Some features of the growing season precipitation fluctuations in the Interior Plains of North America

Long-term growing season precipitation series for the period 1931–1978 at 407 stations in the Interior Plains of North America and for a decreasing number of stations with records dating back to 1875 were subjected to various statistical analyses in order to understand their inter-annual and long-term variability. Only a few stations showed any sign of inhomogeneity, there being no evidence of any significant trend in the precipitation series. The higher degree of spatial variability of seasonal precipitation series over the southern Plains indicated a higher frequency of occurrence of thunderstorms in the sub-region compared with the northern Plains and the Canadian Prairies. The test for normality by Fisher's statistic and the Geary coefficient indicated that about 20 per cent of the precipitation series were not normally distributed. For meaningful statistical analysis, there is always the need to convert the frequency distribution of rainfall series from non-normal to normal. Using only the square root transformation method, most of the non-normal cases were transformed to normal. Subperiods analysis as tested by Cramer's test indicated that, although the 1941–1970 normal was wetter than the preceding non-overlapping 30-year normal period, there had been no significant secular change in the growing season precipitation in the Interior Plains, at least since 1911. Analysis of the precipitation anomalies decade by decade from the 1941–1970 averages indicated that the 1940s and 1960s were generally wet, while other decades, especially the 1930s and 1950s were particularly dry. The 30-year period 1941–1970 was probably the most anomalously wet period in the nearly 104-year time sequence considered. On the whole, the results of the statistical analyses indicated no significant changes in some of the secular features of growing season precipitation totals in the Interior Plains.