Urea concentrations in urine patches deposited during animal grazing can be over ten times higher than typical fertiliser application rates, potentially leading to large ammonia (NH3) losses. The process-based NZ-DNDC model was modified to better simulate soil pH changes and ammonia (NH3) emissions following urine application using data collected from a New Zealand field trial. After modification, simulated 30-day NH3 emissions decreased from 506 to 117 kg N ha−1 compared to measured emissions of 78 ± 3 kg N ha−1 (mean ± standard error) and the Nash–Sutcliffe Efficiency (NSE) for daily NH3 emissions increased from −7.11 to +0.97 for the parameterisation dataset. However, modified model correctly estimated the cumulative emissions for the first 7 days. Using the same parameterisation on an independent dataset from a nearby site gave cumulative 18-day NH3 emissions of 84 kg N ha−1 compared to the measured 48 ± 2 kg N ha−1 (mean ± standard error). However, the NSE for daily NH3 emissions was −0.71, indicating site specific parameterisation might be needed. The sensitivity of NH3 emissions to ±5 and ±10% errors in 4 model parameters was tested. The sensitivities ranged from −0.36 to +0.71. The highest sensitivity was to the rate of NH3 transfer from the soil solution to the atmosphere and the lowest sensitivity was to the rate of urea hydrolysis.