Download PDF for Winter 2015
Agricultural Policy Review home

Late-1990s Climate Shift Impact on Corn Yield in Iowa

Christopher J. Anderson (cjames@iastate.edu), Bruce A. Babcock (babcock@iastate.edu), Yixing Peng (pyixing@iastate.edu), Philip W. Gassman (pwgassma@iastate.edu), and Todd D. Campbell (tdc@iastate.edu)

The next advance in climate science will come out of experiments in forecasting shifts in climate regimes—an extended period of time in which weather conditions have consistent range, such as the Dust Bowl years or the Little Ice Age. A climate regime shift results in a new range of weather conditions for an extended period, so being able to predict a regime shift allows planners to anticipate an emerging weather risk profile that would be expected to persist for 20–30 years. One way a regime change occurs is when slowly varying ocean surface temperatures change from warm to cold. In the Corn Belt, summer rainfall is influenced over 20–30 year periods by two recurring ocean surface temperature patterns: the Pacific Decadal Oscillation (PDO) (Hu and Feng 2001) and the Atlantic Multidecadal Oscillation (AMO) (Hu et al. 2011). Together they have four phases of warm and cold conditions that result in four different spatial patterns for drought risk across the United States (McCabe et al. 2004). While climate scientists will focus on decadal forecast capability for broad temperature and rainfall patterns, the more immediate question for agriculture is, how have climate regime shifts affected yield?

Our approach is to evaluate corn yield response to the late-1990s climate regime shift. From 2000 to 2014, weather volatility produced corn yield shocks that dominated market prices, particularly in 2008, 2010, 2012, and 2013. In 2014, the absence of poor growing conditions sent the corn market down to prices last seen in 2006–2007. The objective of this research is to determine whether corn yield shocks during 2000–2012 resulted from different weather extremes than experienced in 1980–1992.

Table 1. Names and descriptions for predictor variables
Table 1

We develop an empirical model that relates a logarithm of county-level corn yield to temperature, rainfall, and soil moisture. This means our model predicts the change of yield rather than yield itself. We use model predictors based on corn phenological stage development (Table 1) in order to examine interaction among weather extremes, such as 2011 when wet conditions in spring were followed by dry and hot conditions in summer. Model parameters are estimated by the method developed in Yu and Babcock (2011).

Data

Corn production and planted acres are obtained for all 99 Iowa counties from the US Department of Agriculture National Agricultural Statistics Service, and yield is constructed as corn production divided by planted acres. Daily temperature and rainfall data are obtained from values in a one-eighth degree grid dataset produced by an interpolation routine applied to daily measurements of more than 10,000 stations across the United States (Maurer et al. 2002). Temperature and rainfall are aggregated to county scale.

Soil moisture is not widely measured. We use EPIC model version 1102-64 (Izaurralde et al. 2006) to produce a 1980–2012 simulation of soil moisture at 48,084 points from the 1997 Natural Resources Inventory (NRI) from western Minnesota through central Illinois. Each NRI point is provided the 1980–2012 gridded daily weather from the grid point nearest the NRI centroid.

Results

We evaluate weather changes between 1980–1992 and 2000–2012 by comparison of mean values of the predictors. We compute mean values for the entire thirteen year period and focus on volatility with means for only hot-dry summers within the periods. Period mean values are statistically different for all variables except July–August rainfall. The mean growing season conditions in 2000–2012 compared to 1980–1992 began with drier May 1 soil moisture and progressed to wetter and cooler May–June, wetter July 1 soil moisture, and a cooler July–August. Weather during these two regimes is different, but the yield effect of these factors is mixed, resulting in the model predicting a 2.33 bu ac-1 net increase in state average yield. For reference, the model estimates a yield trend of 1.56 bu ac-1, such that the yield effect of the 1990s regime shift is equivalent to 1.5 years of advancement in technology.

