Assessing the sensitivity of fertilizer types and soil variables on nitrous oxide emissions in permanent grasslands using the DNDC model

The adoption and use of improved methodologies including models that reflect more robust emissions accounting procedures and the identification of specific mitigation options for agricultural greenhouse gases are a global concern. In Ireland, country-specific N2O emission factors (EFs) are constrained primarily by short-term measurements and limited coverage of regulating factors. Simulation of N2O emissions from grassland silage plots managed for 42 years with different slurry treatments was performed using the DeNitrification-DeComposition (DNDC95) model. The objective was to assess the long-term impact of management practices on N2O fluxes and EFs, and the sensitivity of the outputs to key inorganic and organic fertilizer management and soil variables. The DNDC performed well for urea, cattle slurry and pig slurry applied at variable rates, delivering EFs on-average of 0.35±0.02, 1.80±0.28 and 1.53±0.41%, respectively. Variation in the derived-EFs could be explained by differences in nitrogen inputs (49%), rainfall (16%) and temperature (10%) and are close to national estimates. Sensitivity analysis of the model demonstrated that N2O EFs were higher with ammonium sulphate compared to CAN and urea fertilizers, and with urea-N at higher rates. The replacement of slurry either after the second or third silage cut by urea decreased EFs significantly. There was a strong correlation with the sensitivity of N2O EFs to soil texture, bulk density, pH and organic carbon (R2=0.96-0.99). The resulting-EFs ranged from 0.28 to 0.41% for urea, 1.12 to 2.07% for cattle slurry, and 1.05 to 1.65% for pig slurry, and the corresponding values on-average were 0.35±0.02, 1.74±0.17 and 1.39±0.12%. These findings show that DNDC95, although requiring more improvement, could provide an accurate representation of the effect of soils, climate and management practices on N2O fluxes and subsequent estimates of disaggregated EFs.