A soil-specific model to predict N₂O emissions from laboratory and field experiments

Abstract

Numerous models have been developed to simulate nitrous oxide (N₂O) emissions from agricultural soils, yet accurately capturing the spatial and temporal variability of soil N₂O fluxes remains a challenge. To better estimate soil N₂O emissions, we developed a statistical model based on the Gaussian function, with parameters that vary according to edaphic properties. A global database of N₂O emissions from agricultural soils, derived from laboratory incubation experiments, was established to parameterize, calibrate and validate the developed model. Simulations demonstrated that incorporating multiple edaphic properties, including soil moisture, mineral nitrogen contents, carbon to nitrogen ratio, silt content, bulk density and soil depth, enabled reliable prediction of N₂O emissions from sieved soils. However, the initially parameterized model significantly overestimated emissions from intact soils. To address this, soil structure correction factors, quantified by bulk soil properties, were introduced into the model. Incorporating these structure corrections enabled the model to successfully predict N₂O emissions from intact soils, highlighting the importance of accounting for soil structure in models. The improved model was then employed to simulate N₂O emissions from different field sites with contrasting agricultural treatments, after further taking into account temperature effects. It effectively captured the temporal dynamics of N₂O fluxes, including the timing and magnitude of N₂O emission peaks, particularly under optimal N additions and long-term tillage. Overall, this soil-specific model provides a robust tool to predict the large spatiotemporal variations in N₂O fluxes across different soils under various environmental settings, which is critical for reducing uncertainty in large-scale estimates.