Seventy-eight years (1900--1977) of corn yield, temperature and precipitation data for the continental U.S.A. are used to extend crop---weather relationships to a spatial scale compatible with the planetary (stationary) atmospheric waves. Both the meteorological and yield anomaly data are represented in terms of principal components.
The year-to-year corn yield variability, expressed as fractional departures from running mean yields, is highest in the Great Plains. The dominant patterns of corn yield, temperature and precipitation show strong spatial coherence over scales of thousands of kilometers. The amplitudes of the dominant yield anomaly patterns are significantly correlated with the monthly amplitudes of several temperature and precipitation patterns during the growing season. The fractions of the yield variance attributable to meteorological fluctuations during individual months within the growing season are as large as 0.30--0.35. The corresponding fractions are 0.60--0.70 when meteorological fluctuations throughout the growing season are included in the analysis. The results suggest an optimum length of 11--13 y for the running mean used to remove temporal trends in the yield data.
Spring, MD 20910, U.S.A.