Impact of nitrogen deposition on N2O exchange

There are a large number of controlling variables and complex interactions that influence the net N2O emission, which would suggest applying a detailed mechanistic model for calculating the effect of atmospheric nitrogen deposition on N2O emissions from European forests. The problem, however, is that the application of such a model on a European scale is limited by the large number of data requirements. Consequently, we used simple, transparent and empirical approaches, with process-based models and empirical data-sets, because these are currently the most feasible to quantify the effect of anthropogenic nitrogen deposition on

N2O emissions.

Regression model based on process-based model

We derived a regression model, predicting N2O emissions as a function of stand and site characteristics and environmental factors, including total nitrogen deposition, based on the predicted N2O emissions (in kg N2O-N/ha/ year) for European forest soils with the PnET-N-DNDC model for a geographic resolution of 50 x 50 km (Kesik et al., 2005). Total nitrogen deposition was calculated from the wet deposition data, used by Kesik et al. (2005) according to the procedure used in PnET-N-DNDC (Li et al., 2000):

[Nltotal = 0.27 + 2.7 [NJrainfall (coniferous forests) (17.6a)

[Nltotal = 0.20 + 1.6 [^N] rainfall (deciduous forests) (17.6b)

where [N]rainfall is the nitrogen concentration in rainfall, which is equal to the wet nitrogen deposition divided by the precipitation. These concentrations were multiplied by the throughfall, which was calculated as a percentage of the precipitation depending on tree species, using precipitation and throughfall data from ~500 level II plots. On average, the throughfall fraction was ~0.7 for conifers (pine, spruce, fir and evergreen oak) and ~0.8 for deciduous trees (oak, beech, birch and hardwoods). The estimated total deposition was on average comparable to the EMEP total nitrogen deposition, but this value was not used, since it was not included in the PnET-N-DNDC calculation. The best result obtained from the regression analyses, while distinguishing between tree species, was:

N2O[kg N/ha/year] = 1.3211 + a . tree species + 0.019925 clay - 0.01329 . Cmin

where clay = clay content (%), Cmin = mean value for the carbon pool in the mineral topsoil (0-30 cm; t C/ha), pH = pH-H2O, Pr = annual precipitation (mm/year), T = mean annual temperature (°C) and Ndep_ tot = total nitrogen deposition (kg N/ha/year). The percentage variance accounted for by this model (r|dj) is 0.42 and the standard error of observations is estimated to be 0.280 kg N2O-N/ha/year.

In the regression model, the impact of tree species was considered in different intercepts, where a is 0 for pine (reference tree), -0.2232 for larch, -0.1604 for fir, -0.0020 for evergreen oak, -0.0727 for spruce, 0.0276 for oak, 0.0396 for hardwoods, 0.1792 for birch and 0.3964 for beech. In general, the results for the deciduous trees were larger than for conifers, comparable to the measurements (Table 17.2). This could be due to differences in canopy structure, acidity of the forest floor and differences in soil moisture, which favour nitrification rather than denitrifica-tion activity in the soil of coniferous forests. The negative relationship with pH is according to expectations since the relative loss of N2O during nitrification and denitrification increases with decreasing pH, and this overrides the stimulating effect of increasing pH on nitrification and denitrification itself. The negative relationship between N2O emissions and carbon content, precipitation and temperature seems opposite to expectations. This is, however, due to the application of denitrification and decomposition (DNDC) on the whole of Europe and the correlations between explaining variables. For example, carbon-rich soils in Scandinavia and the Baltic states have relatively low emissions due to temperature restrictions. The positive impact of carbon is illustrated in the interaction terms of carbon with temperature and precipitation. The estimated average annual N2O emissions calculated with Eq. 17.7 was 0.59 kg N2O-N/ha/year, which is very close to the average value of 0.58 kg N2O-N/ha/ year derived by Kesik et al. (2005), while using the original PnET-N-DNDC model.

