Uncertainties in nitrogen application from manure and synthetic fertilizers and ammonia emissions from agriculture in the Netherlands across different spatial scales

Abstract Quantifying uncertainty in simulated nitrogen losses from agriculture is crucial for emission inventories. In the Netherlands, the INITIATOR model is used to quantify the spatially explicit distribution of livestock manure and synthetic fertilizers, and related NH 3 emissions. Activity data such as livestock numbers, nutrient excretion rates, grazing hours, synthetic fertilizer use, manure transport statistics and emission factors serve as model inputs, each with their own uncertainty. This study uses Monte Carlo simulation to assess how uncertainties in INITIATOR inputs affect the simulated N application rates from manure and synthetic fertilizers, and NH 3 emissions across spatial scales (500 m × 500 m grid, agricultural regions, provincial, national scale). Results at 500 m × 500 m show that the average upper and lower limits of the 95% confidence interval deviated from the median simulation results for livestock manure inputs by 21% (165–245 kg N ha⁻ 1 yr⁻ 1 ) for synthetic fertilizer inputs by 32% (89–168 kg N ha⁻ 1 yr⁻ 1 ), and for total NH 3 emissions from agriculture by 32% (36.0–68.3 kg N ha −1 yr −1 ). Uncertainty decreased strongly when upscaling the results from 500 m × 500 m to the national scale: for manure from 21 to 5%, for synthetic fertilizers from 32 to 26%, and for NH 3 emissions from agriculture from 32 to 9%. Reducing uncertainties in model inputs on excretion rates, synthetic fertilizer use statistics, and farmers acceptance factors for utilizing manure would effectively improve the simulated N manure and synthetic fertilizer rates. Uncertainties in emission factors related to housing systems and manure storage, field application, and grazing should be reduced to improve NH 3 emission estimates.