Abstract
Estimates of global riverine nitrous oxide (N 2 O) emissions contain great uncertainty. We conducted a meta‐analysis incorporating 169 observations from published literature to estimate global riverine N 2 O emission rates and emission factors. Riverine N 2 O flux was significantly correlated with NH 4 , NO 3 and DIN ( NH 4 + NO 3 ) concentrations, loads and yields. The emission factors EF (a) (i.e., the ratio of N 2 O emission rate and DIN load) and EF (b) (i.e., the ratio of N 2 O and DIN concentrations) values were comparable and showed negative correlations with nitrogen concentration, load and yield and water discharge, but positive correlations with the dissolved organic carbon : DIN ratio. After individually evaluating 82 potential regression models based on EF (a) or EF (b) for global, temperate zone and subtropical zone datasets, a power function of DIN yield multiplied by watershed area was determined to provide the best fit between modeled and observed riverine N 2 O emission rates ( EF (a): R 2 = 0.92 for both global and climatic zone models, n = 70; EF (b): R 2 = 0.91 for global model and R 2 = 0.90 for climatic zone models, n = 70). Using recent estimates of DIN loads for 6400 rivers, models estimated global riverine N 2 O emission rates of 29.6–35.3 (mean = 32.2) Gg N 2 O–N yr −1 and emission factors of 0.16–0.19% (mean = 0.17%). Global riverine N 2 O emission rates are forecasted to increase by 35%, 25%, 18% and 3% in 2050 compared to the 2000s under the Millennium Ecosystem Assessment's Global Orchestration , Order from Strength , Technogarden , and Adapting Mosaic scenarios, respectively. Previous studies may overestimate global riverine N 2 O emission rates (300–2100 Gg N 2 O–N yr −1 ) because they ignore declining emission factor values with increasing nitrogen levels and channel size, as well as neglect differences in emission factors corresponding to different nitrogen forms. Riverine N 2 O emission estimates will be further enhanced through refining emission factor estimates, extending measurements longitudinally along entire river networks and improving estimates of global riverine nitrogen loads.