Comparing crop growth models across the contiguous USA with a focus on dry and warm spells

Efficient water use and sustainable crop production are vital for ensuring both food and water security amidst the growing global population and changing climate. With expected increases in hydroclimatic extremes under climate change, irrigated agriculture is projected to rise, further interlinking food and water security. Hence, the need for future crop yield projections has become more urgent, including knowledge about the accuracy of crop growth modeling in terms of crop yield and water consumption. To this end, our study evaluates two crop growth models—AquaCrop-OS and PCR-GLOBWB 2-WOFOST. The coupled hydrological-crop growth model PCR-GLOBWB 2-WOFOST was applied using both its original settings and harmonized with AquaCrop-OS input, resulting in two model experiments that were compared to the AquaCrop-OS results, allowing analyses of differences in both model structure and parametrizations. AquaCrop-OS and PCR-GLOBWB 2-WOFOST (with original and harmonized input) were used to simulate the yield of irrigated maize, soybean, winter wheat and spring wheat across the contiguous United States (CONUS) from 2001 to 2019. Model outputs were compared in terms of crop water consumption, biomass production, and crop yield and the model sensitivity under dry and warm conditions was evaluated. For all considered crops, PCR-GLOBWB 2-WOFOST (original) simulated yields align closely with mean reported yields, with the harmonized version showing some underestimates for soybeans and winter wheat (∼23 %). AquaCrop-OS generally overestimates crop yields (up to ∼34 %). Our results also highlight significant differences in spatial patterns of crop water consumption, biomass, and crop yield between AquaCrop-OS and PCR-GLOBWB 2-WOFOST (original and harmonized) models. Specifically, differences between the AquaCrop-OS and PCR-GLOBWB 2-WOFOST (harmonized) models are attributed to variations in how biomass is converted to yield within each model. Also, AquaCrop-OS is found to be much less sensitive to higher air temperatures than the PCR-GLOBWB 2-WOFOST model. Finally, the differences between PCR-GLOBWB 2-WOFOST original and its harmonized settings emphasize the importance of using local- and up-to-date information on crop-specific parametrization in crop growth modeling. These findings underscore the importance of model selection and parameterization in accurately simulating crop yield and crop water consumption under climate extremes, which is essential for improving agricultural practices under climate change.