This study estimates the yields of maize, soybeans, and wheat as a function of daily interaction of soil moisture and heat. We apply this approach to investigate the impact of each combination of heat-moisture interval on crop yields. It also demonstrates the advantages of using soil moisture metrics over current proxy variables in capturing climate-driven variations. The hypothesis is that soil moisture index and its daily interaction with heat perform better in predicting crop yields in the US compared to the commonly used proxy variables such as precipitation. Specifically, we investigate: 1) the marginal impacts of mean soil moisture content on crop yields; 2) the marginal impact of soil moisture extremes on crop yields, and 3) the marginal impact of compound heat-water extremes on crop yields. This work extends Schlenker and Roberts (2009) model by considering compound heat-water extremes. An important contribution is the introduction of a daily cumulative yield production function considering the interaction of heat and soil moisture. We also suggest a revision of the concepts of beneficial heat (below critical temperature threshold) and harmful heat (above critical temperature threshold) by integrating heat-soil moisture extremes into definitions. The temperature and precipitation are taken from temperature from PRISM (Parameter-elevation Regressions on Independent Slopes Model) as provided by Schlenker and Roberts (2009). We pair a fine-scale daily dataset of precipitation and new time series on predicted daily root-zone soil moisture from WBM (Water Balance Model) driven by the PRISM data set, then we estimate determinants of the county-level yield of maize, soybeans, and wheat as reported by USDA-NASS.

We find that compound heat-water extremes are better predictors of agricultural yields than conventional separated heat and water stress. We also find that a hot-humid incident has a negative impact on yield. It also makes sense as a high soil moisture content can go together with high humidity and nutrient loss through water flows. At high humidity, the plants may have difficulty remaining cool at high temperatures. It is likely also due to waterlogging of soils. This framework is applicable to other climate impact studies.