Figure 1. May–June average rainfall for 1981–1992 (left) and 2000–2012 (right) for crop reporting districts
Figure 1

We point out some aspects of spring rainfall increase, because it presents complicated tradeoffs for machinery decisions, drainage, timeliness of planting, and resilience to summer drought. The change in spring rainfall is not unique to Iowa (Figure 1), but is a large pattern shift across the Corn Belt. In Iowa, yield loss from late planting occurs after May 10–14, such that expected yield loss at May 31 is 10 percent and 30 percent on June 15 (Farnham 2001). Negative correlation between average suitable fieldwork days from April 1 to May 15 and average April–May rainfall in Iowa during 1976–2010 is clear, and a linear regression predicts a reduction of 2.2 fieldwork days for every one inch increase in April–May rainfall. The increase of 1.3 inches in Iowa average April–May rainfall suggests a decrease of roughly three suitable fieldwork days.

We are highly interested in how the climate regime affects yield volatility, and we focus discussion on years with high temperatures in July–August. The role of July 1 soil moisture and July–August rainfall in ameliorating high temperature yield effect is clear in our model predictions. The percent yield loss in Iowa under high July–August temperature drops from 26.25 percent to 10.89 percent if both soil moisture and rainfall are abundant in July–August. Weather during hot-dry summer years is statistically different during the two periods for all variables except May–June rainfall and May–June temperature (Table 1). Comparing 2000–2012 to 1980–1992, the average hot-dry summer growing season sees 2.5 inches more rainfall in spring—adding 0.66 inches to the July 1 soil moisture reservoir—and summer sees 1.5 inches more rainfall and temperatures one degree Fahrenheit cooler. The yield impact of different growing seasons for hot-dry summers is substantial. Our model predicts smaller yield losses from cooler July–August temperatures, more July–August rainfall, and more July 1 soil moisture of 12.6, 11.9 and 4.5 bu ac-1. The effect of May–June rainfall is positive, because of the positive yield effect from July 1 soil moisture, despite its impact on planting delay. The net yield effect of all weather factors during hot-dry summers is a reduction of yield loss by 25.3 bu ac-1.

Final thoughts

The results show the power of knowing yield effects under climate regimes, and it suggests substantial value to forecasts of climate regime shifts if they prove to be skillful. Iowa agriculture can suggest priorities to this work. An immediate priority is clear from the historical sequence of PDO and AMO phases that suggest a combination could occur within the next decade that has higher drought risk. There is urgency for agriculture, then, to identify differences in weather seasonality under past climate regimes and translate this to yield effects. We can then evaluate whether the recent trend of wet springs is characteristic of past regimes, and what types of investments can be made when a regime shift occurs.

References
Farnham, D. 2001. “Corn Planting Guide (PM 1885)”. Iowa State University Extension. Available at https://store.extension.iastate.edu/Product/pm1885-pdf.

Hu, Qi, and Song Feng. “Variations of Teleconnection of ENSO and Interannual Variation in Summer Rainfall in the Central United States.” Journal of Climate 14.11(2001): 2469–2480.

H, Qi, Song Feng, and Robert J. Oglesby. “Variations in North American Summer Precipitation Driven by the Atlantic Multidecadal Oscillation.” Journal of Climate 24.21(2011): 5555–5570.

Knudsen, M.F., M.S. Seidenkrantz, B.H. Jacobsen, and A. Kuijpers. 2011. “Tracking the Atlantic Multidecadal Oscillation through the last 8,000 Years.” Nature Communications 2: 178.

Mantua, N.J., and S.R. Hare. 2002. “The Pacific Decadal Oscillation.” Journal of Oceanography 58(1): 35–44.

McCabe, G.J., M.A. Palecki, and J.L. Betancourt. 2004. “Pacific and Atlantic Ocean Influences on Multidecadal Drought Frequency in the United States.” Proceedings of the National Academy of Sciences 101(12): 4136–4141.

Maurer, E.P., A.W. Wood, J.C. Adam, D.P. Lettenmaier, and B. Nijssen. 2002. “A Long-Term Hydrologically-Based Data Set of Land Surface Fluxes and States for the Conterminous United States.” Journal of Climate 15(22): 3237–3251.