Assessing nitrous oxide emissions in time and space with minimal uncertainty using static chambers and eddy covariance from a temperate grassland

Abstract Where nitrogen input from fertilizer application exceeds plant demands, hotspots of microbially produced nitrous oxide (N 2 O) can exhibit disproportionately high rates of emissions relative to longer periods of time, known as hot moments. Hotspots and hot moments of N2O are sensitive to changes in agricultural management and weather, making it difficult to accurately quantify N 2 O emissions. This study investigates the spatial and temporal variability of N2O emissions using both static chambers (CH) and eddy covariance (EC) techniques, measured at a grassland site subject to four fertilizer applications of calcium ammonium nitrate (CAN) in 2019. Daily mean CH emissions were calculated using the arithmetic method and Bayesian statistics to explicitly account for the log-normal distribution of the dataset. N2O fluxes measured by CH and EC were most comparable when flux measurements were > 115 N 2 O N µg m − 2 hr −1, and EC and CH measurements showed spatial and temporal alignment when CH n ≥ 15. Where n ≤ 5, the Bayesian method produced large uncertainties due to the difficulty of fitting an arithmetic mean from a log-normally distributed data set with few flux measurements. Annual EC fluxes, gap-filled using a multi-variate linear model, showed a strong correlation with measured flux values (R 2 = 0.92). Annual cumulative fluxes by EC were higher (3.35 [± 0.5] kg N ha−1) than CH using the arithmetic (2.98 [± 0.17] kg N ha−1 ) and Bayesian method (3.13 [± 0.24] kg N ha−1), which quantified emission factors of 1.46%, 1.30% and 1.36%, respectively. This study implies that a large sample size and frequent CH flux measurements are necessary for comparison with EC fluxes and that Bayesian statistics are an appropriate method for estimating realistic means and ranges of uncertainty for CH flux data sets.