Estimating fractions of N2O emissions from nitrification and denitrification using data assimilation

Pan, B., Huang, Y., Xia, L., Liang, J., Liu, R., Luo, Y., Du, Z., Chen, D., & Lam, S. K., Estimating fractions of N2O emissions from nitrification and denitrification using data assimilation. Biogeochemistry, 168 (5). DOI: https://doi.org/10.1007/s10533-025-01268-x

Nitrous oxide (N₂O) emissions play a significant role in global warming and stratospheric ozone depletion. Nitrification and denitrification represent the primary pathways of N₂O emissions in agroecosystems. However, modelling the responses of nitrification, denitrification, and subsequent N₂O emissions to soil conditions and nitrification inhibitors remains challenging, as the fractions of N₂O emissions derived from nitrification and denitrification used in model simulations cannot be directly measured. In this study, we estimated soil nitrification, denitrification, N₂O emissions, and their related parameters via data assimilation under various soil moisture levels [water-filled pore space (WFPS) at 50% and 70%], incubation temperature (15, 25 and 35 °C) and nitrification inhibitor application (DMPP, 3MPTZ and C₂H₂) in cereal and vegetable production systems in Australia. We found that the contribution of nitrification to N₂O emissions (i.e., the fraction of N₂O emitted from nitrification, ) decreased with increasing temperature and moisture content, whereas denitrification dominated N₂O production (i.e., the fraction of N₂O emitted from denitrification, ) under 70% WFPS regardless of temperatures. Under fertilizer N application, the use of nitrification inhibitors decreased  but increased . The efficacy of nitrification inhibitors in mitigating N₂O emissions varied with environmental conditions. In this study, we demonstrate the use of data assimilation to constrain key parameters for predicting nitrification, denitrification and associated N₂O emissions in response to soil environments and management practices. Integrating this technique into ecosystem process-based models has the potential to enhance model accuracy by reducing uncertainties and biases.