Rainfall Evaluations

Some of the most important fundamental research in North Dakota took place during the North Dakota Pilot Project (NDPP), a randomized cloud seeding program conducted in McKenzie County from 1969-1972 (Dennis et al., 1975). The NDPP found statistically-significant results that silver iodide seeding of convective clouds on a determinate set of days leads to: (1) an increase in the frequency of rainfall events at the target gauges (0.04 significance level, or 96% confidence), (2) an increase in the average rainfall recorded per rainfall event (0.02 significance level (98% confidence)), and (3) an increase in total rainfall on the target area (0.07 significance level (93% confidence)). These results apply on days with dynamic seedability, that is, days when a cloud model predicted an increase in cloud top height under the influence of silver iodide seeding. The authors estimated there were about 50 days with dynamic seedability each summer during the NDPP. Further, they estimated a potential increase of one inch of rainfall per growing season for western North Dakota from cloud seeding.

Several rainfall evaluations have been conducted on the NDCMP: two looking at ARBCON rainfall data (Wise, 2005 and Tuftedal et al., 2022) and the other at NWS climatic rain gauge data (Smith et al., 2004). The ARBCON rainfall analyses, which take advantage of hundreds of rain gauge locations in western ND, indicated that between the years 1977-2003, the seeded areas and those areas slightly downwind received about 6% to 9% more rainfall than the upwind control areas where no seeding occurred (Wise, 2005). Tuftedal et al., found rainfall increases in six of nine target/control pair comparisons, with two of the ratios having a 95% statistical chance of being greater than 1.0. This is not only positive news for the seeded areas, but it also addresses the concern of some that seeding upwind will reduce downwind precipitation, or “steal someone's rain,” (e.g., DeFelice et al., 2014). Conversely,there is no evidence that cloud seeding contributes to large increases in precipitation at long distances downwind of seeded areas. Previous studies of the NDCMP and other states' seeding programs indicate that downwind effects diminish with increasing distance from the target area. The Smith et al. (2004) study found no statistical difference in rainfall between the NDCMP counties and an upwind control area. The author concluded, however, that the sparse coverage of NWS rain gauges in both areas and the ±0.1 confidence interval would not allow the analysis to determine differences in precipitation over the two areas at a level of ten percent or less, the typical range of effect identified by several other rainfall studies.

One of the earliest evaluations of the effects of cloud seeding on summertime clouds over the plains was produced by Changnon and Huff (1972). A model was developed at the Illinois State Water Survey which projected the increases in precipitation likely to be realized by cloud seeding, based on efforts in Illinois. That model, along with projections specifically developed for North Dakota by Eddy and Cooter (1979), were used by North Dakota State University researchers to develop an estimation of the economic benefits of the North Dakota program (Schaffner, et al. 1983). This evaluation indicated that, for the six most prevalent crops in western North Dakota (wheat, oats, barley, sunflowers, corn grain, and flax), benefits from the additional one inch of rainfall obtainable by cloud seeding ranged from $3.63 to $6.21 per acre, depending upon the region of the state involved.

Another evaluation of rainfall in and near the project areas examined NWS rainfall data for seven years, 1976-82 (Johnson, 1985). Three areas were defined for each storm event: the multi-county target areas (Target), the regions downwind of the Targets (Downwind), and the areas neither in the target nor downwind (Control). Some of the more interesting findings of this analysis showed: 1. Evidence of an overall increase in precipitation downwind of the target (relative to the control area), especially notable in July (15% increase, significant at the 0.08 level). 2. Evidence of increase in the target areas on days with heavier rains (14%increase significant at the 0.13 level). Other findings showed additional increases both in and downwind of the target areas, correlated at various (generally lower) confidence levels with upper air features. The report concluded that, "It appears that seeding for hail suppression in the mode employed in North Dakota likely increases rainfall."

Lastly, daily comparisons of seeded and non-seeded rainfall in North Dakota were conducted from 1976 to 1980 (Eddy 1981). These comparisons used a dense statewide network of more than 500 rain gages and found average rainfall increases of about 15% during the June 6 to July 11 critical agricultural period. There was an increase in the number of stations reporting rain in and downwind of the seeded areas and an increase in the average rain which fell in each rain gage.