Variability and trends of corn heat units on the Canadian prairies

Seasonal accumulations of corn heat units were calculated from temperature records of 85 stations in Alberta, Saskatchewan, and Manitoba. The number of years of data varied from 16 to 72. The objectives of the study were to determine how many years of data were necessary to obtain a reliable estimate of CHU for each station and to examine possible CHU trends with time. Stein's two-stage sampling technique was used to show that about 20 years of data should be used to obtain a reliable estimate of the location mean. When years were taken at random, the number of years of data needed for a location was larger than when taken in sequence. Statistically significant indications of trends in CHU with time were found for only about 11% of the stations. Year-to-year variability was considerably greater than any possible trends in the last half-century. Means and standard errors of the deviations from the 30-year mean (1943–1972) were calculated for all stations. Although these data were not independent because of their geographic proximity, there was an indication that CHU totals have changed in the past half-century. However, future changes in the CHU totals could not be predicted.