Water, temperature, and crop productivity research

Share Tweet Email

This article is part 3 of a 4-part series focusing on agriculture, climate, and change. The first two articles in the series, “Is Kansas agriculture likely to be affected by a changing climate?” and “A brief on global research on drought monitoring” can be found on the eUpdate website in Issues 710 and 712, respectively.


Planting an agricultural crop requires a degree of optimism. In semi-arid western Kansas, water and temperature frequently limit crop productivity. These components of weather, along with sunshine and relative humidity, comprise the weather-related risks which limit the productivity of the crop just planted. Several questions arise concerning weather and climate:

  • Are there periodic behaviors in weather patterns?
  • Are there long-distance signals indicating wetting and drying trends?
  • Is long-term weather forecasting feasible? If so, accurate forecasts can inform the optimism required to plant that crop, infusing an additional hope that the “bet has been hedged”.

Weather forecasting skill is related to ‘teleconnections,’ a topic addressed at the 2018 American Meteorological Association meetings. The El Nino-Southern Oscillation (ENSO) phenomena serves as an example. Warming and cooling trends in the surface waters of the equatorial Pacific Ocean impact fisheries and rainfall in coastal Peru. Indeed, ENSO trends impact the productivity of winter wheat growing in the Texas High Plains. Louis Baumhardt, an USDA-ARS soil scientist, and his colleagues found a degree of association between ENSO patterns and winter wheat yields in the Texas Panhandle (1). Does this ENSO signal convey information about wheat productivity further north, in the Central High Plains?

Winter wheat is vulnerable to drought conditions; wheat can also respond positively to wet conditions, though subject to disease impacts. The Standardized Precipitation-Evapotranspiration Indicator (SPEI) provides a metric for wetting and drying conditions, generally varying between values of -4 and 4 to indicate drying (negative) and wetting (positive) conditions. Aiken et al (2017) compared wheat yields, reported for counties (USDA-NASS) in Kansas (1970 through 2007) against monthly SPEI values, after removing linear historic trends attributed to improved genetics and production technologies (2). R-squared (r2) is a statistical measure that indicates the fraction (0 -1.0) of observed variation which can be accounted for by a regression relationship. Generally, the higher the R-squared value, the better the regression model fits the data. A moderate relationship (R2 = 0.41) emerged for wheat yields reported for counties in western Kansas, indicating positive effects of weather conditions in February, March and April. A weaker relationship (R2 = 0.25) resulted for counties in sub-humid eastern Kansas, indicating both positive (October, February, April) and negative (August, December, May, June) relationships with the SPEI metric. This regression analysis quantified the relationship of winter wheat productivity to weather variation during the growing season. However, the utility of forecasting skill depends on information available prior to planting decisions.

Thus, we evaluated a hypothesized ENSO signal: Is winter wheat grain productivity in western Kansas related to equatorial Pacific Ocean surface temperatures in preceding years? We tested this hypothesis using multiple regression for western Kansas county yield reports and monthly ENSO data for the 24-month period prior to wheat planting (September, year prior to harvest). We found that the answer is a resounding YES. A stronger relationship (R2 = 0.53) resulted from regression analysis (Figure 1).


Figure 1. The in-sample predictive ability for wheat yield variation of multiple regression based on the NINO_3 surface temperatures of the Equatorial Pacific is compared against observed yield variation for wheat yield in Kansas counties W of the 99th Meridian. Observed variation represents variation after removal of a linear historic trend (1970 – 2007 period) as normalized by dividing by the standard deviation of the time series for each county.


Interestingly, the strongest influences were ENSO values 18- and 16-months prior to the wheat-planting period (Figure 2). This indicates that complex patterns in equatorial Pacific Ocean temperatures can convey information that is pertinent to subsequent winter wheat yields in western Kansas.

Figure 2. Regression coefficients of the NINO_3 multiple regression model are shown in relation to corresponding time lag (months prior to a September planting period for winter wheat in Kansas). Positive coefficient values for a given time lag indicate a positive association with expected wheat yields for that lag interval; negative coefficient values indicate a negative association.


In summary, there is opportunity to develop climate-informed decision-support tools for cropping systems in the Central High Plains.

Stay tuned for the last installment of this series focusing the creeping trends of a changing climate.



Rob Aiken, Crops Research Scientist, Northwest Research-Extension Center

Xiaomao Lin, State Climatologist



1) Baumhardt, R.L. S.A. Mauget, R.C. Schwartz and O.R. Jones. 2016. El Nino Southern Oscillation effects on dryland crop production in the Texas High Plains. Agron. J. 108:736-744

2) Aiken, R., X. Lin and Z. Zambreski. 2017. Winter wheat yield responses to climate variation in the U. S. Central Great Plains. ASABE Paper No. 1701661 DOI: https://doi.org/10.13031/aim.201701661

Tags:  climate weather