Optimization of N₂O Emission Parameters in Dryland Spring Wheat Farmland Soil Based on Whale Optimization Algorithm

In order to improve the simulation accuracy of N₂O emissions by using APSIM model, this study used Whale Optimization Algorithm (WOA) to optimize the default parameters related to soil N₂O emissions in the APSIM model to improve the accuracy and applicability of the model in simulating soil N₂O emissions in the semi-arid agricultural region of northwest China, for providing support for precise assessment and management of greenhouse gas emissions in agricultural activities.

This study used field experimental data measured by the Anjiapo integrated long-term positioning test station in Anding District, Dingxi City, Gansu Province from 2020 to 2021, combined with meteorological data provided by the Meteorological Bureau from 1970 to 2021, to optimize the four key parameters of N₂O formation stage in the APSIM model (soil nitrification potential (nitration_pot), concentration of ammonium nitrogen at semi maximum utilization efficiency (nh4_at-half_pot), denitrification coefficient (dnit_rate-coeff), and power term of denitrification water coefficient (dnit_wf_power) using the WOA for single objective and multi parameter optimization. The accuracy of the optimized APSIM soil N₂O emission model was evaluated by comparing the errors between the default parameter simulation values, optimized parameter simulation values, and measured values of the APSIM model.

Through multiple executions of the optimization program, the optimal combination of four parameters was ultimately determined. Among them, the soil nitrification potential was 7.62 mg·kg^-1·d^-1, the concentration of ammonia nitrogen at semi maximum utilization efficiency was 49.3 mg·kg^-1, the denitrification coefficient was 0.00063, and the power term of the denitrification water coefficient calculation was 0.64. Compared with the default parameters of the APSIM model, the coefficient of determination R² increased from 0.432 to 0.719, the root mean square error (RMSE) decreased from 39.42 to 25.37 μg·m^-2·h^-1, and the normalized root mean square error (NRMSE) decreased from 18.51% to 11.92%. The whale algorithm exhibited significant global search capability and fast convergence during the optimization process. The optimized APSIM model significantly improved the accuracy of simulating soil N₂O emissions, indicating that this method could achieve rapid and accurate calibration of model parameters.

By applying WOA, four key parameters were precisely adjusted, which significantly reduced the prediction error of the model and significantly improving the performance of the APSIM soil N₂O emission model. The optimized model has shown higher accuracy and applicability in the semi-arid agricultural region of northwest China, which also proved the effectiveness of the optimization strategy.