Ammonia volatilization from broadcast urea: measurements using a micrometeorological approach and modeling with the Denitrification-Decomposition (DNDC) model

Abstract

Synthetic N fertilizers, such as urea, are one of the main anthropogenic sources of atmospheric ammonia (NH₃). NH₃ volatilization from N fertilizer application can significantly reduce agronomic efficiency (AE), contribute to air pollution, soil acidification and eutrophication of water bodies. Therefore, understanding the processes and factors influencing NH₃ volatilization from broadcast urea is pivotal to improve agricultural sustainability. In Chapter 2 of this thesis, the integrated horizontal flux (IHF) approach was used to measure NH₃ volatilization from eight field experiments under cold weather conditions in Kansas. NH₃ volatilization was measured from circular plots (20-m radius) fertilized with urea and urea amended with urease inhibitor (NBPT) both at rate of 60 kg of N ha⁻¹. The impact of NH₃ volatilization losses on winter wheat was evaluated through experimental plots arranged in a complete randomized block design with treatments consisting of four different rates of application (30, 60, 90 and 120 kg N ha⁻¹) of urea and urea + NBPT and control (0 kg N ha⁻¹) . NH₃ cumulative losses varied from <1% to 29% of applied N. Largest losses occurred when urea was broadcast to soils with high water content followed by a dry period. The use of urease inhibitor NBPT reduced NH₃ volatilization losses in more than 20% at locations where the largest losses (> 25%) occurred. No statistical difference was found in terms of grain yield, N recovery and AE, when comparing urea and urea + NBPT treatments. In Chapter 3, simulations provided by two versions of the Denitrification-Decomposition (DNDC) process-based model (DNDC 9.5 and DNDC v.CAN) were compared with flux data obtained in 29 NH₃ volatilization sampling campaigns. These sampling campaigns were conducted in Kansas and Montana using the IHF method over circular plots. Overall, the DNDC v.CAN simulated NH₃ emissions with smaller average root mean square error ((RMSE) ̅ = 10.9 kg of N ha⁻¹) compared to the DNDC 9.5 ((RMSE) ̅ = 32.8 kg of N ha⁻¹). Our sensitivity analysis showed that soil pH and soil temperature were the main input variables affecting NH₃ volatilization in both models. In addition, our analysis demonstrated several drawbacks that could be improved in future versions of the model to better simulate NH₃ volatilization. These potential areas for improvements the DNDC model versions include: i) limitations in the soil-hydrology water sub-model affected the accuracy of the simulations of the effects of soil water content on urea hydrolysis, which has direct effect on NH₃ volatilization; ii) both models failed to simulate the effects of accumulated precipitation (≥ 20 mm) on NH₃ volatilization during the first 5-15 d post fertilization; iii) future developments of the DNDC should consider adding a more robust routine to simulate the effects of urease inhibitor on NH₃ volatilization and iv) the timing of the NH₃ volatilization peak after fertilization was underestimated by the DNDC v.CAN and largely overestimated by the DNDC 9.5.