The uncertainty of nitrous oxide emissions from grazed grasslands: A New Zealand case study

Abstract Agricultural soils emit nitrous oxide (N 2 O), a greenhouse gas and the primary source of nitrogen oxides which deplete stratospheric ozone. Agriculture has been estimated to be the largest anthropogenic N 2 O source. In New Zealand (NZ), pastoral agriculture uses half the land area. To estimate the annual N 2 O emissions from NZ's agricultural soils, the nitrogen (N) inputs have been determined and multiplied by an emission factor (EF), the mass fraction of N inputs emitted as N 2 O N. To estimate the associated uncertainty, we developed an analytical method. For comparison, another estimate was determined by Monte Carlo numerical simulation. For both methods, expert judgement was used to estimate the N input uncertainty. The EF uncertainty was estimated by meta-analysis of the results from 185 NZ field trials. For the analytical method, assuming a normal distribution and independence of the terms used to calculate the emissions (correlation = 0), the estimated 95% confidence limit was ±57%. When there was a normal distribution and an estimated correlation of 0.4 between N input and EF, the latter inferred from experimental data involving six NZ soils, the analytical method estimated a 95% confidence limit of ±61%. The EF data from 185 NZ field trials had a logarithmic normal distribution. For the Monte Carlo method, assuming a logarithmic normal distribution for EF, a normal distribution for the other terms and independence of all terms, the estimated 95% confidence limits were −32% and +88% or ±60% on average. When there were the same distribution assumptions and a correlation of 0.4 between N input and EF, the Monte Carlo method estimated 95% confidence limits were −34% and +94% or ±64% on average. For the analytical and Monte Carlo methods, EF uncertainty accounted for 95% and 83% of the emissions uncertainty when the correlation between N input and EF was 0 and 0.4, respectively. As the first uncertainty analysis of an agricultural soils N 2 O emissions inventory using “country-specific” field trials to estimate EF uncertainty, this can be a potentially informative case study for the international scientific community.