An exploratory analysis of crop hail insurance data for evidence of cloud seeding effects in North Dakota

Abstract

The basis for the cloud seeding operations of the North Dakota Cloud Modification Project (NDCMP) is first outlined. Then the multiresponse permutation procedures are applied in an analysis of crop hail insurance data for the NDCMP target area and for an upwind control area in eastern Montana. A historical analysis of the annual hail insurance loss ratios for the target area indicates lower hail-loss experience during the NDCMP operational years 1976-88. A corresponding analysis for the control area shows no indication of a difference during those years, suggesting the absence of any significant climatological variation. Analysis of a target-control scatterplot of the loss ratios also indicates that the target area experienced relatively smaller hail losses during the NDCMP period. An inference that the difference can be attributed to the NDCMP seeding operations appears to be justified, and the reduction in hail insurance loss ratios in the target area during the NDCMP years is estimated to be about 45%.