Evaluation of DeNitrification DeComposition model for estimating ammonia fluxes from chemical fertilizer application

DeNitrification DeComposition (DNDC) model predictions of NH3 fluxes following chemical fertilizer application were evaluated by comparison to relaxed eddy accumulation (REA) measurements, in Central Illinois, United States, over the 2014 growing season of corn. Practical issues for evaluating closure were addressed by accounting for fluxes outside the measurement site and differences in temporal resolution. DNDC modeled NH3 fluxes showed no significant differences in magnitude (at p = 0.05) compared to measurements and replicated trends satisfactorily (r(a)(2) > 0.74), during the initial 33 days after fertilizer application, when measured fluxes were to the atmosphere, compared to later time periods when depositional fluxes were measured (r(a)(2) < 0.52). Among the model input parameters, NH3 fluxes were most sensitive to air temperature, precipitation, soil organic carbon, field capacity, pH, and fertilizer application rate, timing, and depth. By constraining these inputs for conditions in Central Illinois, uncertainty in daily NH3 fluxes was estimated to vary from 0% to 70% on a daily basis, during the corn growing season, with the highest uncertainty values estimated for the period of highest positive NH3 fluxes. These results can guide future improvements in DNDC, which is a valuable tool to assist (1) in the development of NH3 emission inventories with high spatial (constrained by the spatial resolution of input parameters) and temporal resolution (daily) and (2) in upscaling emissions from the site (farm) to the regional scale. (C) 2017 Elsevier B.V. All rights reserved.