Laserfiche WebLink
To illustrate the impact of this process, consider the case of the July data for <br />“Watford 2” station in McKenzie County in NDCMP District 2. The records showed <br />data missing for four Julys out of the 53 years of interest. The range of observed values <br />* <br />from the other years was 0.83 - 7.68 inches, with a median of 2.96 inches (mean 3.32 <br />inches). The regression relationship for those 49 years yielded predictand values ranging <br />from 0.99 to 5.78 inches, in the ratio of from 0.29 to 3.40 times the observed value for the <br />year in question, and involving more than twice as many underestimates as overestimates. <br />For the four missing years the regression estimates were 1.84, 1.92, 3.75, and 6.21 inches <br />(average 3.43 inches). These values indicate regression estimates generally in line with <br />the observed values, and suggest that application of this procedure to about 1% of the <br />total database would not strongly influence the statistical results. <br />2.2Test variables <br />The Appendix lists those stations finally included in the analysis. With the small <br />numbers of gages available to represent the rainfall in such large areas (the NDCMP <br />target area exceeds 20,000 square kilometers and the Montana control area is more than <br />twice as large), meaningful estimates of the areal rainfall volumes are impractical. <br />Consequently, the simple sums of the gage data for each respective area and time period <br />were used as the test variables for this analysis. <br />3. Characteristics of the data <br />This section summarizes general characteristics of the primary (season total) data. <br />Table 2 presents some statistical features (medians, means, standard deviations, and <br />correlations) of the various data subsets. <br />The measures of central tendency, considered on a per-gage basis, reflect the <br />general gradient of precipitation in the region, increasing from west (control area) to east <br />(target area), as well as the generally drier conditions in southeast Montana (south <br />control). No significant temporal trends appear in the data, though the indication of a <br />downward trend for the south control approaches a significant level (a correlation of <br />0.229 would be significant at the 95% level). This suggestion of a downward trend for the <br />south control, coupled with the weak indication of an upward trend for the north control, <br />may be responsible for an interesting suggestion of a “seeding effect” when the August <br />data from these two non-seeded areas are compared (Section 5). <br />* <br /> Inch units are used in this report because the NOAA rainfall data are recorded in those units; 1 inch = 25.4 <br />mm. <br />5 <br />