The distributed CERES-Maize model with crop parameters determined through data assimilation assists in regional irrigation schedule optimization

Abstract

Crop model–based irrigation scheduling is essential for improving crop production and water resource allocation. However, the parametrization of crop growth processes, for which relevant genetic coefficients must be determined through field scale experiments, is usually hampered by poor spatial representation; this limitation compromises the robustness of irrigation scheduling at the regional scale. We propose a framework for optimizing regional irrigation schedules; this framework involves using a distributed crop model with crop parameters determined through data assimilation. We first conducted a sensitivity analysis to identify the key genetic coefficients (P2: photoperiod sensitivity coefficient; P5: thermal time from silking to physiological maturity; G3: kernel filling rate during the linear grain filling stage and under optimal conditions) affecting the leaf area index (LAI) in the CERES-Maize model. Subsequently, the field-scale CERES-Maize model was calibrated and validated using data collected from field experiments conducted in 2016 and 2017, respectively. Subsequently, the LAI estimated from 2020 microwave backscattering data was used to assimilate remote sensing information and crop model in order to obtain the key genetic coefficients of maize in the distributed crop model. Finally, on the basis of the distributed crop model, a multiobjective genetic algorithm was executed to optimize irrigation schedules under various meteorological scenarios (e.g., precipitation and reference evapotranspiration). The LAI of maize at a regional scale was accurately estimated using backscattering information extracted from microwave remote sensing images (R2 = 0.94, RMSE = 0.27). Compared with the crop model parameters calibrated at the field scale, the crop model parameters determined through data assimilation resulted in more accurate simulations of LAI, soil water content, and yield at the regional scale. The irrigation schedule based on the distributed model increased the maize yield corresponding to the local current irrigation schedule by 125–851 kg/ha, increased the water use efficiency by 0.02–0.17 kg/m3, and reduced the irrigation amount by 2–60 mm. The proposed framework for estimating genetic coefficients based on data assimilation methods helps in regional irrigation scheduling for other crops.