Spatio-temporal variability of N2O

Spatio-temporal variability of N2O: Spatial, temporal and methodological biases

Soil flux of nitrous oxide (N2O) is unpredictable to measure under natural circumstances and becomes even more difficult when looking at agricultural sites. One reason for this is the temporal uncertainty associated with agricultural soils where fertilizer application, animal voiding and intense rainfall can increase fluxes by several orders of magnitude, and can persist for hours to weeks. eosAC soil gas flux chamberSpatial variability also adds to the unpredictability, where (N2O) fluxes varying by several orders of magnitude in areas smaller than 10 m2.

Eddy covariance measurements capture large scale (100 m2) fluxes, but wash out the extent of spatial variability. However, important fluxes can be missed if they occur on sporadic time scales, and gap filling procedures are limited at best, considering N2O emission peaks can occur over time scales shorter than 24 hour periods. On the other hand, flux chambers can cover spatial variability, but the ability of these methods to capture the temporal variability needed is limited without investing in automated systems. In contrast to eddy covariance, any gap-filling attempts between flux measurements is limited by the unpredictable nature of the spatial variability of (N2O) fluxes.

In order to understand these processes and methods of measurement better, Nick Cowan and colleagues at the Centre for Ecology and Hydrology (CEH) created a model simulation using field scale fluxes using eddy covariance data obtained from an agricultural field during two fertilization events. Using the log-normal spatial distribution pattern typically seen in (N2O) flux chamber measurements, they were able to simulate 10,000 potential chamber distribution measurements.

The Monte-Carlo method allowed the authors to sample from the simulation continuously in order to assess the uncertainty associated with various chamber measurement set ups (time between measurements, number of chamber measurements).

Among the many interesting findings, Cowan et al. (2016)’s simulation results suggest that researchers should be using a minimum of 5 chambers taking measurements every day, or 25 chambers taking measurements every three days, in order to keep error in annual cumulative flux measurements low (<10%).

Systematic bias due to fitting is also an important consideration according to Cowan et al. (2016). Previous studies have suggested that commonly used chamber fitting methods can underestimate fluxes by up to 20%. Assuming this is true, then the cumulative chamber flux measurements have the potential to underestimate (N2O) fluxes by as much as 60% due to the typically log-normal nature of (N2O) flux measurements.


Observed nitrous oxide flux measurements from eddy covariance towers used in Cowan et al. (2016)’s model.

These findings bring to light some interesting insights into common-practice methods in (N2O) flux measurement methodology. By understanding the uncertainties associated with current, accepted flux methodologies can help us gain a better understanding of how to improve them, and reduce associated temporal-spatial uncertainty in (N2O) measurements.

Cowan et al. (2016)’s findings support and augment the conclusions made in the Biases of Temporally Sparse Data and Measurement Scheduling on Flux Estimates white paper and Creelman et al. 2013 paper reviewing biases in commonly used chamber methods. This research, and the research of others, is beginning to demonstrate the need for spatially and temporally dense chamber measurements collected by automated chamber systems in order to properly assess spatial and temporal variability in N2O fluxes.

More from CEH at EGU.


Cowan, N., Levy, P., and Skiba, U., The effect of random and systematic measurement uncertainties on temporal and spatial upscaling of N2O fluxes. EGU General Assembly, Vienna, Austria (2016).

Creelman, C., Nickerson, N., and Risk, D. (2013). Quantifying lateral diffusion error in soil carbon dioxide respiration and estimates using numerical modeling. Soil Sci. Soc. Am. J. 77(3): 699-708.

Related Posts Plugin for WordPress, Blogger...