosen as the basis for this analysis; we focused <br />2 <br />