Figure 17.4 presents a graph showing the predicted N2O-N emissions with PnET-N-DNDC versus the regression model, showing that the regression model deviates from PnET-N-DNDC predictions for higher annual N2O emissions (>1.0 kg N2O-N). The regression model (Eq. 17.7) shows that a change in nitrogen deposition of 1 kg N/ha/year leads

to an increase of ~0.018 kg N2O-N/ha/year. This is 1.8% of the nitrogen input, which is almost a factor of 2 higher than the default N2O emissions factor of 1% used by the Intergovernmental Panel on Climate Change (IPCC) (e.g. IPCC, 1996; Mosier et al., 1998). The value of 1.8% is close to the value derived by Denier van der Gon and Bleeker (2005) for coniferous forests (see also Tables 17.2 and 17.5 and the discussion in the following section on a regression model based on field measurements).

As with carbon sequestration, the effect of elevated atmospheric nitrogen deposition on N2O emissions from European forest soils was assessed by using the empirical approach outlined above. The calculated average N2O emission during 1960-2000 was compared to the emission in the reference year 1960 (reference nitrogen deposition rates) using available data for all level I plots, representing a total area of 162 million hectares. In this way, we estimated the effect of the cumulative additional nitrogen deposition during 1960-2000, compared with that in 1960, on the cumulative additional N2O emissions. The clay content and carbon pool in the mineral soil and the pH-KCl were based on measurements up to a depth of 20 cm. The mean annual precipi tation and the mean annual temperature were based on an interpolation of 30-year average values for a high-resolution grid in Europe during 1970-2000 (Mitchell et al., 2004). N2O-N emission for 1960 was estimated at 66,000 t, corresponding to an average of 0.41 kg N2O-N/ha/year. The difference between this value and the average European estimate with the PnET-N-DNDC model (0.58 kg N2O-N/ha/year) is mainly due to a different schematization: the use of generic soil data for a geographic resolution of 50 x 50 km by Kesik et al. (2005) compared to the measured data at level I plots in this study. Using an average additional nitrogen deposition of 2.8 kg/ha/year during 1960-2000 leads to an average increase of 0.05 kg N2O-N/ha/year. For 162 million hectares of forests, the average additional N2O emission can be estimated at 8100 t N2O-N/year. Comparing this value with the emission in 1960, it follows that the impact of nitrogen deposition on N2O emissions in the last 40 years is ~12%.

Regression models based on field measurements

In the IPCC methodology for accounting N2O emissions from agriculture (IPCC,

Fig. 17.4. Comparison of N2O emission estimates with 2527 grids of 50 x 50 km with an empirical regression model and the PnET-N-DNDC model.
Table 17.5. Estimated nitrous oxide (N2O) emission factors for deciduous forests and coniferous forests. (From Denier van der Gon and Bleeker, 2005.)

N2O emission factor (%)

Type of forests

Meana

Weighted meana

Median

Regression

Coniferous forests

3.7

2.6

2.0

1.4

Deciduous forests

5.7

5.6

4.4

6.4

aIn calculating mean emission each location and/or study is weighted equally, whereas the weighted mean weighs the average by the number of observations.

aIn calculating mean emission each location and/or study is weighted equally, whereas the weighted mean weighs the average by the number of observations.

1996), the N2O released from atmospheric nitrogen following its deposition on, for example, forest soils is simply calculated as a fraction of the amount of NH3-N lost from agriculture. The emission factor (1% of nitrogen lost) for these indirect N2O emissions from agriculture, multiplied with the total amount of NH3-N lost from agriculture, provides a rough estimate of the indirect N2O emissions. The empirical data in Table 17.2 show, however, that the current IPCC default value of 1% for indirect emissions is underestimating the N2O emissions from forests. Results indicate that the N2O emission fraction (derived as N2O emissions divided by nitrogen deposition) is generally higher than 1%, specifically in deciduous forests.

