Optimizing nitrogen topdressing for winter wheat by coupling remote sensing data with the DSSAT model.

Introduction: Excessive fertilization not only causes environmental pollution and degrades water and soil quality but also increases production costs and reduces agricultural sustainability. Methods: Based on two consecutive years of field experiments, this study developed a two-step data assimilation strategy for nitrogen (N) topdressing recommendations for winter wheat. First, a data assimilation system was established by minimising the discrepancy between aboveground dry biomass (AGB) estimated from remote sensing and that simulated by the crop growth model using a particle swarm optimization approach. Second, target yields under varying growth conditions were constructed using the DSSAT model and N economic return curves to enable optimised N fertilization recommendations. Results: AGB monitoring model was developed, achieving satisfactory results in both the calibration and validation datasets, with determination coefficient (R²) (normalised root mean square error (nRMSE)) values of 0.94 (13.62%) and 0.82 (15.42%), respectively. Based on the data assimilation system, the data assimilation stability for AGB and yield are relatively high. The nRMSE values for AGB are 11.20% and 19.44% for the training and validation datasets, respectively. The nRMSE values for yield are 6.35% and 11.22% for the training and validation datasets, respectively. The data assimilation-based recommended fertilization shows a negative power-law relationship with AGB at the jointing stage (R² = 0.65). Under different yield levels, fertilization was reduced by 6.69%-34.08% compared with that under high yield levels. Conclusion: This study balances yield and production costs by developing a data assimilation strategy for N fertilization recommendations, which can maintain high productivity and sustainability.