Evaluation of Parametric Limitations in Simulating Greenhouse Gas Fluxes from Irish Arable Soils Using Three Process-Based Models

Abstract

Globally a large number of process-based models have been assessed for quantification of agricultural greenhouse gas (GHG) emissions. Modelling approaches minimize the presence of spatial variability of biogeochemical processes, leading to improved estimates of GHGs as well as identifying mitigation and policy options. The comparative performance of the three dynamic models (e.g., DNDC v9.4, DailyDayCent and ECOSSE v5+) with minimum numbers of common input parameters was evaluated against measured variables. Simulations were performed on conventionally-tilled spring barley crops receiving N fertilizer at 135 - 159 kg·N·ha-1·yr-1 and crop residues at 3 t·ha-1·yr-1. For surface soil nitrate (0 - 10 cm), the ECOSSE and DNDC simulated values showed significant correlations with measured values (R2 = 0.31 - 0.55, p 0.05). Only the ECOSSE-simulated N2O fluxes showed a significant relationship (R2 = 0.33, p 0.05) with values measured from fertilized fields, but not with unfertilized ones. The DNDC and DailyDayCent models significantly underestimated seasonal/annual N2O fluxes compared to ECOSSE, with emission factors (EFs), based on an 8-year average, were 0.09%, 0.31% and 0.52%, respectively. Predictions of ecosystem respiration by both DailyDayCent and DNDC showed reasonable agreement with Eddy Covariance data (R2 = 0.34 - 0.41, p 0.05). Compared to the measured value (3624 kg·C·ha-1·yr-1), the ECOSSE underestimated annual heterotrophic respiration by 7% but this was smaller than the DNDC (50%) and DailyDayCent (24%) estimates. All models simulated CH4 uptake we