Reducing N2O emissions while maintaining yield in a wheat–maize rotation system modelled by APSIM

Abstract CONTEXT Modelling approaches have already been used to quantitatively assess the trade-offs between crop yield and N2O emissions as impacted by management practices. However, the model's performance in terms of predicting N2O emissions was mainly assessed against total emissions per growing season or year, which may reduce accuracy in modelling due to the uncertainties in total emissions estimated using the manual (static) chamber method. OBJECTIVE Here, a comparison between optimizations for Agricultural Production Systems sIMulator (APSIM) with total N2O emissions and daily N2O emissions was conducted. We further used the validated model to develop simple surrogate models for estimating total N2O emissions in different years and target potential opportunities to reduce N2O emissions while still maintaining the grain yield under long-term climatic conditions. METHODS Five parameters relating to denitrification and nitrification were optimized using differential evolution algorithm for global optimization based on two-year field experimental data at Huantai site in North China Plain, and comprehensive simulation experiments were further conducted under long-term climate variability in an irrigated wheat–maize rotation system. The method of Levenberg-Marquardt was implemented to fit simple surrogate models for estimating total N2O emissions in different years, and Analysis of Variance was used for model comparison. RESULTS AND CONCLUSIONS APSIM model optimized with daily N2O emissions could better simulate soil N2O and nitrate dynamics than that optimized with total N2O emissions. We obtained the posterior distributions of five key parameters to which N2O emissions are sensitive, and demonstrated that original model using default parameters could underestimate the rate of nitrification and denitrification and the subsequent N2O emissions. Total N2O emissions increased exponentially with nitrogen application rate and mean temperature, and IPCC (1% emission factor) could underestimate whole-year N2O emissions when N rate was higher than the optimized nitrogen rate for crop production. We also found that there was potential to optimize nitrogen fertiliser rate to reduce N2O emissions while still maintaining crop yield in the irrigated wheat–maize rotation system. SIGNIFICANCE This study demonstrated the necessity of optimization with daily N2O emissions in improving model accuracy, and the posterior distributions of five parameters relating to N2O emissions offered reference range for future model improvement and applications.