Improving evapotranspiration simulations in the CERES-Maize model under limited irrigation

Limitations on water resources for agriculture in places such as Colorado, USA, have caused farmers to consider limited irrigation as an alternative to full irrigation practices, where the crop is intentionally stressed during specific growth stages in an effort to maximize yield per unit water consumed, or evapotranspiration (ET). While crop growth models such as CERES-Maize provide the ability to evaluate numerous management scenarios without the costs associated with multiyear field experiments, recent studies have shown that CERES-Maize performs well under full irrigation but overestimates ET of corn under limited irrigation management. The primary objective of this study was to improve CERES-Maize ET simulation under limited irrigation management while maintaining accuracy of other important model output responses. Field experiments with corn were performed in northern Colorado, USA from 2006 to 2010, where four replicates each of full (ET requirement supplied by irrigation throughout the season) and limited (no irrigation before the V12 growth stage unless necessary for emergence, then full irrigation afterwards) irrigation treatments were analyzed. The local sensitivity of model input parameters affecting ET was evaluated, prompting changes to the model code with a new dynamic crop coefficient (KCD) as a function of the crop leaf area index. The modified CERES-Maize model more accurately represented ET under full and limited irrigation, for example reducing late-season ET potential from a plant with reduced canopy and more closely matched FAO-56 crop coefficient curves under full irrigation. Using the limited irrigation data for evaluation, the modified model showed significant decreases in model error for seasonal cumulative ET (root mean square deviation RMSD from 80.9mm to 49.9mm) and water productivity (RMSD from 5.97kgha⁻¹mm⁻¹ to 2.86kgha⁻¹mm⁻¹) as compared to the original model. The modified model was subsequently applied to several hypothetical irrigation management strategies, indicating that reducing weekly vegetative stage water applications from 20mm to 2.5mm can increase simulated water productivity by over 15%. While these synthetic water production functions may not be feasible in a production field with natural climate variability, the modified ET model indicates promise for limited irrigation management increasing water productivity.