Simulating CH4 and N2O emissions from direct-seeded rice systems using the DeNitrification DeComposition (DNDC) model

Abstract:

Process-based modeling of CH₄ and N₂O emissions from rice fields is a practical tool for conducting greenhouse gas inventories and estimating mitigation potentials of alternative practices at the scales of management and policy-making. However, few studies have evaluated site-level model performance in side-by-side field trials of various management practices during both the growing season and fallow periods. We empirically evaluated the DeNitrification-DeComposition (DNDC) model for estimating CH₄ and N₂O fluxes in California rice systems under varying management (N fertilizer application rate, type of seeding system, fallow period straw and water management), soil environments, and weather conditions. Five and nine site-year combinations were used for calibration and validation, respectively. The model was parameterized for two cultivars, M206 and Koshihikari, and able to simulate 30% and 78% of the measured variation in yields, respectively. A major strength of DNDC was in estimating general site-level seasonal CH₄ emissions (R² = 0.85). However, a major limitation was in simulating finer resolution of differences in CH₄ emissions (or lack thereof) among side-by-side management treatments (range of 0.2-465% relative absolute deviation). Additionally, DNDC did not satisfactorily simulate fallow period CH₄ emissions, or seasonal and fallow period N₂O emissions across all sites with the exception of a few cases. Specifically, simulated CH₄ emissions were oversensitive to fertilizer N rates, but lacked sensitivity to the type of seeding system and prior fallow period straw management. Additionally, N₂O emissions were oversensitive to fertilizer N rates and field drainage. Sensitivity analysis showed that CH₄ emissions were highly sensitive to changes in the root to total plant biomass ratio. Overall, uncertainty in model predictions was attributed to uncertainty in both the input parameters due to in-field spatiotemporal variability of soil properties, and in the model structure (e.g., genotype by environment interactions, clay effects, and simulation routines for field drainage, and diffusion and ebullition of gasses). These findings have implications for model-directed field research that could improve model uncertainty for application at larger spatial scales.