Fertilization of agricultural soils is the largest source of anthropogenic nitrous oxide (N2O) emissions. Despite decades of study using chamber-based methods to estimate N2O emission by soils, a discrepancy remains between upscaled chamber estimates and regional observations. We hypothesize that extreme temporal and spatial variability of N2O emissions results in underestimation of direct soil N2O emissions due to insufficient measurement frequency and/or measurement locations that do not capture the full range of environmental conditions in a study site or region. To account for temporal and spatial variability, we constructed a new automated chamber system to enable sub-daily measurements (4-hr intervals). We installed 8–16 chambers in Central Iowa across a topographic gradient from a poorly drained upland to a very poorly drained depression where we expected higher N2O emissions from greater soil moisture and denitrification. The soil conditions were typical for the region, where poor drainage frequently limits agricultural productivity. Agricultural management was a conventional corn-soybean rotation with typical fertilization rates during corn years (~168 kg N ha-1 y-1). We collected three years of soil N2O and carbon dioxide (CO2) flux data with accompanying edaphic conditions (soil temperature, moisture, inorganic nitrogen pools, and soil properties).
Results/Conclusions
We found average cumulative N2O emissions in 2017–2019 of 16.2, 3.5, and 7.1 kg N-N2O ha-1 yr-1 respectively. In the two years with fertilizer application, cumulative N2O-N emissions were 9% and 4% by mass compared to the total fertilizer applied. Over the three-year period, we found N2O emissions over twice what was expected under the 2019 International Panel on Climate Change Tier 1 N2O emission factor calculations, which include both fertilizer application and nitrogen (N) recycled in crop biomass. A random forest machine learning analysis suggested that soil CO2 emission rate, volumetric water content, and nitrate (NO3-) concentration were the best predictors of N2O emission. Despite generally higher soil moisture, lower topographies did not exhibit higher N2O emissions. We found low emissions during periods of flooding in the lowest positions, which may offset higher emissions at other times. Using high-frequency auto-chamber measurements from a typical agricultural field site in the Corn Belt, we estimate N2O emissions in reasonable agreement with top down methods.