Derivation of an average N2O emission factor from field measurements data has some methodological artefacts. For example, our data presented in Tables 17.2 and 17.5 clearly indicate that emission factors depend on the way average emissions are calculated. Median values or average values, either weighted by the number of observations or unweighted, assume no N2O emissions when nitrogen deposition is negligible. The average emission factors thus calculated are ~2.5% for coniferous forests and ~5% for deciduous forests. An averaging approach accounting for a certain N2O emission when nitrogen deposition is negligible is linear regression analysis. Results from such an analysis indicate average emission factors of 1.4% for coniferous forests and 5.4% for deciduous forests as shown in Eq. 17.8 (with N2O-N emissions and nitrogen deposition both in kg N/ha/year):

N2O-N emission = 0.088 kg + 0.014 N deposition (r2 = 0.28) for coniferous forests (17.8a)

N2O-N emission = 0.054 N deposition (r2 = 0.29) for deciduous forests

These results are slightly different from those obtained by Denier van der Gon and Bleeker (2005), by removing several outliers, such as an nitrogen-fixing red alder stand with an emission of ~8 N2O-N/ha/ year (see Fig. 17.5). The relatively low N2O emission factor for coniferous forests (1.4%) is due to a relatively large number of low emission observations (Table 17.2) that may not be properly weighted in the regression as it is biased by the sites with many observations. However, this emission factor is closer to the value of 1.8% derived by the PnET-DNDC model (Eq. 17.7).

For the Höglwald spruce site a simple regression model has been derived, based on weekly measured NH4 input via through-fall and weekly measured N2O emissions (Butterbach-Bahl et al., 1998):

Reworking this to the units that are relevant for our calculations on an annual basis, we get:

N2O emission (kg N/ha/year) = 0.41 + 0.0167 NH4-N input (kg N/ha/year)

The formula is derived for NH4 deposition in the range of 0-8 mmol/m2/week, and the regression coefficient of the formula is

Fig. 17.5. N2O-N emission as a function of nitrogen input for (a) deciduous forests and (b) coniferous forests based on the literature. (From Denier van der Gon and Bleeker, 2005.)

r2 = 0.38. According to Eq. 17.9b, 1.67% of the nitrogen deposited is lost as N2O. This estimate again is in good agreement with the independent estimate of 1.4%, derived from literature data for coniferous forests (using linear regression) and the 1.8% derived from the PnET-N-DNDC application (see Eq. 17.7). A much greater impact of nitrogen deposition was found in a pine forest in north-east Germany, where both nitrogen inputs and emissions of N2O, NO and CH4 were measured (Butterbach-Bahl et al., 2002a). Here 14% of atmospheric wet nitrogen input was lost as N2O. Considering a wet/dry deposition ratio of 2 would mean 7% of the total nitrogen input. This high percentage was, however, found at a site with a very high nitrogen input (>20 kg N/ ha/year as wet deposition), largely exceeding nitrogen uptake.

Applying the nitrogen deposition-N2O emission relationships in Eq. 17.8a and Eq. 17.8b leads to an estimated average annual N2O-N emission of 0.3 kg/ha/year for 1960, corresponding to 48600 t N2O-N for the whole of Europe. Using an average additional nitrogen deposition of 2.8 kg/ha/year during 1960-2000, while applying Eq. 17.8a and 17.8b, leads to an average increase of 0.098 kg N2O-N/ha/year, which is equal to 15900 t N2O-N/year.

Comparing this additional emission to the estimated emission in the reference year 1960 suggests that the impact of nitrogen deposition on N2O emissions in the last 40 years has been ~33%. This is three times larger than the estimated contribution of nitrogen deposition on N2O emission increase using Eq. 17.7. Although the discrepancies in the N2O emission estimates by the two approaches are not very large (1960 N2O emissions of 66,000 and 486,00 t N2O-N, respectively and a nitrogen dose-effect curve leading to 8000 and 15,900 t N2O-N/year, respectively), a calculated difference of 12% and 33% additional impact of nitrogen deposition on N2O emissions in the last 40 years is quite significant. Considering both approaches, which are equally relevant, a general conclusion could be that the overall impact of nitrogen deposition on N2O emissions in the last 40 years is ~22% ± 10%. However, the impact of 33% is likely to be an overestimation since the regression for deciduous forests in Eq. 17.8b implies that the background emission is zero, which is not true. In general, the difference in estimates on effects of nitrogen deposition on N2O exchange is largely due to missing information on N2O fluxes under unperturbed, pre-industrial and natural conditions, which would allow us to estimate the magnitude of background emissions.

Was this article helpful?

0 0

Post a comment