Spatiotemporal patterns of maize and winter wheat yields in the United States: Predictability and impact from climate oscillations

Studies have shown linkages of climate oscillations with climate extreme events, such as floods and droughts, which may induce risks in summer and winter crop productions. The goal of this study is to explore spatial and temporal variability of a summer crop (maize) and a winter crop (wheat) yields and its linkages with influential climate oscillations in the United States. The county level yield data over 1960-2016 for maize and winter wheat were aggregated into each of the 260 climate divisions in the rainfed regions of the United States, with the linear yield trend being removed. The rotated Principal Component Analysis (PCA) reveals that the first five principal components explain 79% (maize) and 72% (winter wheat) of the spatial and temporal variability of crop anomalies. The first principal component of crop yield variability is strongly associated with the Atlantic Multidecadal Oscillation (AMO). The results of multiple linear regressions for predicting yield anomalies using climate indices show that, climate indices during the reproductive period of maize explained final yield better than the vegetation period (30% versus 26%), while climate indices for winter wheat during the dormant and reproductive growth periods are similar and not significant (25% versus 28%). Categorical yield forecasts using random forecast techniques show that the low (below 30th percentile) and high (above 70th percentile) yields are well predicted by climate indices. Spatially, AMO is identified as the most important predictor for maize in 46% climate divisions and for wheat in 33% climate divisions. The results from this study may contribute to understanding the risks of large-scale climate oscillations to local-scale crop production and improving crop yield predictions.