Improving DNDC model to estimate ammonia loss from urea fertilizer application in temperate agroecosystems

Process-based biogeochemical models such as the DeNitrification-DeComposition (DNDC) model can provide reliable estimations of agricultural nitrogen loss, information necessary for developing better management practices. The Canadian version of the model (DNDC v.CAN) was recently developed to predict NH3 volatilization after the field application of liquid animal manure; however it did not include a mechanism for simulating soil pH buffering or fertilizer application at depth. Using published data collected from four studies in Qu,bec, Canada, this research included the effects of urea hydrolysis and buffer capacity on soil pH in DNDC v.CAN. The improved model was tested for estimating NH3 loss after surface and incorporated urea applications. Owing to the developments, the NH3 loss estimations of the improved DNDC v.CAN model better correlated to observations in all four field studies (r = 0.95 for P aecurrency sign 0.05) compared to the previous model versions (0.63 and 0.47 respectively). Soil pH predictions were also better correlated (r = 0.56) while the other versions had poor correlations of -0.27 and -0.33, respectively. An assessment of environmental and management factors on NH3 losses using scenario and sensitivity analyses was also carried out. The scenario assessment suggested that incorporated urea application had the highest reduction in cumulative NH3 loss and the sensitivity analysis indicated that soil pH and temperature had the greatest influence on estimates of NH3 loss under the incorporated and split scenarios. The improvements to the mechanisms regulating N losses in DNDC v.CAN will have positive implications to accurately simulate N cycling in agro-ecosystems.