Process-based models play an important role in the estimation of soil N2O emissions from regions with contrasting soil and climatic conditions. A study was performed to evaluate the ability of two process-based models, DAYCENT and DNDC, to estimate N2O emissions, soil nitrate- and ammonium-N levels, as well as soil temperature and water content. The measurement sites included a maize crop fertilized with pig slurry (Quebec) and a wheat-maize-soybean rotation as part of a tillage-fertilizer experiment (Ontario). At the Quebec site, both models accurately simulated soil temperature with an average relative error (ARE) ranging from 0 to 2%. The models underpredicted soil temperature at the Ontario site with ARE from -5 to -7% for DNDC and from -5 to -13% for DAYCENT. Both models underestimated soil water content particularly during the growing season. The DNDC model accurately predicted average seasonal N2O emissions across treatments at both sites whereas the DAYCENT model underpredicted N2O emissions by 32 to 58% for all treatments excluding the fertilizer treatment at the Quebec site. Both models had difficulty in simulating the timing of individual emission events. The hydrology and nitrogen transformation routines need to be improved in both models before further enhancements are made to the trace gas routines.
