There are broadly two methods for measuring N2O emissions: flux chamber and micrometeorological techniques. In chamber techniques, N2O emission rates are commonly determined by enclosing the atmosphere above the source (for example soil, manure or water body) and measuring the increase in headspace N2O concentration over time. These flux chamber techniques have been widely used to measure N2O emissions from soil because they are relatively cheap and simple, are useful for relative comparisons between adjacent treatments and they allow process-based studies of N2O emission from soils. Their main disadvantage is that they measure N2O emission over a relatively small area and that a large number of measurements and chambers are required to deal with the large spatial and temporal variability in emissions. Recent advances in flux chamber techniques include (1) continuous N2O analysis using gas chromatography (GC) (Kiese et al, 2003), tunable-diode laser, Fourier transform infrared (FTIR) (Kelly et al, 2008) or photo-acoustic infrared spectroscopy (Dinuccio et al, 2008) and (2) the use of chambers that open and close automatically (for example Breuer et al, 2000; Kiese et al, 2003; Kelly et al, 2008).
Micrometeorological techniques involve measurements of N2O in the atmosphere at two or more points above the soil surface, in combination with meteorological measurements (for example wind speed, wind direction and air temperature) (Denmead et al, 2000). These techniques measure N2O emissions on a field scale, and thus spatially integrate N2O flux measurements. Their disadvantages are, however, that they require large homogeneous field sites, are less reliable with low wind speed and high atmospheric stability, and require expensive N2O analysis equipment (Fowler et al, 1997; Phillips et al, 2007). Phillips et al (2007) measured N2O fluxes from a flood-irrigated dairy pasture, using a micrometeorological method coupled to a tunable-diode laser, in order to detect rapid changes in N2O fluxes in response to irrigation, grazing and N fertilizer application (Figure 6.3). N2O fluxes remained low following each irrigation event while WFPS was close to saturation (>95 per cent). However, two to three days after irrigation, as soil moisture decreased below ~95 per cent WFPS, N2O emissions increased rapidly and remained high for one to two days. As WFPS approached ~75 per cent, there was an initially rapid decrease in fluxes, followed by a gradual decrease to background levels at a WFPS of about 65 per cent. The magnitude of each N2O flux response to irrigation was greatly influenced by the N input, via either grazing or N fertilizer. This pattern was similar for each irrigation or significant rainfall
Note: Dotted and dashed vertical lines in (a) represent grazing and fertilization events at each site. Dotted horizontal lines in (b) indicate 95 per cent, 75 per cent and 65 per cent WFPS Source: Phillips et al (2007)
event. Micrometeorological methods are well suited to measuring real-time fluxes over a large area in response to management intervention. However, since enclosures and chamber techniques are by far the most commonly deployed method for measuring soil N2O fluxes, some key aspects for their deployment are discussed below.
Fluxes are calculated by determining the increase in headspace N2O concentration over a given period (typically about one hour), and hourly N2O emissions (mg N m~2 h-1) corrected for temperature and the ratio of cover volume to surface area (de Klein et al, 2003). These hourly emission rates are converted to daily fluxes for each enclosure, and then integrated over time using a trapezoidal calculation to estimate total emissions from the enclosed area over a sampling period.
An increase in the headspace N2O concentration may decrease the concentration gradient within the soil and atmosphere, resulting in a declining flux. This could lead to an underestimation of the calculated flux when relying solely on linear models (Conen and Smith, 2000; Davidson et al, 2002). Non-
linear models often result in less biased estimates of the rate of change in headspace N2O concentrations compared to linear models (Healy et al, 1996).
It is important to recognize that deploying enclosures onto the soil surface often modifies the flux that is to be measured (Rochette and Eriksen-Hamel, 2008). These workers assessed 356 studies employing enclosures by using 16 criteria to assess chamber design and technique. They conclude that confidence in absolute flux values was considered 'very low' or 'low' in about 60 per cent of the studies. To ensure N2O flux data is of high quality, Rochette and Eriksen-Hamel (2008) recommend the following six criteria when using enclosures:
1 Use an insulated and vented base-and-chamber design.
2 Avoid chamber heights lower than 10cm.
3 Have a minimum insertion depth of 5cm for the chamber base.
4 If samples are stored for analysis off-site, then use pressurized fixed-volume containers of known efficiency for air sample storage.
5 Include a minimum of three discrete air samples during deployment, including one at time zero.
6 Test non-linearity of changes in headspace concentration with time for estimating dC/dt at time zero.
Similarly, Meyer et al (2001) suggested the following six requirements for accuracy in designing chambers:
1 Ensure that there is sufficient air speed over the soil surface to overcome boundary layer effects resulting from the chamber physical design.
2 Ensure adequate turbulent mixing of the free volume within large (for example around 100-litre) chambers to avoid concentration gradients forming.
4 Ensure equality of ambient pressure within and outside the chamber, which is also achieved by venting a chamber as suggested by Rochette and Eriksen-Hamel (2008).
5 Minimize interference of the chamber with key soil and plant environmental variables.
6 Quantify the rate of flow of air into and out of the chamber when measuring headspace concentrations using continuous or near-continuous N2O analysis.
The high spatial and temporal variability of N2O emissions (Figure 6.3) hampers the estimation of N2O emissions from chamber-based flux measurements. Significant emission events may be underestimated, or indeed missed, when sampling at intervals of several days compared with frequent sampling at regular short time intervals (Smith and Dobbie, 2001), unless an 'event-related' sampling regime (for example more intensive sampling following N inputs or significant rainfall) is adopted. For example, Parkin (2008) evaluated the impact of sam pling frequency on actual cumulative fluxes, and found that a three-day sampling regime resulted in deviations in cumulative emission estimates being within ±10 per cent. This deviation increased with the increase in time interval between sampling, with 21-day sampling intervals resulting in deviations of between +60 per cent and —40 per cent of the actual cumulative flux. Smith and Dobbie (2001) observed that cumulative N2O emission values from sampling at eight-hour intervals were on average 14 per cent higher than values based on samples collected once every three to seven days. This difference was not significant at the 95 per cent confidence interval level, as both overestimations and underes-timations occurred. It is noteworthy that their temporal variation was smaller than the corresponding spatial variation. Less frequent sampling may be useful for relative comparisons from adjacent treatments, subjected to similar conditions. However, using these data to calculate annual fluxes or emission factors should only be applied with caution, as significant peaks of N2O emission may have been missed. Adopting 'event-related' sampling regimes reduces the error associated with temporal variability, and thus improves the accuracy of determining actual fluxes or emission factors.
Consideration of diurnal variation in soil temperatures and fluxes is required when selecting an appropriate time of day for collecting chamber headspace samples. Fluxes calculated from samples collected between 10am and 12 noon were not significantly different from those based on sampling at eight-hour intervals during a period when diurnal variation in temperatures was small (Smith and Dobbie, 2001). Diurnal temperature and flux data collected from a grassland soil in northern Germany (Dittert et al, 2005) would also suggest that midday sampling is representative of the 24-hour period. In situations where diurnal fluctuations in temperature and flux are likely to be much larger, auto-chambers programmed to sample at a high frequency could be deployed (Smith and Dobbie, 2001).
Was this article helpful?