Assessing the impact of soil water deficit and supplemental irrigation scenarios on Ohio’s maize and soybean yields using machine learning models

To understand the relationship between climate variables and irrigation requirements with crop yield, this study evaluates the role of monthly precipitation, temperature, soil water deficit (SWD), and supplemental irrigation on the spatio-temporal variability of county-level maize and soybean yields across Ohio. We combined a soil water balance approach with machine learning to identify key climatic and soil hydrological variables influencing the effects of supplemental irrigation on county-level maize and soybean yields. To model the effect of SWD and weather parameters on yield variability, the Random Forest model performed best with an RMSE of 0.60 Mt/ha and an R² of 0.77 for maize, and with an RMSE of 0.21 Mt/ha and an R² of 0.64 for soybean yields. Maize yields were most influenced by July soil water deficit, September maximum temperature, and August precipitation, whereas soybean yields were primarily affected by precipitation in May and August. Supplemental irrigation of 50.8 mm/month during the summer improved maize yields more than soybean yields, with an average maize yield increase of 598 kg/ha (∼0.6 Mt/ha) relative to rainfed conditions. However, a statistically significant reduction in inter-annual yield variability was found with supplemental irrigation under all conditions for both crops. Irrigation beyond 50.8 mm/month did not yield further significant gains. Yield improvement varied over the years, with higher improvement seen during dry years, particularly for maize and it was more pronounced in years with cooler September months. Southwest Ohio showed a higher average yield (1991–2022) increase for maize, with an average increase of up to 22.6 %, which corresponds to an increase of 1224 kg/ha (∼1.2 Mt/ha). For soybeans, an average yield increase of up to 9.2 % was found, which corresponds to an increase of 207.5 kg/ha (∼0.21 Mt/ha), and counties with higher average yield increases were found in different regions of Ohio. • Machine Learning and Soil Water Deficit (SWD) were used to quantify weather and irrigation effects on Ohio yields. • Maize yield was most strongly influenced by July SWD and September maximum temperature. • Soybean yield was primarily driven by precipitation during August and May. • Soybeans were found to be less sensitive to drought and less responsive to supplemental irrigation than maize. • Spatial and temporal patterns of yield gains with supplemental irrigation varied with climate and crop type.