Enhancing DSSAT–CERES-Wheat model long-term simulation of wheat growth under irrigation and tillage management in semi-arid regions

Sustainable wheat production in semi-arid regions of Iran is constrained by declining water resources. Crop simulation models such as DSSAT, combined with field experiments, provide effective tools to evaluate management strategies including conservation tillage and irrigation scheduling. This study evaluated the DSSAT-CERES-Wheat model to simulate crop growth, soil moisture dynamics, evapotranspiration, and water productivity of the Sirvan wheat cultivar under tillage systems and water stress regimes. A four-year field experiment (2017–2020) was conducted at the Zarghan Research Station, Iran, using a split-plot randomized complete block design. Tillage treatments were no-till (T1), reduced tillage (T2), and conventional tillage (T3), while irrigation included full irrigation and irrigation cutoff at anthesis (GS61), milking (GS71), and soft dough (GS83) stages. Model performance was assessed using root mean square error (RMSE), normalized RMSE (RMSEn), and coefficient of determination (R²). The DSSAT model simulated grain yield with high accuracy during calibration (RMSE = 419.22 kg ha −1 , R² = 0.90), while validation accuracy decreased under severe water stress conditions. Above-ground biomass was underestimated under severe water stress. However, leaf area index and soil organic carbon were simulated accurately across both growth stages. Soil moisture dynamics were well represented (RMSEn = 5–27%), with higher accuracy under full irrigation and improved moisture retention under no-till systems. Scenario analysis indicated that although full irrigation maximized yield, conventional tillage combined with regulated deficit irrigation at GS83 provided the most sustainable balance between yield and water productivity. Overall, the calibrated DSSAT-CERES-Wheat model proved reliable for optimizing wheat management under semi-arid conditions.