Simulating the spatial variability of nitrous oxide emission from cropped soils at the within-field scale using the NOE model

Estimating total N2O emission from agricultural soils is associated with considerable uncertainty due to the very large spatial variability of the fluxes. Thus characterizing the range of variations is of great interest. Modeling N2O fluxes remains challenging, especially at the within-field scale. The aim of this study was to test the ability of a simple process-based model, NOE (Nitrous Oxide Emission), to simulate N2O at scales finer than the field. Six field studies including 30–49 measurements of chamber N2O fluxes and ancillary variables were conducted in a barley/wheat field on hydromorphous soils. Three studies were made on surfaces of ∼10 m2 (defined as the local scale), and three studies along a 150-m transect (defined as the transect scale). First, the model was tested deterministically for predicting the flux spatial patterns, i.e., to try to reproduce the high flux points. Then the denitrification part of the model was tested stochastically for simulating the flux distributions by randomly generating input variables from the measured frequency distributions (Monte Carlo simulation). Measured fluxes were comprised between 0 and 1.5 mg N h−1 m−2. The deterministic prediction of spatial patterns provided a good match with measurements in 1 of the 6 studied cases, in a transect study. Denitrification was assessed to be the main source of N2O in 5 of the 6 cases and the model satisfactorily simulated frequency distributions in 4 cases out of 5, 2 at the local scale and 2 at the transect scale. Thus this study suggests that simple process-based models such as NOE, combined to Monte Carlo methods, can be used to improve simulation of the skewed frequency distributions of N2O fluxes and provide valuable information about the range of spatial variations in N2O fluxes.